{ "schema": "naevum.public-knowledge/4", "name": "Naevum Labs", "url": "https://naevumlabs.com", "contact": "hello@naevumlabs.com", "description": "Naevum Labs is a research and engineering lab building dedicated intelligence environments, persistent agents, institutional memory and compute infrastructure.", "relationships": [ { "from": "agency-engine", "to": "supersell", "type": "foundation-for", "label": "Commercial-intelligence spin-off" }, { "from": "black-box", "to": "conductr", "type": "complementary", "label": "Factory coordination and developer coordination" }, { "from": "black-box", "to": "vault", "type": "complementary", "label": "Research and reusable engineering for custom systems" }, { "from": "synapse", "to": "vagus", "type": "complementary", "label": "Scoped model work and institutional memory" }, { "from": "manna", "to": "anemone", "type": "shared-research-context", "label": "Continuing autonomy; no runtime dependency asserted" }, { "from": "metahive", "to": "metahive-builder", "type": "includes", "label": "Physical and operational infrastructure planning" }, { "from": "everbloom", "to": "vault", "type": "shared-research-context", "label": "Living systems, research and composition; distinct product canon" } ], "environment": { "environment": { "title": "Intelligence, native to your world.", "intro": "Your knowledge, systems and ambitions already form a world. We design the intelligence environment that helps you work within it.", "body": [ "Naevum brings memory, reasoning and continuing action around the people and organizations they serve. A dedicated environment builds an evolving model of your identity, digital footprint, infrastructure and objectives.", "It is personal in the details: the sources it understands, the responsibilities it respects, the preferences you can correct and the work it is allowed to do. We select and evaluate the relevant capabilities around your actual tasks." ], "action": { "label": "Explore your intelligence environment", "href": "/solutions/custom-ai-systems" }, "secondaryAction": { "label": "Discover the systems", "href": "/products" } }, "world_model": { "title": "A world model you can inspect.", "intro": "Connected sources become structured context. Observations keep their origin; inferences remain open to correction. This is the architecture of a dedicated engagement, shaped around the world you choose to connect.", "layers": [ { "id": "identity", "title": "Identity and representation", "represents": "People, organizations, roles, accounts and delegated responsibilities.", "control": "Inspect the active identity, who may act for whom and the source of an attribute." }, { "id": "footprint", "title": "Your digital footprint", "represents": "Selected documents, repositories, communications, applications, devices and services.", "control": "Choose connected sources, exclusions, freshness requirements and access permissions." }, { "id": "knowledge", "title": "Knowledge and time", "represents": "Claims, decisions, episodes, contradictions and the intervals when they apply.", "control": "Follow the evidence, correct a claim and define retention and deletion behavior." }, { "id": "infrastructure", "title": "Infrastructure", "represents": "Assets, services, dependencies, capacity and operating conditions.", "control": "See the topology, current observations, unknowns and responsible operators." }, { "id": "purpose", "title": "Purpose and commitments", "represents": "Objectives, priorities, obligations, success criteria and resource limits.", "control": "Change what the environment is working toward and how success will be judged." }, { "id": "preferences", "title": "Behavior and preferences", "represents": "Your explicit preferences and supported patterns in authorized work.", "control": "Correct an inference and distinguish it from an instruction you deliberately gave." }, { "id": "prediction", "title": "Possible futures", "represents": "Forecasts, horizons, uncertainty and alternative actions.", "control": "Examine assumptions and keep a scenario distinguishable from an observation." }, { "id": "authority", "title": "Authority and consequences", "represents": "Permitted actions, consent, effect identities, results and recovery obligations.", "control": "Inspect what may happen, what happened and what remains uncertain; stop or revise the work." } ] }, "rail_classes": { "personal": "Cybernetic organisms · personal and commercial", "institution": "Bio-inspired systems and foundations · Tiers 2–3", "build": "Foundations · how we build", "foresight": "Bio-inspired systems and cybernetic organisms · prediction", "infrastructure": "Infrastructure · Tier 1 and compute" }, "rail_evidence": { "personal": { "evidence": "E-K7", "text": "Eight in ten workers and leaders say they lack the time or energy to do their work [E-K7]." }, "institution": { "evidence": "E-K4", "text": "Executives and teams spend a quarter of the workweek searching for information [E-K4]." }, "build": { "evidence": "E-A5", "text": "A preliminary MIT study found that only 5% of custom enterprise AI tools reach production [E-A5]." }, "foresight": { "evidence": "E-D1", "text": "Siemens estimates that unplanned downtime costs the 500 largest companies 11% of their revenue [E-D1]." }, "infrastructure": { "evidence": "E-E2", "text": "The IEA expects data-center electricity use to roughly double by 2030 [E-E2]." } } }, "products": [ { "id": "black-box", "name": "Black Box Factory", "category": "AI systems & software development", "url": "https://naevumlabs.com/products/black-box", "description": "Black Box organizes custom AI delivery around a client mandate, architecture, implementation, independent challenge and operational handover. Its selected factory design keeps acceptance tied to the exact change being judged.", "applications": [ { "title": "Turn architectural direction into delivery", "audience": "Organizations commissioning custom AI", "situation": "A difficult brief spans data, software, interfaces and operating responsibilities.", "mechanism": "The Planner defines a mandate; implementation, review and QA preserve its acceptance conditions.", "impact": "The customer can examine how the delivered system answers the original brief." }, { "title": "Continue after a worker disappears", "audience": "Engineering teams", "situation": "A development attempt ends before the work is resolved.", "mechanism": "The source-bound handover retains objective, checks, findings and the next decision.", "impact": "Another attempt can continue from the durable record rather than relying on a departed session." }, { "title": "Preserve independent challenge at scale", "audience": "Product organizations", "situation": "Several implementation streams compete for the same finite engineering capacity.", "mechanism": "Separate role principals and reserved review capacity keep judgment in the delivery path.", "impact": "Increased implementation activity does not silently consume its own acceptance process." } ], "technical_features": [ { "name": "A mandate before an attempt", "description": "Scope, ceilings, budget and expiry are bound to an owner-confirmed mandate. The local Planner translates architectural direction into work the factory can admit." }, { "name": "Roles are separate principals", "description": "Coder, Reviewer and QA have distinct responsibilities. A result is challenged outside the author’s context; the merge authority checks the required verdict rather than accepting the author’s own success claim." }, { "name": "The handover is a record", "description": "Source-bound packets carry objective, implementation, checks and unresolved findings between short-lived attempts. Restarting a session does not require reconstructing intent from memory." }, { "name": "Reserve capacity for the challenge", "description": "Implementation and review compete for finite resources. Review capacity is reserved so a growing build queue does not consume the capacity needed to judge its results." }, { "name": "Acceptance follows the exact source", "description": "Repository gates, source identity, review evidence and intended delivery environment remain connected. A passing check on one revision cannot qualify a different change." } ], "operating_context": null, "mechanism": [ { "name": "Define", "description": "State the purpose and acceptance conditions in a source-bound mandate." }, { "name": "Build", "description": "Create a bounded implementation attempt against that mandate." }, { "name": "Challenge", "description": "Independent review and QA examine the identified candidate. Corrections produce a new source version for another review." }, { "name": "Deliver", "description": "Bind an accepted change to its exact source, review evidence and intended environment." } ], "related": [ "https://naevumlabs.com/products/agency-engine", "https://naevumlabs.com/products/vault", "https://naevumlabs.com/products/conductr" ], "editorial": { "lead": "From an important problem to a system you can run.", "audience_job": "Black Box Factory is Naevum's custom AI delivery service for leaders with one operating problem, so that a separate reviewer checks the system before their team runs it.", "chapters": [ { "id": "client-artifacts", "kind": "process", "title": "The brief becomes a set of inspectable artifacts.", "intro": "A useful AI prototype still has to work with the business’s data, interfaces and operating responsibilities. The commissioning team needs a clear path from that brief to software it can examine and run.", "items": [ { "title": "The mandate", "text": "Define the work, operating boundaries and outcomes that acceptance must examine.", "input": "A workflow that needs to change.", "output": "A bounded system brief and acceptance conditions." }, { "title": "The architecture", "text": "Choose components, data contracts and integration points appropriate to the problem.", "output": "A design whose responsibilities can be examined." }, { "title": "The accepted artifact", "text": "Bind the built software, review findings and QA evidence to an exact source and intended environment.", "output": "A release and the record supporting its acceptance." } ], "body": [ "Bring Black Box a bounded workflow that crosses those responsibilities. Start with its intended user journey and the conditions that would make the delivered result acceptable; the mandate, architecture and independently examined artifact carry that intent into handover." ] }, { "id": "independent-challenge", "kind": "mechanism", "title": "Creation and acceptance have separate responsibilities.", "intro": "Black Box’s selected factory architecture gives Planner, Coder, Reviewer, QA and service distinct powers.", "body": [ "The Planner translates architectural intent into a proposed mandate. An eligible owner confirms it. Implementation is a bounded attempt; independent review concerns the exact source being judged. The service cannot manufacture the reviewer’s acceptance.", "The design reserves capacity for review and QA. It includes acceptance criteria before implementation, checks on the installed artifact and exploratory examination of surfaces that fixed tests do not yet cover.", "The selected development direction gives Planner, Coder, Reviewer and QA different identities and bounded attempts. The role-separated triad is an architectural proposal; an admitted project pipeline still needs its own qualification. Reviewer capacity and credentials are reserved as independent resources." ] }, { "id": "handover-and-operation", "kind": "constraints", "title": "Delivery includes the next operator.", "intro": "A working artifact needs a way to be installed, observed, recovered and changed.", "items": [ { "title": "Carry the unresolved work", "text": "Durable handovers retain source, result, findings and the next decision. Incoming material is treated as evidence rather than new authority." }, { "title": "Make operation concrete", "text": "The engagement specifies controls, documentation, recovery and support responsibilities." }, { "title": "Keep the route accurate", "text": "Naevum delivers through project-specific build, review and release workflows. The full unattended factory loop is a development direction." } ] } ], "questions": [ { "question": "Do I need the whole Naevum portfolio?", "answer": "No. The architecture selects the components the problem needs. A bounded integration or a single well-defined product can be the right engagement." }, { "question": "What should I bring to the first conversation?", "answer": "Describe the work you want to change, its users, current systems, operating constraints and how you would recognize a useful result." } ], "relationships": [ { "type": "foundation", "id": "agency-engine", "label": "Application architecture", "text": "A composable foundation where it suits the product.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "uses", "id": "vault", "label": "Engineering knowledge", "text": "Reusable mechanisms and their assumptions.", "url": "https://naevumlabs.com/products/vault" }, { "type": "related", "id": "conductr", "label": "Development discipline", "text": "Harness practice and independent verification.", "url": "https://naevumlabs.com/products/conductr" } ], "action": { "label": "Discuss a system to build", "href": "https://naevumlabs.com/contact?context=black-box" }, "example": { "title": "A useful review changes the candidate.", "intro": "Illustrative scenario. This is a proposed role-separated delivery trace, not a live autonomous factory demonstration.", "states": [ { "id": "candidate", "label": "Candidate", "title": "The implementation answers the brief but misses a failure case.", "body": "An identified source candidate carries the plan and its acceptance conditions into review.", "records": [ { "name": "Mandate", "value": "Preserve a user’s pending request through interruption." }, { "name": "Coder result", "value": "A candidate restores the normal successful path." }, { "name": "Independent challenge", "value": "The reviewer finds that an uncertain effect can be duplicated." }, { "name": "QA question", "value": "Can the same operation resume without a new external effect?" } ] }, { "id": "correction", "label": "Corrected candidate", "title": "The next attempt resolves the review finding.", "body": "A durable attempt record connects the correction to the original mandate. Delivery remains its own decision.", "records": [ { "name": "Correction", "value": "Retain operation identity during reconciliation." }, { "name": "Verification", "value": "Exercise the interrupted route against the acceptance condition." }, { "name": "Independence", "value": "Reviewer and QA use their separate identities and capacity." }, { "name": "Handover", "value": "Provide source, artifact and verification evidence to the operator." } ] } ] } }, "anatomy": { "representation": "An illustrative development process from bounded brief through candidate review and handover.", "signature": "Construction and independent challenge", "parts": [ { "id": "definition", "name": "Brief and architecture", "description": "Turn the desired workflow into responsibilities, interfaces and inspectable acceptance conditions." }, { "id": "attempts", "name": "Attempt and resource scheduler", "description": "Bind one eligible attempt to its source and finite capacity." }, { "id": "implementation", "name": "Implementation bench", "description": "Transform the task into an artifact and correct identified defects." }, { "id": "challenge", "name": "Independent challenge", "description": "Review and QA examine the same candidate through separate responsibilities." }, { "id": "acceptance", "name": "Exact-source comparator", "description": "Join candidate, current base, policy, checks and independent verdicts." }, { "id": "handover", "name": "Handover and continuity", "description": "Preserve artifacts, unresolved work and original uncertain attempts for the next operator." } ] } }, { "id": "synapse", "name": "Synapse", "category": "Private computation & model orchestration", "url": "https://naevumlabs.com/products/synapse", "description": "Synapse binds model work to an identified source, a chosen observer and a permitted effect. Cumulative disclosure history and durable recovery preserve the meaning of a task when work is concurrent or interrupted.", "applications": [ { "title": "Use outside intelligence on a bounded task", "audience": "Institutions and R&D teams", "situation": "A useful model needs context, while the organization holds a much larger sensitive source.", "mechanism": "Select a source-bound projection and account for its disclosure to the actual observer before dispatch.", "impact": "The organization can inspect the approved context, purpose and destination of the model work." }, { "title": "Recover an interrupted agent attempt", "audience": "Software teams", "situation": "A connection drops after a request may already have reached its provider.", "mechanism": "The durable effect remains uncertain and reuses its exact identity during reconciliation.", "impact": "Recovery can establish what happened without silently repeating an action or forgetting prior exposure." }, { "title": "Put a budget on an inference route", "audience": "Compute operators", "situation": "Concurrent requests can pass a separate check before either has charged its usage.", "mechanism": "Transactional reservation binds admission to the request, with settlement linked to received evidence.", "impact": "The inference route has a durable basis for concurrent resource accounting.", "link": "/products/metahive" } ], "technical_features": [ { "name": "Immutable context admission", "description": "Context is bound to a source snapshot, provenance, holder, actor, client and workspace. Registration and disclosure authority remain distinct." }, { "name": "Cumulative observer history", "description": "Admission can consider what a destination has already received across several effects." }, { "name": "Reserve before transport", "description": "Effect identity and cumulative call and byte allowances are reserved before dispatch, including concurrent work." }, { "name": "Source and effect binding", "description": "Results belong to the job, step, source and policy they concern. Deduplication and recovery use that identity." }, { "name": "Durable uncertainty", "description": "An interrupted effect retains its history until reconciled. Abandonment does not erase disclosure." }, { "name": "Bounded transformation and recovery", "description": "Source selection, restricted projection, recovery objects and returned-edit validation have explicit supported-route and language limits." } ], "operating_context": { "label": "Used in Naevum’s engineering workflows", "description": "Synapse is used in Naevum’s engineering workflows. The context, admission and recovery mechanisms described here retain their own supported-route scope." }, "mechanism": [ { "name": "Reserve", "description": "In the admissible comparison branch, bind the fixed candidate and reserve review-014 before transport." }, { "name": "Interrupt", "description": "One admitted attempt loses its reply. Keep exposure accounting and outcome uncertainty under review-014." }, { "name": "Reconcile", "description": "Recover the existing result and its receipt. Reconciliation sends no new request." } ], "related": [ "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/manna" ], "editorial": { "lead": "Each proposed view is checked against what its observer has already received and what remains unresolved.", "audience_job": "Synapse is a private model boundary for organizations that use outside AI models, so that each provider receives only what a task needs and each send through it is recorded.", "chapters": [ { "id": "context-journey", "kind": "process", "title": "One task. A carefully chosen view.", "intro": "An institution wants a model to help with a document. The institution holds much more information than that task requires.", "items": [ { "title": "Bind the source", "text": "Register the context against an immutable source snapshot and its provenance. Holder, actor, client and workspace identify the intended setting.", "input": "An identified source and a bounded task.", "output": "A context record that can be examined." }, { "title": "Choose the observer", "text": "The destination and what it has already received matter to disclosure. Registering context does not itself authorize transmission." }, { "title": "Reserve the effect", "text": "Reserve task identity and cumulative call and byte allowances before transport. Concurrent work must account for the same finite budget." } ], "body": [ "The source owner assigns permitted information classes. Before dispatch, the capability fixes the observer and model, task policy, expiry and stopping conditions. The receiving model does not grant itself that authority." ] }, { "id": "observer-history", "kind": "mechanism", "title": "A series of small views can reveal a larger picture.", "intro": "A useful disclosure boundary considers the observer’s cumulative history.", "body": [ "Two individually limited requests may expose more together than either does alone. Synapse tracks what a particular observer has already received across effects, so admission can account for that accumulated view.", "Source selection, task policy and allowed effects are distinct decisions. A supported coding projection also needs validation of returned edits and a recovery path back to the source. Each integration must specify the transformations and languages it supports." ] }, { "id": "interrupted-effect", "kind": "constraints", "title": "A missing answer does not mean nothing happened.", "intro": "A connection disappears after dispatch. The model may have received the context even though its result never arrived.", "items": [ { "title": "Keep the identity", "text": "The job, step, source and policy binding remain attached to the effect." }, { "title": "Reconcile before repeating", "text": "Inspect the existing effect and its receipts. A fresh identifier cannot safely stand in for an unresolved attempt." }, { "title": "Recover with the record", "text": "Received results, interrupted work and abandonment retain their disclosure history. A recovery decision follows the known state." } ] } ], "questions": [ { "question": "Does Synapse protect every model, IDE and operating-system route?", "answer": "Coverage is specified for each supported integration and deployment. A protected task route does not establish universal IDE interception or complete operating-system containment." }, { "question": "What does a dedicated deployment add?", "answer": "It can change custody, compute location and the parties involved. Context admission, observer history and action controls still need to be designed and tested for that deployment." } ], "relationships": [ { "type": "related", "id": "vagus", "label": "Institutional context", "text": "Temporal memory supplies source-aware knowledge.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "related", "id": "manna", "label": "Continuing execution", "text": "An agent needs boundaries for every admitted effect.", "url": "https://naevumlabs.com/products/manna" }, { "type": "deployment", "id": "metahive", "label": "Compute choices", "text": "Dedicated infrastructure is one deployment option.", "url": "https://naevumlabs.com/products/metahive" } ], "action": { "label": "Design your model boundary", "href": "https://naevumlabs.com/contact?context=synapse" } }, "anatomy": { "representation": "An illustrative role map. Configured routes and profiles determine actual capabilities and limits.", "parts": [ { "id": "0", "name": "Local context", "description": "Retains original context and the record of prior exposure." }, { "id": "1", "name": "Local processing", "description": "Transforms local material before a view is proposed." }, { "id": "2", "name": "Compose and route", "description": "Composes a bounded request and binds its destination." }, { "id": "3", "name": "Parallel check · upper", "description": "Checks an independently routed request before transport." }, { "id": "4", "name": "Parallel check · middle", "description": "Checks an independently routed request before transport." }, { "id": "5", "name": "Parallel check · lower", "description": "Checks an independently routed request before transport." }, { "id": "6", "name": "Scoped work · array", "description": "Processes only the admitted view for this request." }, { "id": "7", "name": "Scoped work · layers", "description": "Processes only the admitted view for this request." }, { "id": "8", "name": "Scoped work · core", "description": "Processes only the admitted view for this request." }, { "id": "9", "name": "Return check", "description": "Treats returned observations as untrusted until checked." }, { "id": "10", "name": "Reconciliation and history", "description": "Relates observations and consequences to the same originating identity." }, { "id": "11", "name": "Result control", "description": "Separates checked observations from proposals for application effects." }, { "id": "12", "name": "Application executor", "description": "Performs only an independently authorized application effect." }, { "id": "13", "name": "Output and receipt", "description": "Records the output and consequence of that authorized effect." }, { "id": "14", "name": "Adaptive check · upper", "description": "Adds a temporary checking route while requests need it." }, { "id": "15", "name": "Adaptive check · lower", "description": "Adds a temporary checking route while requests need it." } ] } }, { "id": "vagus", "name": "Vagus", "category": "Institutional memory & attention", "url": "https://naevumlabs.com/products/vagus", "description": "Vagus connects temporal knowledge to attention and operational decisions. Claims retain their source and validity; signals accumulate over time; outcome estimates keep their uncertainty.", "applications": [ { "title": "An institution remembers its decisions", "audience": "Institutions", "situation": "A policy changes, but earlier claims and the reasons for past decisions still matter.", "mechanism": "Validity intervals, authority ancestry and contested claims preserve the history of understanding.", "impact": "A new decision can distinguish the current position from an earlier or unsupported one." }, { "title": "Attention across a technical estate", "audience": "Engineering organizations", "situation": "Signals arrive from many projects, workers and resource windows.", "mechanism": "Temporal attention integrates repeated observations and uses failure and outcome memory.", "impact": "The operator sees why a matter deserves attention without treating every new observation as an emergency." }, { "title": "Learn which work succeeds", "audience": "Operational teams", "situation": "Different attempts consume different resources and produce different outcomes.", "mechanism": "Outcome predictions and delayed reward joins retain the task and source behind each observation.", "impact": "Selection can improve from recorded experience while the operator keeps control of authority." } ], "technical_features": [ { "name": "Claims within validity intervals", "description": "Temporal recall can recover the claim that applied at an earlier point. Source pointers and contested assertions remain attached to the answer." }, { "name": "A rebuildable evidence index", "description": "Claims, lessons, failures and episodes retain their roles. A derived index can be reconstructed without inventing a second authoritative history." }, { "name": "Integrate-and-fire attention", "description": "Signals accumulate, decay and cross thresholds. Refractory behavior limits repeated firing; missing observability can itself contribute a signal." }, { "name": "Hierarchical outcome estimates", "description": "Success proxies, elapsed time and resource estimates can share evidence across levels while preserving prior strength and uncertainty." }, { "name": "Failure memory with hysteresis", "description": "Measured harms can add decaying caution or quarantine context. That experience changes caution without granting a new capability." } ], "operating_context": { "label": "Used for Naevum’s memory and work records", "description": "Vagus is used for Naevum’s memory and work records, connecting source evidence to signals, attention and ongoing tasks." }, "mechanism": [ { "name": "Remember", "description": "Keep episodes, lessons, claims and failure memory with their provenance and holder." }, { "name": "Attend", "description": "Integrate signals over time, with decay, refractory behavior and explicit unknown states." }, { "name": "Adapt", "description": "Compare a retained estimate with a later identified outcome. Keep uncertainty and useful failures available for the next decision." } ], "related": [ "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/products/manna", "https://naevumlabs.com/products/vault" ], "editorial": { "lead": "An organization should not have to learn the same lesson twice.", "audience_job": "Vagus is institutional memory for organizations, so that anyone can ask why a decision was made and get who decided, when, and on which sources.", "chapters": [ { "id": "then-and-now", "kind": "comparison", "title": "What did we know when we made the decision?", "intro": "A delivery date moves. The new date is current, but the old commitment still explains yesterday’s plan.", "items": [ { "title": "At the earlier date", "text": "A temporal claim is resolved within its validity interval. A historical question can recover the claim that applied at that time.", "input": "The original commitment and its supporting source.", "output": "The answer appropriate to the earlier decision." }, { "title": "With today’s evidence", "text": "A replacement claim changes the current answer. Source pointers and contested claims preserve the reason for that change.", "input": "A revised commitment, its source and effective date.", "output": "The current answer alongside the changed history." } ] }, { "id": "attention", "kind": "mechanism", "title": "Which change deserves attention now?", "intro": "A large organization produces more signals than anyone can examine individually.", "items": [ { "title": "Accumulate relevance", "text": "Signals charge a subject’s attention potential. Decay, thresholds and refractory periods shape when it becomes worth examining." }, { "title": "Notice missing visibility", "text": "A quiet source and an unobserved source are different conditions. Missing observability can itself contribute an attention signal." }, { "title": "Remember the failure", "text": "Measured harms create decaying failure contexts. Hysteresis can preserve appropriate caution as conditions change; a warning cannot grant a new capability." } ] }, { "id": "outcomes", "kind": "evidence", "title": "Experience should carry its uncertainty.", "intro": "Vagus connects memory and attention to estimates of task outcomes, elapsed time and resource use.", "body": [ "Predictions of task outcomes, elapsed time and resource use enter the record before the outcome arrives. Later observations test the original estimate; they cannot silently replace it with a prediction written after the fact.", "Hierarchical models can share evidence across a portfolio, organization, project and worker while retaining the uncertainty of a small sample. Worker selection can explore eligible options; an empty evidence base cannot support a confident winner.", "The memory index can be rebuilt from source-connected claims, lessons, failures and episodes. A derived lesson remains distinguishable from a binding instruction. Useful evaluation asks whether people recover the right evidence and make better-supported decisions." ] } ], "questions": [ { "question": "Is this search across documents?", "answer": "Document retrieval is one part. Vagus adds temporal claims, operating signals, attention and outcome context so that a source can influence ongoing work while keeping its history." }, { "question": "How would an institution start?", "answer": "Choose a bounded set of sources and recurring decisions. Establish access boundaries, a fixed evaluation question set and the workflow where the resulting evidence will be used." } ], "relationships": [ { "type": "uses", "id": "synapse", "label": "Control the context", "text": "Choose which institutional knowledge may reach a model.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "related", "id": "manna", "label": "Continue the work", "text": "Memory can inform a continuing objective.", "url": "https://naevumlabs.com/products/manna" }, { "type": "related", "id": "agency-engine", "label": "Build the workspace", "text": "A domain interface can make knowledge useful in daily operations.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Discuss institutional intelligence", "href": "https://naevumlabs.com/contact?context=vagus" }, "example": { "title": "Monday’s decision, Thursday’s knowledge.", "intro": "Illustrative scenario. An internal delivery commitment changes. Examine the earlier decision and the present recommendation without rewriting history.", "states": [ { "id": "monday", "label": "Monday", "title": "The earlier plan used the commitment then in force.", "body": "A historical question resolves the claim valid on Monday and retains the document that supported it.", "records": [ { "name": "Question", "value": "Why did we reserve the launch window?" }, { "name": "Valid claim", "value": "Delivery was scheduled for Friday." }, { "name": "Source", "value": "The original supplier commitment." }, { "name": "Decision", "value": "Keep the Friday launch reservation." } ] }, { "id": "thursday", "label": "Thursday", "title": "The recommendation changes with the new evidence.", "body": "The revised source changes the current answer. The earlier commitment still explains Monday’s work.", "records": [ { "name": "Current claim", "value": "The supplier now commits to Tuesday." }, { "name": "Attention", "value": "The changed dependency can cross the configured attention threshold." }, { "name": "Next decision", "value": "Revisit the launch reservation with the project owner." }, { "name": "Outcome test", "value": "Record the expected delay before observing the actual delivery." } ] } ] } }, "anatomy": { "representation": "An illustrative process for source-grounded recall, attention and outcome comparison.", "signature": "Temporal archive and bounded recall", "parts": [ { "id": "records", "name": "Evidence registration", "description": "Append source-connected claims, episodes, lessons and measured failures as distinct records." }, { "id": "interpretation", "name": "Temporal interpretation", "description": "Resolve validity, knowledge time, scope and conflicting authority for the requested question." }, { "id": "recall", "name": "Bounded recall workbench", "description": "Find relevant evidence and assemble a bounded source-bearing answer." }, { "id": "attention", "name": "Accumulated attention", "description": "Integrate eligible change, retain standing causes and distinguish blindness from resolution." }, { "id": "outcomes", "name": "Outcome and selection advice", "description": "Compare a stored estimate with a later identified outcome and retain uncertainty." }, { "id": "maintenance", "name": "Maintenance and evaluation", "description": "Rebuild derived context and retain useful failures without replacing source history." } ] } }, { "id": "manna", "name": "Manna", "category": "Persistent personal & industrial agency", "url": "https://naevumlabs.com/products/manna", "description": "Manna gives a continuing agent purpose, an internal organization and a way to regulate its work. Personal and Industrial editions connect explicit laws, sensing, memory, admitted action and recovery to different operating responsibilities.", "applications": [ { "title": "A project that continues tomorrow", "audience": "Personal edition", "situation": "A person has a continuing objective spread across sessions, files and decisions.", "mechanism": "A genome retains the objective and envelope; heartbeats and memory resume work with its earlier context.", "impact": "The next session can begin with the objective, prior decisions and outstanding approvals already connected." }, { "title": "A process with a safe stopping point", "audience": "Industrial edition", "situation": "An operational process must react to observations while respecting equipment and responsibility boundaries.", "mechanism": "A pipeline separates perception, proposal and gated action; the journal records the resulting state.", "impact": "Operators can inspect why a step was admitted, stop the process and recover from its record." }, { "title": "Specialists working toward one purpose", "audience": "Engineering teams", "situation": "Several specialized agents need a shared direction and finite resources.", "mechanism": "Constellations combine bounded mandates with attention allocation and fault isolation.", "impact": "A failure or urgent signal can be addressed locally while the broader objective remains intact." } ], "technical_features": [ { "name": "Constitutional composition", "description": "A supervisor can examine the union of laws within a defined fragment while organisms retain their journals and failure domains." }, { "name": "Resource-aware regulation", "description": "Load, memory and thermal observations inform a declared pressure model. Missing measurements remain unknown." }, { "name": "Attention with protected obligations", "description": "An allocator can preserve due work, limit an organ’s share and expose lost capabilities. Dwell and hysteresis reduce unstable switching." }, { "name": "Continuity checks", "description": "Journal ancestry, required state, behavioral probes and self-description contribute to checking a successor against the work it continues." }, { "name": "Bounded predictive research", "description": "A synthetic scheduling organ records forecasts before outcomes and studies lookahead under uncertainty. Its scope is distinct from a general-purpose world model." }, { "name": "Outcome-aware evaluation", "description": "Structure, policy, artifact and objective achievement have separate meanings. An objective needs units, horizon, baseline and an evaluator." } ], "operating_context": { "label": "Running in a controlled environment", "description": "Manna runs continuously in a controlled environment. It has made millions of decisions and observations and has learned its host’s operating patterns through observation and self-improvement. Industrial applications have their own operating mandate, equipment integration and acceptance." }, "mechanism": [ { "name": "Declare", "description": "Describe purpose, means, operating envelope and the flow of work in an inspectable genome." }, { "name": "Continue", "description": "Select relevant material, compare it with retained context, and assemble a revised artifact. Keep the result and unfinished work in the journal." }, { "name": "Act within bounds", "description": "Present an identified request under its declared envelope. Couple the applicable grant before acting, then retain the action and independent evidence of its outcome." } ], "related": [ "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/lab/anemone" ], "editorial": { "lead": "Give your work a life beyond the next prompt.", "audience_job": "Manna is a continuing agent for owner-operators and industrial teams, so that work on your goal continues while you are away and your decisions wait for you.", "chapters": [ { "id": "continuing-purpose", "kind": "narrative", "title": "The project continues. The conditions change.", "intro": "You leave a research task unfinished. When you return, its objective still exists, the available resources have changed, and an outstanding decision needs you.", "body": [ "Manna gives a continuing agent an internal organization. Its genome states purpose and laws. Specialized cells and pipelines describe how work happens, while a journal records the consequences of admitted actions.", "A Personal agent can carry a project between sessions. An Industrial system needs an operating mandate around a process, its equipment and the people responsible for it. The editions share a language; their responsibilities differ." ], "intro_role": "editorial-statement" }, { "id": "regulation", "kind": "mechanism", "title": "An agent should notice its own condition.", "intro": "Load, memory pressure and thermal observations can change how a continuing system allocates its attention.", "items": [ { "title": "Sense the condition", "text": "Interoception reads resource observations and preserves missing measurements as unknown. A pressure model can increase vigilance or lengthen the interval between ticks." }, { "title": "Protect the obligation", "text": "Attention allocation can protect due commitments, limit one organ’s share and expose capabilities lost under resource pressure. Dwell and hysteresis reduce unstable switching." }, { "title": "Admit the action", "text": "A pipeline separates observation and proposals from effects. Its operating envelope and gates determine which action may happen; extra urgency does not expand that authority." } ] }, { "id": "continuity", "kind": "process", "title": "Resume the purpose, with its history intact.", "intro": "Recovery is an engineering question: which state, obligations and behavior must survive?", "items": [ { "title": "Retain ancestry", "text": "Journal ancestry and required state connect a successor to the work it continues." }, { "title": "Change an organ deliberately", "text": "A new organ starts from a reviewed template, reserves resources and retains its identity through activation. Retirement checks for outstanding obligations and protected functions before the organ can leave.", "input": "A proposed capability and its declared resource cost.", "output": "An admitted lifecycle transition with the continuing purpose preserved.", "boundary": "A pending human decision remains pending after the transition." }, { "title": "Examine behavior", "text": "Behavioral probes and self-description contribute to continuity checks. A process starting again is only one part of recovery." }, { "title": "Judge the objective", "text": "Execution, artifact quality and objective achievement answer different questions. A useful outcome needs units, a time horizon, a baseline and an evaluator." }, { "title": "Prove a rollback", "text": "A rollback certificate records the intended restoration; a separate check must establish what was actually restored. External consequences remain part of the history.", "boundary": "Structure, policy, artifact quality and useful outcomes retain separate assurance judgments." } ] }, { "id": "research-organs", "kind": "evidence", "title": "Prediction and growth have their own research questions.", "intro": "Manna explores bounded lookahead, supervised organ lifecycles and resource-aware development.", "body": [ "Manna’s predictive organ currently studies synthetic scheduling. Forecasts enter a ledger before outcomes arrive, so an experiment can test whether bounded lookahead improves the later decision.", "A personal or industrial engagement selects its capabilities around the actual operating task and evaluates their contribution." ] } ], "questions": [ { "question": "What belongs in Personal Manna?", "answer": "Continuing work around a person’s projects: retain an objective, observe relevant changes, prepare useful work and return decisions that require the person’s authority." }, { "question": "How does Industrial Manna differ?", "answer": "Its mandate is designed around an operating process, resources, permitted interventions and recovery responsibilities. Equipment integration and task acceptance are evaluated for the intended setting." } ], "relationships": [ { "type": "uses", "id": "vagus", "label": "Memory and attention", "text": "Temporal context can inform continuing work through a defined integration.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "uses", "id": "synapse", "label": "Admitted model work", "text": "Model context and effects need their own boundaries.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "research", "id": "anemone", "label": "Consequential autonomy", "text": "A separate lab-owned application of continuing autonomous operation.", "url": "https://naevumlabs.com/lab/anemone" } ], "action": { "label": "Discuss your Manna edition", "href": "https://naevumlabs.com/join?product=manna" }, "example": { "title": "Resume the project. Keep the outstanding decision.", "intro": "Illustrative scenario. A research project pauses while resource availability changes. Continuation preserves its purpose and responsibilities.", "states": [ { "id": "pause", "label": "Before the pause", "title": "The project has useful work and an unresolved approval.", "body": "The journal connects the objective, the completed source review and a proposed external action.", "records": [ { "name": "Purpose", "value": "Prepare an evidence-backed research brief." }, { "name": "Completed", "value": "Source review and a draft outline." }, { "name": "Outstanding", "value": "Owner approval for an external data request." }, { "name": "Resource state", "value": "A compute-heavy analysis is still pending." } ] }, { "id": "resume", "label": "On resumption", "title": "A smaller work allocation preserves the same obligation.", "body": "Resource observations can change pacing or the proposed organ. They do not settle the owner’s outstanding decision.", "records": [ { "name": "Continue", "value": "Retain the source review, outline and causal ancestry." }, { "name": "Adapt", "value": "Prepare the lighter analysis under the current resource envelope." }, { "name": "Keep pending", "value": "The external request still needs owner approval." }, { "name": "Check", "value": "Probe the restored behavior and verify any proposed rollback separately." } ] } ] } }, "anatomy": { "representation": "An illustrative continuing-work process.", "signature": "Source-led productive work machine", "parts": [ { "id": "reference", "name": "Purpose and progress", "description": "Compare observed progress with a retained objective without equating a completed operation with objective achievement." }, { "id": "evidence", "name": "Evidence and contextual memory", "description": "Align selected sources with retained context, validity and unresolved gaps." }, { "id": "regulation", "name": "Attention and condition regulation", "description": "Allocate finite work and pacing from purpose, known resource condition and retained faults." }, { "id": "workbench", "name": "Specialised workbench", "description": "Transform source material and retained context into a useful revised artifact." }, { "id": "effect-interface", "name": "Bounded effect interface", "description": "Hold an identified external request, couple only its applicable grant, and issue one admitted action." }, { "id": "continuity", "name": "Consequence and continuity", "description": "Retain work results, action records, independent evidence and unfinished obligations across continuation." } ] } }, { "id": "anemone", "name": "Anemone", "category": "Naevum’s automated multi-market fund", "url": "https://naevumlabs.com/lab/anemone", "description": "Anemone is Naevum’s own multi-market fund, engineered for fully automated operation. Its continuing mandate gives the lab a consequential setting for experiments, calibration and operational learning.", "applications": [ { "title": "Operate an automated multi-market fund", "audience": "Naevum’s own fund", "situation": "Market conditions keep changing after any one decision is made.", "mechanism": "A continuing mandate connects observation, constrained action and subsequent evidence.", "impact": "The operating system can be examined across a sequence of decisions rather than a single recommendation." }, { "title": "Calibrate against sustained operation", "audience": "Autonomy research", "situation": "A mechanism must be understood across conditions and over time.", "mechanism": "Continuing experiments and calibration are grounded in the fund’s retained operating history.", "impact": "New ideas can be investigated in the context of the system’s actual responsibilities and prior behavior." }, { "title": "Study an autonomy boundary with consequences", "audience": "Research partners", "situation": "A general autonomy claim needs a concrete operating setting.", "mechanism": "Anemone is a distinct fund application within Naevum’s broader autonomy programme.", "impact": "The discussion can be about mandate, intervention and observed behavior without treating a research profile as an investment offer." } ], "technical_features": [ { "name": "A continuing multi-market mandate", "description": "Naevum’s proprietary automated fund maintains a continuing operating objective across changing markets. Observation, decision, execution and subsequent evidence belong to the same system." }, { "name": "Authority and risk stay in the loop", "description": "The fund’s operating boundaries shape which actions can proceed. An autonomous decision remains accountable to the mandate and the limits governing the portfolio." }, { "name": "Calibration through sustained operation", "description": "The system runs over time with continuing experiments and calibration. The operating record supports the investigation of decisions and consequences across changing conditions." }, { "name": "A fund as an autonomy system", "description": "Observation, decision and action stay connected over time. The operating record gives Naevum a concrete setting in which to examine continuing autonomous behavior." } ], "operating_context": { "label": "Naevum’s own automated fund", "description": "Anemone is Naevum’s own automated multi-market fund in ongoing operation. The public profile makes no investment intake or comparative return claim." }, "mechanism": [ { "name": "Observe", "description": "Read changing market conditions in the context of a continuing mandate." }, { "name": "Constrain", "description": "Keep decision making inside the fund’s operating and risk boundaries." }, { "name": "Respond", "description": "Retain the permitted action and its execution record. Compare independent later conditions with the decision context to inform continuing work." } ], "related": [ "https://naevumlabs.com/products/manna" ], "editorial": { "lead": "Autonomy, living with its decisions.", "audience_job": "Anemone is Naevum’s own automated multi-market fund, a distinct application of continuing autonomous operation.", "chapters": [ { "id": "own-business", "kind": "narrative", "title": "A continuing mandate in the lab’s own business.", "intro": "A market decision remains exposed to what happens next. Naevum uses its own automated multi-market fund to study that continuing loop of observation, constrained action and later consequences.", "body": [ "Anemone operates in Naevum’s own business, with months of ongoing experiments and calibration. Conditions change, decisions consume resources, and later observations can challenge the reasoning behind an earlier action.", "The continuing operation puts observation, decision-making and consequence into the same research setting.", "The relevant question is how an autonomous operating mandate continues as conditions change. Explore Anemone as Naevum’s own research and operating setting, then follow how a decision returns as evidence for later examination." ], "intro_role": "editorial-statement" }, { "id": "continuing-loop", "kind": "process", "title": "Observe, decide, constrain, revisit.", "intro": "A continuing process follows its decisions through changing conditions.", "items": [ { "title": "Observe", "text": "Maintain an operating view relevant to the continuing mandate." }, { "title": "Decide within constraints", "text": "Relate a proposed action to the mandate, resource limits and consequences." }, { "title": "Return to the result", "text": "Retain operating observations for later evaluation. An experiment result and a fund outcome need distinct interpretation." } ] }, { "id": "research-role", "kind": "evidence", "title": "Consequences sharpen the engineering questions.", "intro": "Continuity, intervention, calibration and recovery matter when the system keeps operating between instructions.", "body": [ "The application gives the lab a continuing setting in which to examine how an autonomous process remains coherent as conditions, observations and responsibilities change." ] } ], "questions": [ { "question": "Can I invest through this page?", "answer": "Anemone is the lab’s own fund. This page provides a route for research and technical discussion, with no investment intake." }, { "question": "Does the site publish fund returns?", "answer": "The profile focuses on operating design and research context. Fund performance figures are not published here." } ], "relationships": [ { "type": "research", "id": "manna", "label": "Shared autonomy questions", "text": "Continuing purpose and regulation; no runtime dependency is asserted.", "url": "https://naevumlabs.com/products/manna" } ], "action": { "label": "Discuss consequential autonomy", "href": "https://naevumlabs.com/contact?context=anemone" } } }, { "id": "agency-engine", "name": "Agency Engine", "category": "Governed application foundation", "url": "https://naevumlabs.com/products/agency-engine", "description": "Agency Engine separates a shared kernel, service and interface composition, product identity and organization context. Distinct applications can specialize their experience around common controls and versioned contracts.", "applications": [ { "title": "Launch another product from the same core", "audience": "Product organizations", "situation": "A new application needs its own workflow, language and interface.", "mechanism": "Compose a product-owned shell and selected services above the sealed governance core.", "impact": "The team can change the experience without rebuilding approvals, evidence and custody controls." }, { "title": "Make an organizational installation", "audience": "Enterprises", "situation": "A common application must reflect local permissions and operating context.", "mechanism": "Installation and organization layers supply context while policy can only narrow inherited authority.", "impact": "Local specialization retains a comprehensible relationship to the foundation’s controls." }, { "title": "Let a useful method return to the platform", "audience": "Software teams", "situation": "A product discovers a mechanism that another product could use.", "mechanism": "Promote the reusable part through the core’s versioned interfaces while preserving its provenance.", "impact": "Shared engineering can improve without copying a product’s private business context." } ], "technical_features": [ { "name": "Runtime product composition", "description": "A resolved declaration connects tooling, backend and frontend. Invalid configuration has a refused state." }, { "name": "A protected control set", "description": "Policy, evidence, approvals, model access, disclosure and custody form a protected set around replaceable service contracts. Coverage is stated for the implementation paths." }, { "name": "A shared policy decision model", "description": "Default denial, overriding prohibitions, principal-bound approval, recipient constraints and autonomy ceilings give routes a common decision vocabulary." }, { "name": "Organization as a separate layer", "description": "Reusable product identity and methods are distinct from a specific organization’s people, client relationships and operational records." }, { "name": "Durable process design", "description": "The selected orchestration design connects adaptive planning to typed plans, durable execution and an observable workspace." }, { "name": "Explicit memory changes", "description": "Promotion, generalization, retention and typed forgetting follow declared rules whose usefulness can be evaluated separately." } ], "operating_context": null, "mechanism": [ { "name": "Preserve the core", "description": "Keep gates, ledger, capabilities, approvals and inference admission in the sealed governance set." }, { "name": "Compose the product", "description": "Validate the product declaration and select its domain services and interface vocabulary." }, { "name": "Tighten the context", "description": "Apply organization restrictions while assembling the work context. Compile typed steps, perform the selected method and retain its transitions." } ], "related": [ "https://naevumlabs.com/products/supersell", "https://naevumlabs.com/products/black-box" ], "editorial": { "lead": "One foundation. Products with a life of their own.", "audience_job": "Agency Engine is the shared application core for product teams building AI apps, so that each new app starts with approvals, records and permissions already in place.", "chapters": [ { "id": "product-layer", "kind": "comparison", "title": "A common core can become a very different experience.", "intro": "A new AI application needs its own workflows, language and interface. Rebuilding the shared controls for every product can leave each application with a different approach to permissions, evidence and recovery.", "items": [ { "title": "The kernel", "text": "Shared contracts and controls establish the common operating basis." }, { "title": "The runtime", "text": "Durable work objects, process transitions and execution ports carry the selected product into operation. Replaying recorded transitions reconstructs state without a new reasoning call." }, { "title": "The product", "text": "Selected services, interface composition and domain methods give an application its own behavior and vocabulary." }, { "title": "The organization", "text": "People, client relationships, mandates and operational records belong to the organization using the product." } ], "body": [ "Use Agency Engine when you need a distinct product or an organization-specific installation on a shared application foundation. Begin with the product’s work objects and operating responsibilities, then compose its services around the protected controls.", "SuperSell, built on the engine, organizes commercial conversations; a vertical installation for a law firm or a logistics operator would look nothing like it. Their users encounter distinct products on one governed core." ] }, { "id": "extension-seam", "kind": "mechanism", "title": "Make the extension seam explicit.", "intro": "A resolved product declaration travels from tooling through the backend to the frontend.", "body": [ "The application consumes that composition at runtime. Invalid configuration has a refused state rather than silently exposing the whole engine. Replaceable services use contracts around a protected set of policy, evidence, approvals, model access, disclosure and custody controls.", "The decision model can apply default denial, overriding prohibitions, principal-bound approval and recipient constraints. Coverage is defined for specific startup and execution paths; an application still needs its own threat model." ] }, { "id": "work-process", "kind": "process", "title": "A useful product carries work through a process.", "intro": "The selected orchestration design connects adaptive planning to a typed plan, durable execution and a workspace where people can inspect progress.", "items": [ { "title": "Compile the plan", "text": "Translate proposed work into typed steps and declared execution ports." }, { "title": "Retain the transition", "text": "A durable process record supports replay of recorded transitions and recovery." }, { "title": "Improve the memory", "text": "Promotion, retention and forgetting have explicit rules. Recurrence can suggest a useful lesson; evaluation determines whether that lesson improves future work." } ] }, { "id": "agent-studio", "kind": "mechanism", "title": "An agent has a version you can examine.", "intro": "The studio makes a created agent an inspectable product object, with a template, bounded settings and a recorded history.", "items": [ { "title": "Start from a contract", "text": "Strict templates and independent cases define the behavior expected from the configured agent. Injection canaries examine selected failure conditions." }, { "title": "Keep the earlier version", "text": "Replacing an agent retains the previous version. Restoring an earlier configuration creates a new version, preserving the intervening record." }, { "title": "Separate draft from deployment", "text": "Creating and testing an agent does not itself deploy it remotely. The target environment and its authority require their own release decision." } ] } ], "questions": [ { "question": "Must every product include every engine service?", "answer": "Product composition selects the services and interface that serve its domain. Shared governance and compatibility contracts define how that specialization can operate." }, { "question": "Can different organizations use the same product?", "answer": "The architecture separates reusable product identity from each organization’s people and records. Deployment, data isolation and available workflows are scoped for the intended release." } ], "relationships": [ { "type": "product-expression", "id": "supersell", "label": "Commercial assistance", "text": "A distinct product built from the engine.", "url": "https://naevumlabs.com/products/supersell" }, { "type": "delivery", "id": "black-box", "label": "Build the product", "text": "Architecture, integration and delivery around a selected application.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Discuss a product foundation", "href": "https://naevumlabs.com/contact?context=agency-engine" }, "example": { "title": "Change the product, retain the protected seam.", "intro": "Illustrative scenario. A shared foundation supports a sales application and an organizational workbench with different user experiences.", "states": [ { "id": "sales", "label": "Sales application", "title": "Product vocabulary follows commercial work.", "body": "The selected services connect business knowledge, offers and invited conversations.", "records": [ { "name": "Kernel", "value": "Shared contracts and protected governance set." }, { "name": "Runtime", "value": "Durable work transitions and execution ports." }, { "name": "Product", "value": "Commercial methods and the SuperSell experience." }, { "name": "Organisation", "value": "The seller’s people, relationships and current records." } ] }, { "id": "workbench", "label": "Work organization", "title": "A different product selects a different working method.", "body": "The workbench specializes briefs, procedures, review and delivery without turning every organization into an engine fork.", "records": [ { "name": "Product change", "value": "Replace the domain service through its contract." }, { "name": "Preserve", "value": "The sealed policy, effect and evidence boundary." }, { "name": "Organization change", "value": "Apply the client’s roles and procedures in its own layer." }, { "name": "Version history", "value": "Retain replaced agent versions; restoration creates a new version." } ] } ] } }, "anatomy": { "representation": "An illustrative process for composing product services and retaining organization work.", "signature": "Product composition and durable work", "parts": [ { "id": "foundation", "name": "Versioned foundation", "description": "Preserve shared contracts and the declared protected control set." }, { "id": "composition", "name": "Product composition resolver", "description": "Validate the declaration and select domain services and interface vocabulary." }, { "id": "context", "name": "Organization context", "description": "Supply organization records and restrictions independently from reusable product identity." }, { "id": "compiler", "name": "Plan and process compiler", "description": "Translate proposed work into typed steps, dependencies, budgets and available routes." }, { "id": "services", "name": "Contracted service workbench", "description": "Perform the selected domain method through its eligible execution ports." }, { "id": "record", "name": "Durable work and versions", "description": "Reconstruct work from transitions and keep replaced methods/agents identifiable." } ] } }, { "id": "supersell", "name": "SuperSell", "category": "Commercial intelligence & execution", "url": "https://naevumlabs.com/products/supersell", "description": "SuperSell’s invitation-led experience connects a person’s question to the business knowledge needed for a useful answer. Offers, context, consent and the next appropriate action belong to one continuing relationship.", "applications": [ { "title": "Answer the person who invited you in", "audience": "People and service businesses", "situation": "Someone asks for help in a private thread and expects a relevant answer.", "mechanism": "The conversation retrieves permitted personal context, the approved offer and earlier replies.", "impact": "The answer can address that person’s request while preserving their channel and contact preferences." }, { "title": "Keep a small team’s commercial memory", "audience": "Founders and small teams", "situation": "A prospect moves between a public page, a question and a private conversation.", "mechanism": "Person, offer, touch and conversation records share one commercial context.", "impact": "The team can continue the relationship without repeatedly reconstructing the business or the discussion." }, { "title": "Improve a campaign from its outcomes", "audience": "Commercial organizations", "situation": "More activity alone does not reveal which approach helps the audience.", "mechanism": "Registered experimental arms connect channel and creative choices to observed outcomes and stopping rules.", "impact": "The team can make an inspectable decision about what to continue, change or stop." } ], "technical_features": [ { "name": "Invitation-led assistance", "description": "The current experience direction begins with a person’s expressed need and an invitation into their conversation." }, { "name": "Source-aware business context", "description": "Offers and business assertions can retain origin, effective conditions and disagreements rather than becoming one undifferentiated prompt." }, { "name": "Two views of the relationship", "description": "The person’s question and the business’s review process are designed together so each can understand the proposed next step." }, { "name": "Communication with conditions", "description": "Proposed communication is related to a principal, recipient and the relevant consent and approval conditions." }, { "name": "A bounded first product", "description": "Modular product composition supports a useful core around one offer or journey before additional capabilities are introduced." } ], "operating_context": null, "mechanism": [ { "name": "Invite", "description": "A reply, invitation or consent establishes the permitted personal channel and its scope." }, { "name": "Help", "description": "Retrieve the person’s context and approved business material to compose a relevant response." }, { "name": "Continue", "description": "Retain the person’s need as business facts change and revise the proposal. Keep any permitted communication and observed reply with the conversation." } ], "related": [ "https://naevumlabs.com/products/agency-engine" ], "editorial": { "lead": "Your business context. A better offer. A conversation worth having.", "audience_job": "SuperSell is a sales application for businesses that answer customers directly, so that every answer uses today’s offers and capacity, and any commitment beyond approved policy waits for a person.", "chapters": [ { "id": "correct-the-mission", "kind": "process", "title": "Correct one assumption. Change the next conversation.", "intro": "An offer still reflects last month’s availability. Before the next customer conversation, the seller needs the proposal to use the business’s current capacity and corrected facts.", "items": [ { "title": "Correct the business memory", "text": "A seller replaces an outdated delivery assumption with the current capacity and supporting source.", "input": "An offer still based on last month’s availability.", "output": "Corrected context that informs the next proposal." }, { "title": "Rework the offer", "text": "The proposed offer and the next useful action use the corrected facts. Reasons, assumptions and expected cost remain available to the seller.", "output": "A draft that fits the actual capacity." }, { "title": "Invite the conversation", "text": "Listening, qualification, composition and checking have separate work and permissions. The person decides whether to continue the exchange.", "boundary": "A prepared message remains distinct from an authorized send and its observed result." } ] }, { "id": "the-invitation", "kind": "narrative", "title": "Begin with the person’s question.", "intro": "Someone is considering an offer. They want a relevant answer, a clear explanation and room to make their own decision.", "body": [ "SuperSell’s experience direction starts with an invitation into that conversation. Business knowledge supplies the context: what is offered, who it suits, what the terms mean and which next step would be useful.", "The personal space belongs to the person. Assistance should earn a continued place through relevance and consent. This invitation-led experience guides product development; its usefulness must be demonstrated with real interactions." ], "intro_role": "editorial-statement" }, { "id": "two-sides", "kind": "comparison", "title": "The same relationship, seen from both sides.", "intro": "A helpful exchange depends on what the person needs and what the business actually knows.", "items": [ { "title": "For the person", "text": "Ask about suitability, compare an offer’s conditions and request useful material. Decide whether to continue or involve a human." }, { "title": "For the business", "text": "Keep offers, conversations and follow-up connected to their sources. An answer should remain consistent with the current business context." }, { "title": "Across the relationship", "text": "Retain changes and conflicting assertions with their source and time. A previous answer should be explainable when an offer changes." } ] }, { "id": "next-step", "kind": "mechanism", "title": "A suggested response and a sent message are different events.", "intro": "Agency Engine supplies the product foundation for domain context and work processes.", "body": [ "Proposed communications are tied to a principal, recipient and the relevant consent and approval conditions. A business can examine what a response relies on before it becomes an external action.", "A useful first engagement can focus on one offer and one assistance journey. Evaluate resolved needs, accepted next steps and continued consent; message volume alone says little about whether a person was helped." ] }, { "id": "one-mission-one-effect", "kind": "mechanism", "title": "Two useful assistants should not send the same intention twice.", "intro": "Two work cells can prepare the next step for the same seller and channel. Coordination belongs to their shared mission.", "body": [ "A shared semantic lock relates the competing work to the same intention. Identity and channel rate accounting, together with a stable effect record, keep the proposals from becoming independent conflicting actions.", "A mission receipt distinguishes what was proposed, authorized and observed, alongside what remains unknown. A campaign-learning design fixes its population, observation window and mature trial definition before judging the result; it does not promise calibrated revenue advice." ] } ], "questions": [ { "question": "Is SuperSell an automatic outreach service?", "answer": "Its current experience direction is invitation-led assistance. Any commercial workflow needs explicit channel, recipient and consent conditions; the product should fit those relationships." }, { "question": "What can we explore first?", "answer": "Choose one customer question and the business knowledge needed to answer it. Prototype the person’s experience and the business review process together." } ], "relationships": [ { "type": "foundation", "id": "agency-engine", "label": "Shared product foundation", "text": "Domain services, memory and policy underpin the experience.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "delivery", "id": "black-box", "label": "Custom commercial workflow", "text": "Adapt the system to the offer, channels and responsibilities.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Explore SuperSell access", "href": "https://naevumlabs.com/join?product=supersell" }, "example": { "title": "A corrected fact changes the offer.", "intro": "Illustrative scenario. A seller has limited implementation capacity. An outdated assumption has reached a proposed offer.", "states": [ { "id": "assumption", "label": "Earlier assumption", "title": "The draft promises a date the business cannot support.", "body": "The proposal depends on an old capacity note in the business memory.", "records": [ { "name": "Business fact", "value": "The earlier note says a team is available this month." }, { "name": "Offer draft", "value": "Begin implementation immediately." }, { "name": "Conversation", "value": "No message has been authorized." }, { "name": "Useful check", "value": "Inspect the source behind the promised date." } ] }, { "id": "correction", "label": "Corrected context", "title": "The next proposal fits the actual capacity.", "body": "The seller corrects the source. Offer development and the proposed next conversation use the updated context.", "records": [ { "name": "Corrected fact", "value": "The next team is available next month." }, { "name": "Revised offer", "value": "Scope the work now; propose a later implementation date." }, { "name": "Next action", "value": "Invite a planning conversation with its assumptions and cost visible." }, { "name": "Observation", "value": "Record the person’s actual response separately from the proposed send." } ] } ] } }, "anatomy": { "representation": "An illustrative process for preparing and correcting an offer before communication.", "signature": "Invited relationship and grounded composition", "parts": [ { "id": "relationship", "name": "Relationship intake", "description": "Retain the expressed need, replies and invitation scope." }, { "id": "business", "name": "Business evidence and offer basis", "description": "Retrieve current offer facts and retain corrected assumptions." }, { "id": "composition", "name": "Offer and response workbench", "description": "Join the actual need with current evidence and assemble a suitable draft." }, { "id": "communication", "name": "Review and communication interface", "description": "Check grounding and applicable communication conditions under the shared mission." }, { "id": "outcomes", "name": "Conversation and outcome register", "description": "Keep proposal, authorized send and actual response distinct." } ] } }, { "id": "extrasensory", "name": "Extrasensory", "category": "General-purpose predictive & preventive intelligence", "url": "https://naevumlabs.com/products/extrasensory", "description": "Extrasensory connects operating histories and changing conditions to risk, uncertainty and intervention decisions. Its predictive approach carries the evidence basis and decision constraints into the response.", "applications": [ { "title": "Follow condition on an operating vessel", "audience": "Marine engineering", "situation": "Load, condition and sensor histories change while equipment remains in service.", "mechanism": "Exposure-aware diagnostics and time-to-event estimates connect the condition record to risk.", "impact": "The engineering team can investigate how operating context changes the evidence behind an assessment." }, { "title": "Start with a sparse failure history", "audience": "Industrial operators", "situation": "An asset has limited local failure data but belongs to an understood equipment class.", "mechanism": "Class priors and condition adjustments form an explicit cold-start basis.", "impact": "The operator can inspect what the estimate inherits and what the local observations contribute." }, { "title": "Compare the next useful intervention", "audience": "Asset-intensive organizations", "situation": "Inspection, continued operation and maintenance carry different practical constraints.", "mechanism": "Criticality and uncertainty inform a bounded comparison of available interventions.", "impact": "The decision can retain its assumptions and be revisited when subsequent evidence arrives." } ], "technical_features": [ { "name": "Time-to-event reasoning", "description": "Longitudinal histories distinguish censored observations, delayed entry, competing events and recurrent events. A common leakage policy governs which evidence is eligible for a prediction." }, { "name": "Condition-adjusted cold starts", "description": "Class-level priors and operating condition inform risk when local failure history is sparse. Inferred failure risk remains distinct from a scheduled maintenance countdown." }, { "name": "Load and condition are different signals", "description": "Multichannel diagnostics and load normalization distinguish changing operating regime from evidence of deterioration. Input alignment and exposure are part of the model’s meaning." }, { "name": "Calibration with explicit assumptions", "description": "Grouped partitions and finite-sample conformal ranks retain the basis of uncertainty estimates. Coverage and calibration are evaluated for the actual data and target population." }, { "name": "Prediction leads to a constrained choice", "description": "Criticality and operating context shape the comparison between observation, inspection and intervention. Subsequent evidence can update a bounded model without silently rewriting its decision limits." } ], "operating_context": { "label": "Running for months in a pilot installation", "description": "The predictive-preventive engine has been running for months in a pilot installation, learning from live operating data. Each new application has its own integration, data and acceptance." }, "mechanism": [ { "name": "Establish the history", "description": "Reconstruct dated events and operating exposure, distinguishing condition changes from load and unreliable sensing." }, { "name": "Estimate the risk", "description": "Combine class assumptions with eligible local evidence. Carry the calibration basis, applicability and forecast horizon with the estimate." }, { "name": "Choose an intervention", "description": "Use the retained estimate and operating constraints to choose monitoring, inspection or an intervention proposal. Compare independent later evidence with the original forecast." } ], "related": [ "https://naevumlabs.com/products/manna", "https://naevumlabs.com/products/vagus" ], "editorial": { "lead": "See the change before it becomes the problem.", "audience_job": "Extrasensory is a bio-inspired prediction system for fleet and plant engineers, so that they decide earlier whether to keep running, inspect or intervene, and know how sure the estimate is.", "chapters": [ { "id": "condition-change", "kind": "comparison", "title": "Keep operating. Inspect. Intervene.", "intro": "In an illustrative asset history, a condition signal rises while the operating load also changes. The useful question is which response the evidence supports.", "items": [ { "title": "Keep operating", "text": "Check whether the signal is explained by load and whether the observation is current. An unchanged schedule is not evidence of unchanged risk." }, { "title": "Inspect", "text": "Choose an observation that can reduce the uncertainty relevant to the decision." }, { "title": "Intervene", "text": "Compare the proposed action with operating constraints, its consequences and the responsible person’s authority." } ] }, { "id": "history-and-risk", "kind": "mechanism", "title": "The history contains more than failures.", "intro": "Operating duration, changing condition and incomplete observations all affect how a forecast should be interpreted.", "body": [ "Longitudinal histories include assets that remain in service and records that end before a failure is observed. Those censored observations matter. Sparse histories may begin with class priors whose assumptions remain visible.", "Condition and load can adjust the estimate, while uncertainty and applicability travel with it. A confidence score needs a calibration study before it can be read as a coverage guarantee. An idle asset does not automatically become a repaired asset." ] }, { "id": "decision-record", "kind": "process", "title": "Follow the forecast into the later outcome.", "intro": "Predictive operations need a record of what was known before the decision.", "items": [ { "title": "Record the forecast", "text": "Retain the observation window, horizon, uncertainty and model maturity." }, { "title": "Record the intervention", "text": "Distinguish a proposed action, an approved action and an applied action." }, { "title": "Evaluate prospectively", "text": "Examine false alarms, lead time, calibration and adjudicated outcomes. Avoided downtime needs a defensible comparison." } ] } ], "questions": [ { "question": "Does a forecast automatically control equipment?", "answer": "Prediction and intervention are separate. An engagement defines the decision workflow, permitted actions, operators and integration with the actual process." }, { "question": "How much history is needed?", "answer": "That depends on the asset class, data quality and question. Discovery establishes what can be estimated from existing evidence, which priors are needed and what must be measured next." } ], "relationships": [ { "type": "related", "id": "vagus", "label": "Outcome context", "text": "Source-aware histories and attention can support operational decisions.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "related", "id": "manna", "label": "Continuing observation", "text": "A persistent process can connect observations to a bounded response.", "url": "https://naevumlabs.com/products/manna" } ], "action": { "label": "Discuss a predictive operation", "href": "https://naevumlabs.com/contact?context=extrasensory" }, "example": { "title": "A warmer pump. What changed with it?", "intro": "Illustrative equipment case: a pump’s reported operating temperature rises while its operating load also changes. This is a decision workflow, not a fault diagnosis or a measured remaining-life estimate.", "states": [ { "id": "forecast", "label": "Forecast", "title": "The observation has two competing explanations.", "body": "The temperature change may track the changed load, or it may remain unexplained at comparable conditions. Preserve the load, sensor history and observation time before choosing an intervention.", "records": [ { "name": "Asset", "value": "A synthetic industrial pump." }, { "name": "Observed change", "value": "A rising reported operating-temperature trend while load changes." }, { "name": "Competing explanations", "value": "A load-related response or a persistent condition that the current context does not explain." }, { "name": "Current decision", "value": "Obtain a comparable-load observation and check the sensing path before inferring a fault." } ] }, { "id": "inspect", "label": "Inspection decision", "title": "The next observation changes the inspection decision.", "body": "Compare an aligned reading under an equivalent operating load, within the operator’s safe procedure. Check the sensor and operating context rather than treating a trend alone as a diagnosis.", "records": [ { "name": "If the difference resolves", "value": "Continue the agreed monitoring path and retain the explanation with its evidence." }, { "name": "If the difference persists", "value": "Escalate for the operator’s equipment inspection and intervention decision." }, { "name": "If the reading is unreliable", "value": "Investigate the sensing path; do not convert missing evidence into a healthy condition." }, { "name": "Later evaluation", "value": "Keep the earlier estimate, inspection and observed outcome separate for prospective evaluation." } ] } ] } }, "anatomy": { "representation": "An illustrative equipment decision-support process.", "signature": "Exposure, diagnosis and a prospective estimate", "parts": [ { "id": "history", "name": "Exposure and event-history reconstruction", "description": "Assemble eligible longitudinal history without turning incomplete observation into a failure event." }, { "id": "diagnostics", "name": "Load, condition and sensing comparator", "description": "Distinguish a load-related response, persistent difference and unreliable sensing." }, { "id": "estimator", "name": "Prior and local-evidence estimator", "description": "Combine an explicit class basis with eligible local evidence while retaining maturity and horizon." }, { "id": "qualification", "name": "Calibration and applicability", "description": "Carry calibration basis and applicability with the estimate rather than displaying an unconditional confidence score." }, { "id": "decision", "name": "Constrained decision workbench", "description": "Select useful monitoring, inspection or an intervention proposal from the actual unresolved question and constraints." }, { "id": "prospective-record", "name": "Forecast and prospective-outcome register", "description": "Freeze the estimate before the decision, then compare genuinely later independent evidence." } ] } }, { "id": "metahive", "name": "Metahive", "category": "Research, compute & Tier 1", "url": "https://naevumlabs.com/products/metahive", "description": "Metahive connects compute services to the sites and resources that support them. Workload admission, usage identity, energy attribution and operating responsibility give capacity a concrete setting.", "applications": [ { "title": "Bring a workload into a compute service", "audience": "Enterprises and research institutions", "situation": "A workload needs the right capacity, data boundary and result criteria.", "mechanism": "Admission binds a task contract to eligible resources and a workload-specific verifier.", "impact": "The service can account for accepted useful work rather than equating resource time with a correct result." }, { "title": "Connect several physical sites", "audience": "Infrastructure operators", "situation": "Distributed capacity has different energy, equipment and operating conditions.", "mechanism": "Site constraints, fenced attempts and common job records connect each contribution to its origin.", "impact": "Operators can investigate where work ran and how its result was accepted." }, { "title": "Plan capacity around real demand", "audience": "Institutional compute buyers", "situation": "A hardware list says little about the inference workload the institution needs.", "mechanism": "Workload and facility models connect memory, throughput, power and cooling.", "impact": "Procurement discussions can expose the physical assumptions behind the proposed service.", "link": "/products/metahive/builder" } ], "technical_features": [ { "name": "Admission before workload", "description": "Tenant and site budgets reserve the maximum admitted cost before dispatch and settle against the observed result." }, { "name": "Usage with an identity", "description": "Tenant, product, unit and time connect usage to the delivered service. Reconciliation examines scheduler and service records." }, { "name": "An explicit energy allocation", "description": "Hourly metered site energy and tenant IT use support an allocation method. Estimates and allocations remain labeled as such." }, { "name": "Deployment tiers with prerequisites", "description": "Dedicated endpoints, managed capacity and stronger custody designs have different physical and operating requirements." }, { "name": "Site and network as distinct scopes", "description": "Hardware, energy, hosting and the distributed platform retain their own design and delivery responsibilities." } ], "operating_context": null, "mechanism": [ { "name": "Locate capacity", "description": "Combine compute, power, cooling, network and operating responsibilities into a candidate service envelope." }, { "name": "Connect the system", "description": "Match workload and custody requirements to eligible capacity, then reserve a bounded cost under one operation identity." }, { "name": "Plan useful work", "description": "Reconcile observed service use with the original reservation. Attribute site-hour energy using matching workload evidence and retain idle energy." } ], "related": [ "https://naevumlabs.com/products/metahive/builder" ], "editorial": { "lead": "Useful compute. A clear place in the world.", "audience_job": "Metahive is the research, compute and Tier 1 line of Naevum Labs for organizations and hosts, so that a team’s folders become a working environment on infrastructure it can name.", "chapters": [ { "id": "research", "kind": "narrative", "title": "Research with a physical goal.", "intro": "Metahive has a research track prepared for European funding.", "intro_role": "editorial-statement", "body": [ "What survives testing goes into the Vault, Naevum’s method and component atlas, with what it assumes and what proved it, before it becomes an organ or a program." ] }, { "id": "sites-bridge-hosts", "kind": "mechanism", "title": "Sites, a bridge and a network of hosts.", "intro": "Metahive designs container micro data centers, starts on rented EU servers, and operates sealed units that hosts own. A workload keeps one identity from reservation to energy account.", "items": [ { "title": "Container sites", "text": "Micro data centers in standard containers, planned with Metahive Builder. The sites are in development." }, { "title": "Managed bridge", "text": "Rented EU servers that Metahive operates carry the first workloads while the sites are built." }, { "title": "Host units", "text": "Standard sealed units that hosts own and Metahive operates. Idle capacity is sold as preemptible work." }, { "title": "Trust tiers", "text": "Standard for ordinary data. Custody runs on confidential-computing hardware behind a private endpoint per customer, with open-weight models first." }, { "title": "Reserve before dispatch", "text": "Before a job starts, Metahive reserves the most it may cost against the customer’s and the site’s budgets." }, { "title": "Settle against the result", "text": "When the job ends, its usage (customer, product, unit and time) is reconciled with the scheduler’s and the service’s own records." }, { "title": "Explain the energy", "text": "Hourly site metering and each customer’s IT use support a stated allocation method. Allocated or estimated energy is labeled as such." } ], "body": [ "Idle consumption stays in the site’s energy account instead of disappearing from a workload-only total. Placement, custody and energy answer three different questions: where work runs, which boundary holds its data, and how measured site energy is assigned." ] }, { "id": "tier-1", "kind": "process", "title": "The tier every client starts with.", "intro": "Tier 1 is included with the stack. An organization shares its folders and gets a working environment; the full list of capabilities and their states is on the platform page.", "items": [ { "title": "Share a folder", "text": "Point Metahive at the folders you already have. It reads the text files and records the path and date of each one. PDF and Office files are in development." }, { "title": "Ask the knowledge base", "text": "A knowledge base and a memory are built from the index, and every answer leads back to its files." }, { "title": "Approve the programs", "text": "Metahive proposes programs, skills and workflows. A person approves each one before it is installed." }, { "title": "Connect your agent tools", "text": "Command-line coding agents connect through one governed gateway. Connectors for the Claude app, the ChatGPT app and VS Code are in development." }, { "title": "Check the record", "text": "Every consequential action lands in a hash-chained, Merkle-anchored, timestamped record that you can verify." } ] } ], "questions": [ { "question": "Can we request capacity or discuss hosting?", "answer": "Yes. Describe the workload or site and the location, custody and operating requirements. The discussion establishes what service or development scope is available." }, { "question": "Is Builder a running data center?", "answer": "Builder models proposed configurations and operating scenarios. It supports planning; a site’s construction, commissioning and service operation are separate stages." } ], "relationships": [ { "type": "uses", "id": "metahive-builder", "label": "Plan the physical system", "text": "Explore equipment, capacity and energy before commitment.", "url": "https://naevumlabs.com/products/metahive/builder" }, { "type": "related", "id": "synapse", "label": "Design the AI boundary", "text": "Combine infrastructure choices with context and effect controls.", "url": "https://naevumlabs.com/products/synapse" } ], "action": { "label": "Discuss compute or hosting", "href": "https://naevumlabs.com/contact?context=metahive" }, "example": { "title": "A workload has a place and an energy account.", "intro": "Illustrative scenario. A sensitive institutional workload is evaluated against tenancy, custody and physical resource constraints.", "states": [ { "id": "placement", "label": "Placement", "title": "Select a site only after the context boundary is defined.", "body": "Available compute does not by itself establish an admissible place for the work.", "records": [ { "name": "Request", "value": "A workload with declared resources and sensitivity." }, { "name": "Custody", "value": "The selected tenancy and observation boundary." }, { "name": "Admission", "value": "Reserve the admitted operation and cost before dispatch." }, { "name": "Placement", "value": "Choose an eligible physical execution site." } ] }, { "id": "accounting", "label": "Reconciliation", "title": "Keep workload use, facility overhead and idle energy distinct.", "body": "Energy attribution follows the tenant, site and hour. Uncertainty in a returned result remains visible until reconciled.", "records": [ { "name": "IT use", "value": "Attribute the workload’s observed resource consumption." }, { "name": "Facility overhead", "value": "Retain cooling and other site requirements in the account." }, { "name": "Idle energy", "value": "Keep unattributed consumption explicit under the selected allocation method." }, { "name": "Settlement", "value": "Reconcile observed work and the original reservation under the same identity." } ] } ] } }, "anatomy": { "representation": "An illustrative process for comparing site constraints and reconciling identified resource use.", "signature": "Workload placement and site-hour accounting", "parts": [ { "id": "site", "name": "Site and service model", "description": "Combine compute, power, cooling, network and operating responsibilities into a candidate envelope." }, { "id": "placement", "name": "Workload and custody matcher", "description": "Compare required resources and sensitivity with an eligible service boundary." }, { "id": "reservation", "name": "Reservation and operation register", "description": "Bind maximum admitted cost before work under one identity." }, { "id": "usage", "name": "Service-use reconciliation", "description": "Compare scheduler/service evidence and settle the original reservation." }, { "id": "energy", "name": "Meter and allocation workbench", "description": "Attribute site-hour metering using matching IT use without hiding idle energy." } ] } }, { "id": "metahive-builder", "name": "Metahive Builder", "category": "3D micro data center design & simulation", "url": "https://naevumlabs.com/products/metahive/builder", "description": "Metahive Builder couples a 3D site hierarchy with electrical, thermal, energy and inference scenarios. A configuration change reveals its constraints, and each calculation retains its units, method and input lineage.", "applications": [ { "title": "Design the datacenter before the fitout", "audience": "Datacenter designers", "situation": "Racks must fit the room while power, cooling and floor constraints remain credible.", "mechanism": "Nested geometry checks connect equipment placement to its building and plot.", "impact": "A layout can reveal a physical conflict before it is mistaken for an available capacity." }, { "title": "See what changes with another model", "audience": "AI infrastructure teams", "situation": "A model’s precision and context length change memory and inference requirements.", "mechanism": "Weights, KV cache, parallelism and rooflines propagate the workload into the configuration.", "impact": "The team can compare a workload envelope instead of buying from accelerator count alone." }, { "title": "Compare an energy and investment scenario", "audience": "Operators and investment planners", "situation": "A denser design changes electrical demand, heat, annual energy and cost assumptions.", "mechanism": "Power and thermal propagation share a scenario basis with annual weather, supply and storage.", "impact": "The decision retains units, input quality and sensitivity instead of presenting one unexplained ROI figure." } ], "technical_features": [ { "name": "Device → rack → room → building → plot", "description": "Nested placement gives every device a physical context. Containment, equipment fit, aisle clearance, floor load and site constraints can be checked as the configuration changes." }, { "name": "Power propagates into heat", "description": "Nameplate, provisioned peak and utilization are different power bases. PSU losses, IT heat, cooling and facility demand are propagated through the same scenario." }, { "name": "Inference has a memory and throughput bound", "description": "Model weights, KV cache, precision, tensor parallelism and compute/memory rooflines determine the workload envelope. Installed accelerator count alone cannot describe useful capacity." }, { "name": "A year can change the design", "description": "Weather, energy supply, storage and cooling assumptions shape annual scenarios. Compare design choices on the same input basis and expose the assumptions that dominate the result." }, { "name": "Every result has units and provenance", "description": "Inputs retain their source, quality grade and method. Rated values, assumptions and measured observations remain distinguishable inside a visually precise design." }, { "name": "Engineering and investment stay connected", "description": "A physical configuration can inform capital, operating-cost and sensitivity scenarios. Financial planning retains its own assumptions; a detailed model does not turn an estimate into a promised return." } ], "operating_context": null, "mechanism": [ { "name": "Place", "description": "Configure devices, racks, rooms, buildings and a plot, with containment, aisle and floor-load checks." }, { "name": "Propagate", "description": "Follow IT load through electrical losses and heat to cooling, facility demand and energy supply." }, { "name": "Compare", "description": "Evaluate compute precision, model-memory limits, inference rooflines and scenario ranges on a common basis." } ], "related": [ "https://naevumlabs.com/products/metahive" ], "editorial": { "lead": "Move one rack. See what else changes.", "audience_job": "Metahive Builder is a 3D planning tool for compute-site planners, so that they see whether space, power, cooling or memory runs out first, before buying hardware.", "chapters": [ { "id": "rack-to-site", "kind": "mechanism", "title": "A placement changes more than the floor plan.", "intro": "A planned rack move can change equipment fit, electrical loading and cooling across the room. The planner needs to see those consequences while comparing the layout and the workload it is meant to support.", "items": [ { "title": "Device → rack", "text": "Check equipment fit, weight, clearances and electrical demand." }, { "title": "Rack → room", "text": "Follow cooling, airflow, floor loading and containment across neighboring equipment." }, { "title": "Room → site", "text": "Connect the building and plot to power paths, land use and operating assumptions." } ], "body": [ "Use Builder while evaluating a fitout, a different device mix or a changed inference workload. Start with the physical hierarchy and sourced assumptions, then inspect which constraints change as the configuration changes. Its engineering and financial scenarios retain their input quality and assumptions." ] }, { "id": "energy-and-inference", "kind": "comparison", "title": "Two limits meet in the same configuration.", "intro": "A compute design must support both its physical demand and its intended workload.", "items": [ { "title": "The energy model", "text": "Nameplate rating, provisioning peak, component load and utilization-dependent energy remain separate. PSU losses, redundancy and cooling add their own requirements." }, { "title": "The inference model", "text": "Weights, KV cache, tensor-parallel placement, batch and concurrency determine memory needs. Compute and memory rooflines bound the workload a configuration can support." }, { "title": "The operating day", "text": "Hourly scenarios connect generation, storage, grid caps and cooling. A favorable design-point ratio cannot stand in for an annual energy quantity." } ] }, { "id": "compare-designs", "kind": "process", "title": "Compare the change under the same assumptions.", "intro": "Consider replacing a rack configuration while keeping the workload and ambient scenario fixed.", "items": [ { "title": "Change one input", "text": "Update equipment or workload and retain the comparison basis." }, { "title": "Find the binding constraint", "text": "Inspect the electrical, thermal or memory limit that prevents the intended operation." }, { "title": "Open the calculation", "text": "A result keeps its units, method, input lineage and source quality. Assumptions and defaults remain available for scrutiny." } ] } ], "questions": [ { "question": "Are the inference figures measured benchmarks?", "answer": "Builder produces calculated bounds and scenarios from its declared models and inputs. Hardware benchmarking is a separate activity." }, { "question": "Can Builder guarantee project returns?", "answer": "It informs technical feasibility and operating assumptions. Financial feasibility also requires independently checked costs, demand, financing and contractual terms." } ], "relationships": [ { "type": "foundation", "id": "metahive", "label": "Infrastructure program", "text": "A planning tool within the wider compute and site direction.", "url": "https://naevumlabs.com/products/metahive" }, { "type": "related", "id": "synapse", "label": "Deployment constraints", "text": "Custody and model access can become design requirements.", "url": "https://naevumlabs.com/products/synapse" } ], "action": { "label": "Discuss a site scenario", "href": "https://naevumlabs.com/contact?context=metahive-builder" }, "example": { "title": "One changed workload. A traceable design decision.", "intro": "A synthetic inference scenario follows concurrency into KV memory, a capacity choice and a one-hour energy scenario. Every numerical input is an assumption; none is a hardware benchmark.", "states": [ { "id": "site", "label": "Initial workload", "title": "The first workload fits the assumed memory domain.", "body": "Sixteen sequences require 40 GiB. The selected 48 GiB option fits this memory calculation.", "records": [], "artifact": { "kind": "inference", "inputs": { "layers": 32, "kv_heads": 8, "head_dimension": 128, "bytes_per_element": 2, "context_tokens": 8192, "weight_gib": 20, "workspace_gib": 4, "other_it_kw": 0.45, "facility_ratio": 1.2, "feed_limit_kw": 1.3, "cooling_limit_kw": 1.4, "concurrency": 16, "capacity_gib": 48, "device_kw": 0.55 } } }, { "id": "workload", "label": "More concurrency", "title": "The same option no longer fits.", "body": "Thirty-two sequences need 56 GiB against 48 GiB. The workload remains unresolved; no successful operating-energy result is assigned.", "records": [], "artifact": { "kind": "inference", "inputs": { "layers": 32, "kv_heads": 8, "head_dimension": 128, "bytes_per_element": 2, "context_tokens": 8192, "weight_gib": 20, "workspace_gib": 4, "other_it_kw": 0.45, "facility_ratio": 1.2, "feed_limit_kw": 1.3, "cooling_limit_kw": 1.4, "concurrency": 32, "capacity_gib": 48, "device_kw": 0.55 } } }, { "id": "constraints", "label": "Capacity option", "title": "A separate equipment assumption changes the scenario.", "body": "The assumed 80 GiB option fits the memory requirement. Its independently assumed design power is higher; the feed, cooling and facility scenario are checked again.", "records": [], "artifact": { "kind": "inference", "inputs": { "layers": 32, "kv_heads": 8, "head_dimension": 128, "bytes_per_element": 2, "context_tokens": 8192, "weight_gib": 20, "workspace_gib": 4, "other_it_kw": 0.45, "facility_ratio": 1.2, "feed_limit_kw": 1.3, "cooling_limit_kw": 1.4, "concurrency": 32, "capacity_gib": 80, "device_kw": 0.8 } } } ] } }, "anatomy": { "representation": "An illustrative design calculation under explicit equipment and workload assumptions.", "signature": "Containment, physical limits and workload calculation", "parts": [ { "id": "configuration", "name": "Hierarchical configuration", "description": "Place devices within racks, rooms, buildings and plot with sourced assumptions." }, { "id": "physical", "name": "Electrical and thermal propagation", "description": "Keep distinct power bases and propagate losses, heat and cooling demand." }, { "id": "inference", "name": "Inference workload model", "description": "Derive memory and throughput bounds from the workload and selected equipment." }, { "id": "energy", "name": "Time and energy scenario", "description": "Balance load, supply, storage and grid under the selected time basis." }, { "id": "comparison", "name": "Constraint and alternative comparison", "description": "Join physical and workload results and expose the binding constraint." } ] } }, { "id": "conductr", "name": "Conductr", "category": "Coding-agent work coordination", "url": "https://naevumlabs.com/products/conductr", "description": "Conductr is a component-based harness for AI-assisted development. Context, planning, execution, verification and human checkpoints have inspectable roles, with experiments for evaluating changes to the process.", "applications": [ { "title": "Understand an agent’s working behavior", "audience": "Individual developers", "situation": "An agent follows a process whose promises are otherwise difficult to inspect.", "mechanism": "Observation hooks compare behavior with declared promises and retain telemetry.", "impact": "The developer gains a concrete record for investigating recurring process failures." }, { "title": "Test a proposed harness improvement", "audience": "AI engineering teams", "situation": "A change appears promising but could weaken behavior elsewhere.", "mechanism": "The offline measure harness compares it against a RED-baseline corpus.", "impact": "The change has an explicit experiment rather than a success claim based on one attractive output." }, { "title": "Replace a component without a new harness", "audience": "Developer-tool builders", "situation": "Context, verification or routing needs a different implementation.", "mechanism": "Interface-bounded components separate behavior from the assembly’s private policy and wiring.", "impact": "A team can investigate one mechanism while preserving the responsibilities of the surrounding system." } ], "technical_features": [ { "name": "Component slots with contracts", "description": "Context acquisition, planning, orchestration, execution, routing and verification have explicit roles in the harness." }, { "name": "Acceptance outside the author’s report", "description": "Externally authored acceptance conditions give verification a basis independent of a worker claiming success." }, { "name": "Several kinds of verification", "description": "Standing, structural and behavioral checks cover different questions. Coverage is assessed against a declared denominator." }, { "name": "Observation and steering", "description": "Telemetry and live-loop control are separate concerns. Advisory pacing can respond to the highest-risk area touched." }, { "name": "Paired harness experiments", "description": "A declared task and acceptance basis support comparison of a particular harness change. Optional containment keeps its actual scope." } ], "operating_context": null, "mechanism": [ { "name": "Frame", "description": "Give the implementation attempt an explicit objective and operating context." }, { "name": "Review", "description": "Compare the result with the intended change in a separate review step." }, { "name": "Continue", "description": "Retain specific findings and the next attempt’s context. Compare paired results and resource costs when the observations are available." } ], "related": [ "https://naevumlabs.com/products/black-box" ], "editorial": { "lead": "Give AI development a working discipline.", "audience_job": "Conductr is a coordination harness for engineering leaders whose coding agents share one repository, so that only work checked against tests the agent did not write reaches the main branch.", "chapters": [ { "id": "one-change", "kind": "process", "title": "Carry one change from context to verification.", "intro": "A team is changing its agent harness and needs to understand how the change affects the work. Context acquisition, execution and acceptance need to be examined separately from the agent’s own report of success.", "items": [ { "title": "Acquire context", "text": "Select the source and dependencies relevant to the requested change." }, { "title": "Declare acceptance", "text": "Define the acceptance conditions outside the worker’s own success report." }, { "title": "Run and examine", "text": "A bounded attempt returns a change and verification evidence. The next decision follows the actual findings." } ], "body": [ "Use Conductr to inspect an agent’s working process or compare a harness component on a declared task. Begin with the relevant source and an independently defined acceptance condition; keep the task and resource basis fixed when comparing attempts." ] }, { "id": "replaceable-slots", "kind": "mechanism", "title": "Inspect the harness one component at a time.", "intro": "Change one component while keeping the task, seed, acceptance oracle and resource budget fixed.", "body": [ "Contract-defined slots cover context, planning, execution, sandboxing, routing, credentials, verification, policy, memory and measurement. The fixed orchestration order lets a team isolate a component’s contribution.", "Independent test authority and a commit-boundary gate examine the artifact separately from the executor’s confidence. Standing, structural and behavioral checks retain their own scope; coverage needs a declared denominator." ] }, { "id": "prove-the-change", "kind": "evidence", "title": "Evaluate a harness change as an experiment.", "intro": "A paired experiment can hold the task and acceptance basis steady while changing one part of the development process.", "body": [ "The proving-ground direction makes costs, correctness and failure modes inspectable for a particular comparison. A containment integration is evaluated against the specific routes it is intended to control.", "Discuss the integration you have in mind, and we will scope it with you." ] }, { "id": "degraded-component", "kind": "constraints", "title": "A fallback is part of the component contract.", "intro": "A harness must state what happens when an integration is unavailable or a component cannot complete its work.", "body": [ "The selected degradation behavior belongs to the specific integration and execution policy. An experiment exercises that degraded route alongside the successful path, keeping the task, oracle, seed and budget fixed.", "A harness fallback is not automatically a privacy boundary. Conductr’s execution/degradation rules and Synapse’s context-admission rules have distinct responsibilities and evaluation cases." ] } ], "questions": [ { "question": "Must Black Box run on Conductr?", "answer": "Conductr contributes development-harness ideas and tooling. Black Box’s role and delivery architecture is a separate system; a mandatory runtime dependency is not asserted." }, { "question": "Can we use a different provider or verification component?", "answer": "The component architecture defines those seams. A concrete integration must check provider capability, custody, acceptance and the behavior of the replacement." } ], "relationships": [ { "type": "related", "id": "black-box", "label": "Delivery architecture", "text": "Independent challenge informs the larger factory process.", "url": "https://naevumlabs.com/products/black-box" }, { "type": "research", "id": "vault", "label": "Reusable methods", "text": "Component evaluation and evidence can inform reuse.", "url": "https://naevumlabs.com/products/vault" } ], "action": { "label": "Discuss development workflow", "href": "https://naevumlabs.com/contact?context=conductr" }, "example": { "title": "Replace one component. Keep the experiment fixed.", "intro": "Illustrative scenario. A harness change can be studied without changing the task or declaring an unsupported performance win.", "states": [ { "id": "baseline", "label": "Baseline", "title": "Record the reference execution conditions.", "body": "An independent evaluation defines the unit of work and expected outcome before comparing alternatives.", "records": [ { "name": "Fixed inputs", "value": "Task corpus, seed, budget and acceptance oracle." }, { "name": "Selected component", "value": "The context assembly slot." }, { "name": "External check", "value": "Independent tests judge the work artifact." }, { "name": "Measurement", "value": "Useful completion, failures and resource cost." } ] }, { "id": "alternate", "label": "Alternative", "title": "A component change has a bounded comparison.", "body": "The alternate configuration runs under the same rules. Its observed outcome determines whether the change helps.", "records": [ { "name": "Change", "value": "Replace only the context component through its contract." }, { "name": "Preserve", "value": "The orchestration order and authority boundaries." }, { "name": "Compare", "value": "Report the actual paired results and uncertainty." }, { "name": "Commit boundary", "value": "A successful execution still passes the applicable commit gate." } ] } ] } }, "anatomy": { "representation": "An illustrative work harness with a fixed task, oracle and retained findings.", "signature": "Controlled experiment and independent examination", "parts": [ { "id": "reference", "name": "Task and acceptance reference", "description": "Fix the experiment basis outside the worker success report." }, { "id": "context", "name": "Context and planning", "description": "Select relevant dependencies and form bounded work through explicit component contracts." }, { "id": "execution", "name": "Ordered execution", "description": "Carry work through the selected execution/routing interfaces." }, { "id": "examination", "name": "Independent examination", "description": "Judge the identified artifact with distinct verification kinds." }, { "id": "experiment", "name": "Experiment and continuation", "description": "Retain actual paired results, resource cost and the next decision." } ] } }, { "id": "vault", "name": "The Vault", "category": "Evidence-led research & invention", "url": "https://naevumlabs.com/products/vault", "description": "The Vault organizes reusable engineering as components, methods, contracts and evidence. Composition carries unresolved assumptions into the assembled system, where a test can determine whether reuse is justified.", "applications": [ { "title": "Find the mechanism before building again", "audience": "R&D and software teams", "situation": "A new problem may overlap earlier engineering in another project.", "mechanism": "Typed component and method registers retain contracts, scientific references and lineage.", "impact": "The team can inspect a candidate and the evidence behind its reuse." }, { "title": "Challenge an invention early", "audience": "Research organizations", "situation": "A plausible idea may reproduce known prior art or fail under a simple counterexample.", "mechanism": "Prior-art search, negative controls and evidence grades stay attached to the technical claim.", "impact": "Research effort can move toward a discriminating experiment before the idea becomes a product promise." }, { "title": "Carry knowledge between projects", "audience": "Institutions", "situation": "A useful engineering result should survive its original author and project.", "mechanism": "The selected method retains inputs, contract, limitations and its experimental record.", "impact": "A later team receives a technical argument it can inspect and reproduce." } ], "technical_features": [ { "name": "Scientific evidence travels with the method", "description": "A mechanism retains its scientific references or the explicit search that found none. Failed searches, counterexamples and refutations remain useful research outcomes." }, { "name": "Components and methods are different registers", "description": "One register describes what a system does; another describes how work is performed. Typed identities, contracts and lineage connect candidates without confusing a workflow with executable capability." }, { "name": "Evidence has a promotion ladder", "description": "Assertion, available artifact, functional evidence, independent reproduction and replication carry different grades. A negative control tests that the proposed verification can actually fail." }, { "name": "Composition preserves the argument", "description": "Candidate discovery, compatibility checks and discriminating experiments connect a question to a reusable result. The reason a component was selected remains available to the next project." }, { "name": "Invention is a technical claim", "description": "Mechanism, technical effect, prior art and a deliberate disclosure decision form the invention path. Candidate claims, filed applications and granted rights remain distinct records." } ], "operating_context": null, "mechanism": [ { "name": "Investigate", "description": "Connect a question to scientific references, prior work and explicit search results." }, { "name": "Challenge", "description": "Keep evidence grades, counterexamples and negative controls with the mechanism." }, { "name": "Compose", "description": "Compare interfaces, timing and environmental assumptions before combining methods and components. Carry unresolved obligations into the target experiment with their evidence." } ], "related": [ "https://naevumlabs.com/products/black-box", "https://naevumlabs.com/products/vagus" ], "editorial": { "lead": "Build on more than a promising idea.", "audience_job": "The Vault is a method and component atlas for heads of engineering, so that a team reusing a mechanism inherits its assumptions and turns each open one into a test.", "chapters": [ { "id": "engineering-question", "kind": "narrative", "title": "What can we reuse, and what must still be proved?", "intro": "A team wants to combine a temporal memory component with a workflow. Matching function signatures is only the beginning.", "body": [ "The Vault connects components, methods, contracts, evidence and lineage. A component describes a mechanism. A method describes how work is performed. Both need enough context for another team to decide whether reuse is justified.", "Negative findings belong beside successful candidates. A failed experiment can save another team from accepting an assumption that did not survive testing." ], "intro_role": "editorial-statement" }, { "id": "composition-obligations", "kind": "mechanism", "title": "A connection can carry an unfinished obligation.", "intro": "Typed ports describe interface compatibility; assume/guarantee contracts describe the environment that the component needs.", "items": [ { "title": "Inspect the interface", "text": "Check what a port accepts and produces, including cadence, absence and ordering where relevant." }, { "title": "Carry the assumption", "text": "A composition preserves obligations its parts have not discharged. Connecting two APIs cannot erase those requirements." }, { "title": "Turn the gap into a test", "text": "An unresolved timing or delivery assumption becomes a specific experiment against the intended environment." } ] }, { "id": "reuse-and-adoption", "kind": "evidence", "title": "The next use is part of the evidence.", "intro": "A catalog entry becomes more useful when another system can reproduce its effect and adopt it under stated conditions.", "body": [ "Recurring mechanisms across independently developed systems can suggest candidates for reuse. Their novelty, assumption independence and downstream value remain questions to examine.", "Research, an accepted component and a patent have different evidential and legal meanings. Public examples use cleared material, and adoption is described only where an actual downstream use can be identified." ] } ], "questions": [ { "question": "Is the Vault a document archive?", "answer": "Documents support it, but its organizing units are mechanisms, methods, contracts and evidence. The goal is to make engineering knowledge usable in another system." }, { "question": "What can a research engagement deliver?", "answer": "A technical question, source-backed alternatives, a falsifiable experiment, a reusable artifact with its contract, and an adoption decision grounded in the result." } ], "relationships": [ { "type": "delivery", "id": "black-box", "label": "From component to system", "text": "Delivery can integrate a candidate under explicit acceptance conditions.", "url": "https://naevumlabs.com/products/black-box" }, { "type": "related", "id": "agency-engine", "label": "Product composition", "text": "Reusable services need compatible contracts and upgrade boundaries.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Discuss a research question", "href": "https://naevumlabs.com/contact?context=vault" }, "example": { "title": "Two compatible types can still leave an obligation.", "intro": "Illustrative scenario. A sensor component and a controller agree on units but disagree about timing.", "states": [ { "id": "composition", "label": "Initial composition", "title": "The output type fits. The cadence does not.", "body": "A typed connection alone cannot establish that a controller receives sufficiently fresh observations.", "records": [ { "name": "Producer guarantee", "value": "Emit a temperature observation at its declared cadence." }, { "name": "Consumer assumption", "value": "Receive an observation within a shorter freshness window." }, { "name": "Unresolved obligation", "value": "The composition must account for delayed or missing input." }, { "name": "Decision", "value": "Keep compatibility unresolved rather than declaring the connection accepted." } ] }, { "id": "adapter", "label": "Revised contract", "title": "An explicit adapter changes what the consumer can rely on.", "body": "The proposed assembly documents stale-state behavior and the remaining assumptions before target acceptance.", "records": [ { "name": "Adapter", "value": "Expose observation age and an explicit absent state." }, { "name": "Consumer change", "value": "Hold action when the freshness condition is not met." }, { "name": "Propagated assumption", "value": "The target operator supplies the agreed sensing cadence." }, { "name": "Acceptance", "value": "Test absence and delayed-input behavior in the intended target." } ] } ] } }, "anatomy": { "representation": "An illustrative process for investigating claims, comparing contracts and retaining unresolved obligations.", "signature": "Evidence shelves and contract composition", "parts": [ { "id": "research", "name": "Question and research trace", "description": "Retain the question, relevant prior work and explicit negative search results." }, { "id": "register", "name": "Component and method register", "description": "Discover typed candidates while distinguishing capability from working method." }, { "id": "evidence", "name": "Evidence and falsifier bench", "description": "Grade the specific claim and test whether the proposed verification can fail." }, { "id": "composition", "name": "Composition and obligation bench", "description": "Compare typed interfaces, QoS and environmental assumptions." }, { "id": "experiment", "name": "Target experiment and adoption", "description": "Test the gap in the intended setting and retain the actual result for reuse." } ] } }, { "id": "everbloom", "name": "Everbloom: Crown of the Day", "category": "A shared, consequential game world", "url": "https://naevumlabs.com/products/everbloom", "description": "A King and Queen shape Everbloom’s shared fairytale kingdom. Decisions cascade through player roles, validated campaigns change the world, and durable state and the Chronicle connect one reign to the next.", "applications": [ { "title": "Rule a kingdom together", "audience": "Players", "situation": "A King chooses a response to a threat while a Queen shapes the kingdom’s cultural life.", "mechanism": "Their decisions branch into tasks for court roles and clubs, with shared consequences.", "impact": "One decision becomes an activity people can undertake together in a changing place." }, { "title": "Arrive before the kingdom is full", "audience": "New players", "situation": "An important role is empty when someone joins a growing world.", "mechanism": "A visibly wicker stand-in carries the responsibility until a player takes it over.", "impact": "The newcomer can enter a functioning world and see the role they can make their own." }, { "title": "Leave something in the world’s history", "audience": "Players and world creators", "situation": "A successful reign should remain meaningful after its rulers change.", "mechanism": "The Chronicle connects actual contributions and decisions to persistent world records.", "impact": "Participation can become a recognizable part of the kingdom’s continuing story." } ], "technical_features": [ { "name": "Two crowns and distributed roles", "description": "Royal choices cascade into structured responsibilities throughout a shared kingdom." }, { "name": "Validated generated campaigns", "description": "Campaign packs are checked for structure, permitted effects and economy limits before entering the world." }, { "name": "Server-owned identity", "description": "The server supplies identifiers rather than granting identity authority to generated text." }, { "name": "Durable settlement", "description": "Crown, treasury and world state retain consistency across interrupted closes, retries and server processes." }, { "name": "A Chronicle from events", "description": "Recorded decisions ground the day’s story, with player identity handled separately from generated prose." }, { "name": "Visible world continuity", "description": "Wicker NPCs keep unclaimed roles present. Build lineage supports examination of generated world parts." } ], "operating_context": null, "mechanism": [ { "name": "Inhabit", "description": "Enter a shared geography with persistent player records and roles in the kingdom." }, { "name": "Reign", "description": "Turn a royal choice into role-specific tasks. Each contribution progresses independently, and settlement commits its completed consequence." }, { "name": "Remember", "description": "Record settled consequences in the Chronicle and pass the world’s history into the next reign. Keep unfinished contributions distinct from completed changes." } ], "related": [ "https://naevumlabs.com/products/vault" ], "editorial": { "lead": "Two crowns. Everyone changes the kingdom.", "audience_job": "Everbloom is a shared fairytale game for players who like to build together, so that one reign’s choices change the kingdom the next players inherit.", "chapters": [ { "id": "one-day", "kind": "narrative", "title": "A King and Queen choose. A kingdom answers.", "intro": "A royal decision reaches far beyond the throne. Players take roles, carry out the work and leave a visible consequence in a shared fairytale world.", "body": [ "Everbloom connects a daily reign to a continuing kingdom. Royal choices cascade into structured responsibilities across player roles. Wicker NPCs keep unclaimed roles present, with a visible distinction between people and stand-ins.", "Imagine a decision to prepare a new festival: the ruling choice becomes tasks, those tasks affect the kingdom, and the Chronicle retains what happened. This illustrative journey follows the game’s decision and consequence model." ], "intro_role": "editorial-statement" }, { "id": "world-grammar", "kind": "mechanism", "title": "The story can surprise you. The world still has rules.", "intro": "Generated campaigns enter the world through a validated campaign format.", "items": [ { "title": "Bound the event", "text": "Structural checks, permitted effects and economy limits constrain a campaign before it can change the world." }, { "title": "Keep identity with the server", "text": "Server stamps replace generated identifiers. A model’s text cannot assign itself authority over world identity." }, { "title": "Share one reign", "text": "Durable crown and settlement logic keep a kingdom coherent across servers, interrupted closes and repeated delivery." } ] }, { "id": "chronicle", "kind": "process", "title": "The kingdom remembers what changed it.", "intro": "A completed action belongs in the shared state before it becomes part of the day’s story.", "items": [ { "title": "Commit the consequence", "text": "Treasury and world updates distinguish pending work from durable settlement." }, { "title": "Write from events", "text": "The Chronicle derives its account from recorded decisions. Player identities are handled separately from generated prose." }, { "title": "Begin the next day", "text": "The next reign inherits the kingdom’s state and history. The world’s continuing condition gives later choices their context." } ] } ], "questions": [ { "question": "Is this a single-player story?", "answer": "The design centers on a shared kingdom, a King and Queen, distributed player roles and persistent consequences across a reign." }, { "question": "How can creators and partners get involved?", "answer": "Discuss living-world design, creative collaboration or access to the current experience. We can connect the conversation to the relevant stage of Everbloom’s development." } ], "relationships": [ { "type": "research", "id": "vault", "label": "Reusable world mechanisms", "text": "State, composition and evidence contribute to broader engineering research.", "url": "https://naevumlabs.com/products/vault" }, { "type": "related", "id": "metahive-builder", "label": "Another kind of world", "text": "A physical simulation explores a different set of rules and consequences.", "url": "https://naevumlabs.com/products/metahive/builder" } ], "action": { "label": "Discuss Everbloom", "href": "https://naevumlabs.com/contact?context=everbloom" }, "example": { "title": "The Lantern Feast becomes part of the kingdom.", "intro": "A fictional reign, with fictional players, shows a shared campaign becoming something people do—and something the next reign inherits.", "states": [ { "id": "choice", "label": "The choice", "title": "The King and Queen choose a festival.", "body": "A campaign gives the decision a structured form. Its events and effects must fit the kingdom’s rules before they enter play.", "records": [], "artifact": { "kind": "kingdom", "campaign": "The Lantern Feast", "decree": "Open the river garden for the autumn feast. Repair the footbridge first, then invite the kingdom to the market square.", "day": "Day 18", "contributions": [ { "person": "Mira", "role": "Gardener", "task": "Prepare the river garden for the feast.", "state": "proposed" }, { "person": "Tovan", "role": "Smith", "task": "Repair the footbridge before visitors arrive.", "state": "proposed" }, { "person": "Elin", "role": "Baker", "task": "Stock the market stalls for the celebration.", "state": "proposed" } ], "places": [ { "name": "River garden", "before": "Closed for planting", "after": "Open for the feast", "changed": false }, { "name": "Footbridge", "before": "Passage restricted", "after": "Crossing restored", "changed": false }, { "name": "Market square", "before": "An ordinary market day", "after": "Lantern stalls and a shared feast", "changed": false } ], "chronicle": { "title": "A feast is proposed", "text": "The court has chosen a lantern feast. The garden, crossing and market are still unchanged; the kingdom is waiting for its people to do the work.", "events": [ "Court decision recorded", "Three contributions proposed" ] } } }, { "id": "roles", "label": "The roles", "title": "The choice becomes work across the kingdom.", "body": "Responsibilities cascade to player roles. Unclaimed roles remain present through wicker NPCs, so the social structure stays intelligible.", "records": [], "artifact": { "kind": "kingdom", "campaign": "The Lantern Feast", "decree": "Open the river garden for the autumn feast. Repair the footbridge first, then invite the kingdom to the market square.", "day": "Day 18", "contributions": [ { "person": "Mira", "role": "Gardener", "task": "Prepare the river garden for the feast.", "state": "in-progress" }, { "person": "Tovan", "role": "Smith", "task": "Repair the footbridge before visitors arrive.", "state": "complete" }, { "person": "Elin", "role": "Baker", "task": "Stock the market stalls for the celebration.", "state": "in-progress" } ], "places": [ { "name": "River garden", "before": "Closed for planting", "after": "Open for the feast", "changed": false }, { "name": "Footbridge", "before": "Passage restricted", "after": "Crossing restored", "changed": true }, { "name": "Market square", "before": "An ordinary market day", "after": "Lantern stalls and a shared feast", "changed": false } ], "chronicle": { "title": "The crossing is restored", "text": "Tovan has repaired the footbridge. Mira is still preparing the river garden, while Elin gathers provisions. The feast is promised, but the whole campaign is not yet complete.", "events": [ "Footbridge repair completed", "Garden preparation in progress", "Market provisions in progress" ] } } }, { "id": "consequence", "label": "The world", "title": "The next reign inherits the result.", "body": "Settled changes become part of the shared kingdom. The Chronicle draws on recorded events, giving tomorrow’s choices a history.", "records": [], "artifact": { "kind": "kingdom", "campaign": "The Lantern Feast", "decree": "Open the river garden for the autumn feast. Repair the footbridge first, then invite the kingdom to the market square.", "day": "Day 18", "contributions": [ { "person": "Mira", "role": "Gardener", "task": "Prepare the river garden for the feast.", "state": "complete" }, { "person": "Tovan", "role": "Smith", "task": "Repair the footbridge before visitors arrive.", "state": "complete" }, { "person": "Elin", "role": "Baker", "task": "Stock the market stalls for the celebration.", "state": "complete" } ], "places": [ { "name": "River garden", "before": "Closed for planting", "after": "Open for the feast", "changed": true }, { "name": "Footbridge", "before": "Passage restricted", "after": "Crossing restored", "changed": true }, { "name": "Market square", "before": "An ordinary market day", "after": "Lantern stalls and a shared feast", "changed": true } ], "chronicle": { "title": "The first lanterns on the river", "text": "On the eighteenth day, the court opened the Lantern Feast. Mira’s garden welcomed the kingdom, Tovan’s repaired bridge carried the visitors, and Elin’s market stalls fed the gathering. The next reign inherits an open garden and a restored crossing, along with the memory of who made them possible.", "events": [ "River garden opened", "Footbridge restored", "Market celebration completed", "Campaign settled into world history" ] } } } ] } }, "anatomy": { "representation": "An illustrative game design: completed contributions change the world and enter its history.", "signature": "Inhabited reign, contributions and Chronicle", "parts": [ { "id": "reign", "name": "Shared reign and world reference", "description": "Retain crowns, current world conditions and the next reign context." }, { "id": "campaign", "name": "Campaign grammar and role distributor", "description": "Constrain a royal choice into structured role-specific tasks." }, { "id": "contributions", "name": "Contribution work areas", "description": "Keep garden, bridge and market tasks independently in progress or complete." }, { "id": "settlement", "name": "Settlement assembly", "description": "Commit actual completed consequences under the authoritative world rules." }, { "id": "chronicle", "name": "Chronicle and next-reign memory", "description": "Write from retained decisions/consequences and pass world history onward." } ] } } ], "technology": [ { "id": "technology-software-development", "name": "Software development", "url": "https://naevumlabs.com/technology/software-development", "description": "Naevum connects Black Box’s delivery architecture, Agency Engine’s protected application core, Conductr’s replaceable harness and the Vault’s component contracts. Each solves a different failure between a promising prototype and a system people can depend on.", "applications": [ { "title": "Turn architectural direction into delivery", "audience": "Organizations commissioning custom AI", "situation": "A difficult brief spans data, software, interfaces and operating responsibilities.", "mechanism": "The Planner defines a mandate; implementation, review and QA preserve its acceptance conditions.", "impact": "The customer can examine how the delivered system answers the original brief." }, { "title": "Continue after a worker disappears", "audience": "Engineering teams", "situation": "A development attempt ends before the work is resolved.", "mechanism": "The source-bound handover retains objective, checks, findings and the next decision.", "impact": "Another attempt can continue from the durable record rather than relying on a departed session." }, { "title": "Launch another product from the same core", "audience": "Product organizations", "situation": "A new application needs its own workflow, language and interface.", "mechanism": "Compose a product-owned shell and selected services above the sealed governance core.", "impact": "The team can change the experience without rebuilding approvals, evidence and custody controls." } ], "technical_features": [ { "name": "Mandate record", "description": "Scope, ceiling, resource budget, expiry and confirmed plan revision bind a factory work package. A proposed plan is separate from effective authority." }, { "name": "Independent acceptance identity", "description": "The acceptance condition, candidate revision, findings and artifact hashes travel together. A source change requires checks against the new candidate." }, { "name": "Five application packages", "description": "Kernel, services, UI hooks, UI kit and screens sit below a product-owned frontend/backend shell. The declared sealed-set bindings are checked at startup and in contract tests; this closed-list guard is not process isolation." }, { "name": "Monotone policy composition", "description": "Engine → product → installation → organization → workspace/seat. Later policy can narrow the permitted set; it cannot widen it." }, { "name": "Four-operator contract algebra", "description": "Pipe, parallel, feedback and wrap operate over typed ports and assume–guarantee contracts. Unresolved assumptions propagate outward." }, { "name": "Request/offer compatibility", "description": "Port checks include cadence, ordering and absence tolerance. A slower producer fails a faster consumer’s declared requirement; uncomparable rates stay unknown." }, { "name": "Harness experiment manifest", "description": "Baseline and treatment pin model, context strategy, loop, task battery, repetitions and seed basis. Statistical gates are declared separately from tracked diagnostics." }, { "name": "Versioned distribution", "description": "Semantic package versions, dependency boundaries and software bills of materials keep component changes attributable to an application release." } ], "editorial": { "lead": "Naevum connects Black Box’s delivery architecture, Agency Engine’s protected application core, Conductr’s replaceable harness and the Vault’s component contracts. Each solves a different failure between a promising prototype and a system people can depend on.", "audience_job": "Naevum's software development method is a protected core, a replaceable harness and contract-carrying components for VPs of engineering, so that a prototype becomes a system another team can operate.", "chapters": [ { "id": "accepted-change", "kind": "process", "title": "A prototype is only one part of the system.", "intro": "A model can answer correctly while its application loses approvals, repeats a tool action or fails when another team takes over. Our development architecture carries the same objective through planning, implementation, independent challenge and the exact delivered artifact.", "body": [ "Black Box defines the delivery responsibilities. Agency Engine supplies application controls and product seams. Conductr separates the coding harness into replaceable components. The Vault checks the assumptions carried by reusable engineering. Together they give a custom system an architecture for how it is built as well as how it runs." ], "items": [ { "title": "Mandate and acceptance", "text": "The factory design begins with an owner-confirmed scope, permitted effects, budget and expiry. An independently authored acceptance condition precedes implementation.", "input": "A defined user journey and operating context.", "output": "A bounded work package with its acceptance basis.", "boundary": "The planner proposes work; confirmation belongs to the responsible owner." }, { "title": "Separate implementation and challenge", "text": "Coder, reviewer and QA responsibilities are separated. Findings refer to the source they examined, so a correction changes what must be checked.", "input": "One work package and exact source revision.", "output": "A candidate, findings and source-bound checks.", "boundary": "A successful attempt does not authorize its own release." }, { "title": "Durable continuation", "text": "A handover packet retains decisions, findings and the next unresolved obligation when an attempt ends.", "input": "The attempt’s artifacts and observed results.", "output": "A record from which a successor can continue.", "boundary": "Uncertain effects and failed acceptance remain visible." } ] }, { "id": "product-seams", "kind": "mechanism", "title": "A new product keeps its own experience and inherits a protected core.", "intro": "Agency Engine separates the kernel, service kit, UI hooks, component kit and screens from the product-owned shell. Imports point towards the foundation; product-specific composition sits above it.", "items": [ { "title": "Sealed governance services", "text": "Gate and decision logic, the ledger, write boundary, capabilities, evidence, approvals, model admission and connector/store custody form the sealed set. The architecture declares these services non-replaceable. A closed-list startup guard and contract tests check the declared bindings; this is not process isolation or protection against arbitrary new bindings.", "input": "Product composition and requested capabilities.", "output": "An application whose permitted extensions have explicit boundaries." }, { "title": "Five configuration layers", "text": "Engine defaults, product configuration, installation, organization and workspace or seat merge in that order. A lower layer may tighten inherited policy.", "input": "Versioned defaults plus local context.", "output": "The effective policy for that installation.", "boundary": "A local preference cannot grant a capability denied by an earlier layer." }, { "title": "Service and release contracts", "text": "Services declare whether they belong in the client or a separately operated cell. Versioned packages and per-layer software bills of materials define a release independently from a product’s appearance.", "input": "Selected service contracts and package versions.", "output": "A product dependency graph and upgrade boundary.", "boundary": "The client/cell tag describes the boundary; actual placement belongs to the deployment profile." }, { "title": "Durable semantic work", "text": "Agency’s semantic orchestration design separates task/output contracts from individual attempts. A typed dependency graph drives readiness; journal replay reconstructs work state without another reasoning call. An uncertain effect blocks automatic reissue.", "input": "A versioned intent graph, prior events and declared output contract.", "output": "Reconstructed task state, eligible work and retained unresolved effects.", "boundary": "Exactly-once external consequences require matching support at the effect sink." } ] }, { "id": "contract-composition", "kind": "comparison", "title": "Compatibility includes the assumptions between components.", "intro": "The Vault’s composition checker carries assumptions as data. Matching a function signature does not establish that two components agree about timing, order or who owns a resource.", "items": [ { "title": "Pipe and parallel", "text": "Typed ports connect sequential or concurrent components. Request/offer checks compare cadence, ordering and whether a value may be absent.", "input": "Producer guarantees and consumer requirements.", "output": "Connections plus compatibility findings." }, { "title": "Feedback and wrap", "text": "Feedback closes a declared loop; a wrapper adapts a component’s environment. Assumptions that neither a neighbor nor the environment satisfies remain obligations of the composition.", "input": "Component contracts and the proposed arrangement.", "output": "A composed contract with unresolved obligations retained.", "boundary": "An uncomparable event rate remains unknown; it does not become a compatibility pass." } ] }, { "id": "harness-components", "kind": "mechanism", "title": "The harness has parts that can be tested separately.", "intro": "Conductr defines a context engine, planner, orchestrator, executor, sandbox, router, verifier, policy gate and event store as interface-bounded slots. The orchestrator coordinates their order; a component can be replaced without pretending its responsibilities disappeared.", "items": [ { "title": "Externally defined acceptance", "text": "A work node names files, the intended change, forbidden changes and an acceptance test. The executor works against that external oracle.", "input": "A manifest node and seeded repository.", "output": "An edit and its acceptance result." }, { "title": "Assurance outside the live loop", "text": "Verifier findings and trust events are observed separately from the agent’s ongoing context. A policy gate may act at the commit boundary.", "input": "The candidate diff and declared promises.", "output": "Findings, policy decisions and event records.", "boundary": "Conductr’s degrading integration seams are distinct from Synapse’s disclosure-admission controls." }, { "title": "Controlled harness experiments", "text": "The proving-ground manifest fixes task battery, model, context strategy, repetitions and seed basis for baseline and treatment. Gated metrics and tracked metrics have separate roles.", "input": "Two harness configurations on the same task basis.", "output": "A comparison tied to its tasks, variation and acceptance rule.", "boundary": "A measured harness improvement is specific to that experiment." } ] }, { "id": "delivery-example", "kind": "narrative", "title": "One approval workflow, from request to recovery.", "intro": "In an illustrative composed workflow, an institution commissions document review with a restricted appendix and two approval roles.", "body": [ "The application composition defines the two roles and protects the model door. Synapse supplies the context-admission boundary. A work package names the permitted document view, expected review artifact and interrupted-request recovery. Independent acceptance follows the candidate’s exact source, while the handover retains unresolved findings and the operating procedure.", "This is how the technology enters a custom engagement: a concrete workflow assembled from explicit controls and individually understood dependencies. The factory triad is the selected delivery design; its automated execution is qualified separately for each project." ] } ], "questions": [ { "question": "What does Naevum contribute beyond an agent writing code?", "answer": "The contribution is the surrounding architecture: protected application controls, contract-aware reuse, independent acceptance, replaceable execution components and durable delivery records. These mechanisms connect a business workflow to an inspectable system." }, { "question": "Which parts are shared and which are custom?", "answer": "The application foundation supplies shared governance and selected services. Domain logic, interfaces, integrations, local policy and acceptance conditions belong to the engagement. Factory automation and deployment boundaries are qualified for the named project." } ], "relationships": [ { "type": "delivery", "id": "black-box", "label": "Delivery system", "text": "Architecture, implementation, independent challenge and handover.", "url": "https://naevumlabs.com/products/black-box" }, { "type": "foundation", "id": "agency-engine", "label": "Product architecture", "text": "Compose distinct applications above common controls.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "related", "id": "conductr", "label": "Development harness", "text": "Context, attempts and verification.", "url": "https://naevumlabs.com/products/conductr" }, { "type": "uses", "id": "vault", "label": "Reusable engineering", "text": "Contracts and evidence for composition.", "url": "https://naevumlabs.com/products/vault" } ], "action": { "label": "Discuss a software platform", "href": "https://naevumlabs.com/contact?context=ai-products" } }, "anatomy": { "representation": "An illustrative engineering method for composing interfaces, producing candidates and examining exact-source evidence.", "signature": "Versioned artifact path and independent examination", "parts": [ { "id": "definition", "name": "Work-package definition", "description": "Make the objective and acceptance question concrete before implementation." }, { "id": "contracts", "name": "Reusable-contract bench", "description": "Compare candidate mechanisms and carry unresolved environmental assumptions." }, { "id": "application", "name": "Application composition", "description": "Seat domain services above protected contracts and narrowing configuration." }, { "id": "implementation", "name": "Bounded implementation harness", "description": "Transform the work package into a candidate through declared work slots." }, { "id": "examination", "name": "Independent examination", "description": "Examine the exact candidate through separate structural and behavioral questions." }, { "id": "handover", "name": "Release and handover register", "description": "Retain the artifact, evidence, recovery method and remaining decisions." } ] } }, { "id": "technology-adaptive-control", "name": "Controllers & adaptive control", "url": "https://naevumlabs.com/technology/adaptive-control", "description": "Naevum’s control systems connect measured load, uncertainty and outstanding obligations to bounded changes in attention, pace and resource allocation. Manna supplies organism regulation; Vagus supplies temporal attention and modulation.", "applications": [ { "title": "Follow condition on an operating vessel", "audience": "Marine engineering", "situation": "Load, condition and sensor histories change while equipment remains in service.", "mechanism": "Exposure-aware diagnostics and time-to-event estimates connect the condition record to risk.", "impact": "The engineering team can investigate how operating context changes the evidence behind an assessment." }, { "title": "A process with a safe stopping point", "audience": "Industrial edition", "situation": "An operational process must react to observations while respecting equipment and responsibility boundaries.", "mechanism": "A pipeline separates perception, proposal and gated action; the journal records the resulting state.", "impact": "Operators can inspect why a step was admitted, stop the process and recover from its record." }, { "title": "Attention across a technical estate", "audience": "Engineering organizations", "situation": "Signals arrive from many projects, workers and resource windows.", "mechanism": "Temporal attention integrates repeated observations and uses failure and outcome memory.", "impact": "The operator sees why a matter deserves attention without treating every new observation as an emergency." } ], "technical_features": [ { "name": "Interoceptive record", "description": "Load, available memory and thermal observations remain alongside pressure, its component terms and explicit unknown status. Pressure is bounded to [0, 1]." }, { "name": "Reference pacing law", "description": "The inspected design begins stretching at pressure 0.6 and bounds the multiplier to [1, 4]. Deployment tuning is part of the operating profile." }, { "name": "Obligation state", "description": "Each obligation carries identity, organ, remaining attention need, deadline, weight, protected status and served count. Lost-organ detection retains its open obligations." }, { "name": "Attention allocation", "description": "Protected due work is allocated first; ordinary work follows urgency and slack, capped salience, dwell and a rolling per-organ share. Failed outcomes damp ordinary claims." }, { "name": "Antithetic-inspired PI state", "description": "Two integral accumulators update from observation and reference; their paired cancellation feeds proportional/integral terms and a bounded output. The recommendation does not set a verdict." }, { "name": "Temporal attention states", "description": "New, standing and cleared conditions are distinct. Decay and refractory timing retain history between observations; stale evidence remains distinguishable from resolution." }, { "name": "Bounded modulation", "description": "Pressure, threat, opportunity, patience and plasticity affect exploration, attention threshold and maintenance rates through clamped gains. Permissions are outside that rate API." }, { "name": "Deployment contract", "description": "The inspected Manna interoception and attention modules are opt-in reference implementations, not active by default. Control modules and operating integrations have separate activation profiles. Sensor basis, parameter ranges, intervention conditions and recovery are fixed for the commissioned system." } ], "editorial": { "lead": "Naevum’s control systems connect measured load, uncertainty and outstanding obligations to bounded changes in attention, pace and resource allocation. Manna supplies organism regulation; Vagus supplies temporal attention and modulation.", "audience_job": "Adaptive control is how Naevum's software paces itself under load, for operations managers, so that due work and approvals hold while the system slows the rest.", "chapters": [ { "id": "bounded-feedback", "kind": "process", "title": "A busy system must preserve the work that matters.", "intro": "New requests arrive while a worker is late, memory is constrained and a verification step is due. A fixed-rate loop keeps spending; a simple priority queue can neglect recovery. Our control architecture keeps observations, commitments and permitted responses separate.", "items": [ { "title": "Measure the operating condition", "text": "Manna’s interoception reads load, available memory and thermal limits. Unknown observations remain explicitly marked; a missing sensor is not a healthy reading.", "input": "Timestamped resource observations.", "output": "Bounded pressure plus the terms and unknowns behind it." }, { "title": "Allocate attention to obligations", "text": "The attention allocator retains each obligation’s identity, owner, remaining need, deadline and protected status. Due protected work receives service before ordinary claims.", "input": "Open obligations and fresh organ heartbeats.", "output": "An allocation within a finite per-tick and per-organ budget." }, { "title": "Adjust within an envelope", "text": "Rate changes, scheduling and resource allocation remain within declared bounds. Pressure can tighten behavior while the action gate retains authority.", "input": "Condition estimates and the current mandate.", "output": "A bounded change followed by another observation.", "boundary": "A controller cannot lower an approval requirement or expand a capability." } ] }, { "id": "stable-response", "kind": "mechanism", "title": "Several control laws solve different parts of the problem.", "intro": "Naevum uses bounded feedback and stateful selection rather than one opaque “health score”. The result retains enough internal state to explain why behavior changed.", "items": [ { "title": "Physiological pacing", "text": "The reference interoception design takes the maximum of known load, memory and thermal pressure terms. Above its selected pressure threshold, it stretches the tick interval by a bounded multiplier.", "input": "Normalized known pressure terms.", "output": "A pace multiplier and an evidence record.", "boundary": "The reference configuration stretches by at most 4×; this is a design parameter, not a throughput promise." }, { "title": "Dwell and fairness", "text": "The allocator gives the current focus a dwell bonus and applies a rolling per-organ cap to ordinary work. A noisy specialist cannot purchase the entire shared attention budget.", "input": "Urgency, slack, capped salience and recent allocations.", "output": "Service assignments with protected-work exceptions." }, { "title": "Setpoint regulation", "text": "Vagus’s vital-sign regulator keeps two integral states, updates them from measured condition and reference, and applies a paired cancellation step before producing a clamped PI output.", "input": "A measured condition, reference and previous regulator state.", "output": "A bounded velocity recommendation.", "boundary": "An unmeasured interval does not advance the regulator; this discrete software law carries no imported biological stability guarantee." } ] }, { "id": "temporal-control", "kind": "mechanism", "title": "A persistent condition and a new event need different responses.", "intro": "Vagus maintains a potential for each subject. New evidence adds a full weighted charge; repeated unchanged evidence contributes a smaller standing term. Decay, firing and recovery are explicit state transitions.", "items": [ { "title": "Accumulation and trust", "text": "Charge is weighted by condition and evidence trust. Measured signals contribute fully, derived signals contribute at reduced weight and untrusted signals do not charge the attention potential." }, { "title": "Threshold and refractory state", "text": "Crossing the attention threshold can produce an item. A refractory period prevents repeated immediate firing while the ongoing condition remains visible as standing." }, { "title": "Blindness and resolution", "text": "A stale sensor cannot silently clear the condition it used to observe. The system retains the standing issue with stale evidence until its own cause is measured resolved." } ] }, { "id": "control-example", "kind": "comparison", "title": "The same architecture preserves different responsibilities.", "intro": "The control laws serve a declared operation; they do not decide what the organization is allowed to do.", "items": [ { "title": "A research workload", "text": "A resource-constrained personal agent slows ordinary work while retaining its objective and pending approvals. A recovery obligation remains identifiable.", "input": "Resource observations and due work.", "output": "A bounded schedule without discarding commitments." }, { "title": "An institutional estate", "text": "Vagus combines repeated engineering signals, changing attention and historical failure evidence. An operator sees why an issue remains active or has become unobservable.", "input": "Freshness-aware project and infrastructure observations.", "output": "A prioritized attention record with its basis." }, { "title": "An industrial process", "text": "The architecture separates equipment observations, a proposed response and the permitted actuation path. Physical interfaces and safety limits are specific to the commissioned installation.", "input": "Process condition and the operator’s mandate.", "output": "A recorded proposal and admitted effect." } ] } ], "questions": [ { "question": "Does adaptation mean rewriting the system’s permissions?", "answer": "No. A controller changes behavior within its operating envelope. A change to authority needs its own authorized process." } ], "relationships": [ { "type": "uses", "id": "manna", "label": "Regulated execution", "text": "Resource condition can shape continuing work.", "url": "https://naevumlabs.com/products/manna" }, { "type": "uses", "id": "vagus", "label": "Attention dynamics", "text": "Signals, decay and modulation allocate attention.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "related", "id": "extrasensory", "label": "Intervention reasoning", "text": "Forecasts inform bounded operational decisions.", "url": "https://naevumlabs.com/products/extrasensory" } ], "action": { "label": "Design a continuing operation", "href": "https://naevumlabs.com/contact?context=persistent-agents" } }, "anatomy": { "representation": "An illustrative feedback method that relates valid observations to finite work and retained obligations.", "signature": "Bounded regulation and protected service", "parts": [ { "id": "measurement", "name": "Measurement and normalization", "description": "Align independent operating terms without replacing unknown readings with zero." }, { "id": "obligations", "name": "Obligation store and allocator", "description": "Distribute finite service while retaining protected due work." }, { "id": "regulators", "name": "Regulator-state bank", "description": "Keep pacing, integral/reference regulation and temporal attention distinct." }, { "id": "distribution", "name": "Bounded rate distributor", "description": "Apply finite rates and declared dwell/hysteresis to eligible work." }, { "id": "response", "name": "Response and evidence register", "description": "Retain actual service and independently observed condition after the change." } ] } }, { "id": "technology-bio-inspired-intelligence", "name": "Bio-inspired technologies", "url": "https://naevumlabs.com/technology/bio-inspired-intelligence", "description": "Naevum gives memory, attention, prediction, action selection, regulation and protected model interaction different computational jobs. The connections between those jobs form a continuing intelligence environment.", "applications": [ { "title": "Use outside intelligence on a bounded task", "audience": "Institutions and R&D teams", "situation": "A useful model needs context, while the organization holds a much larger sensitive source.", "mechanism": "Select a source-bound projection and account for its disclosure to the actual observer before dispatch.", "impact": "The organization can inspect the approved context, purpose and destination of the model work." }, { "title": "Specialists working toward one purpose", "audience": "Engineering teams", "situation": "Several specialized agents need a shared direction and finite resources.", "mechanism": "Constellations combine bounded mandates with attention allocation and fault isolation.", "impact": "A failure or urgent signal can be addressed locally while the broader objective remains intact." }, { "title": "Learn which work succeeds", "audience": "Operational teams", "situation": "Different attempts consume different resources and produce different outcomes.", "mechanism": "Outcome predictions and delayed reward joins retain the task and source behind each observation.", "impact": "Selection can improve from recorded experience while the operator keeps control of authority." } ], "technical_features": [ { "name": "Six differentiated functions", "description": "Temporal memory, signal integration, forward prediction, eligible-action selection, bounded modulation and constitution-governed execution have separate state and contracts." }, { "name": "Leaky attention reference", "description": "The inspected Vagus configuration uses a six-hour half-life, a 24-hour refractory period and a capped firing potential of 4. These are selected reference-implementation parameters, not neuronal measurements." }, { "name": "Trust-weighted signals", "description": "Measured evidence charges attention fully, derived evidence at half weight and untrusted evidence at zero. Freshness failures feed an explicit unknown driver." }, { "name": "Standing versus surprise", "description": "Fingerprint changes add the full active-driver charge. Unchanged observations add a smaller elapsed-time term, preserving chronic conditions without repeated immediate interruption." }, { "name": "Hierarchical outcome model", "description": "Beta–Binomial posteriors estimate a declared outcome target. Prior strength is fitted and bounded; sibling/ancestor evidence avoids counting a branch’s observations twice." }, { "name": "Resource forecast model", "description": "Token use and wall time use log-normal models with medians and 80% intervals in the inspected forward-model implementation. Their coverage needs its own evaluation." }, { "name": "Posterior-guided selection", "description": "Thompson sampling compares eligible worker posteriors and retains an exploration route. Observational task histories do not establish a causal superiority claim." }, { "name": "Constrained organ lifecycle", "description": "Template digest, purpose, parent, resource grant, lifetime and outstanding obligations remain attached to an organ. Retirement preserves unresolved effects and protected functions." }, { "name": "Engineering scope", "description": "These are computational mechanisms and selected architecture, with module-specific activation and qualification. They do not assert consciousness, biological equivalence or neuromorphic hardware." } ], "editorial": { "lead": "Naevum gives memory, attention, prediction, action selection, regulation and protected model interaction different computational jobs. The connections between those jobs form a continuing intelligence environment.", "audience_job": "Bio-inspired design is how Naevum divides memory, attention, prediction and action among separate organs, for research leads, so that the system prepares for demand before it arrives.", "chapters": [ { "id": "mechanisms-over-metaphor", "kind": "mechanism", "title": "Intelligence needs an internal organization.", "intro": "An institution can have abundant documents and capable models yet still lose the reason for a decision, interrupt the wrong person or repeat a failed action. Our architecture combines dedicated organs whose state and exchanges can be inspected.", "items": [ { "title": "Hippocampal memory", "text": "Vagus indexes claims, episodes, lessons and failures while canonical records retain source, time and authority. Recall can reconstruct an earlier evidence state.", "input": "A scoped question and time basis.", "output": "A bounded set of evidence-bearing memories or a reason for abstention." }, { "title": "Thalamic attention", "text": "A subject-specific leaky potential accumulates trust-weighted evidence. Threshold crossing, refractory state and standing-condition logic decide what deserves examination.", "input": "Fresh signals, their provenance and the previous potential.", "output": "A new, standing or cleared attention item." }, { "title": "Cerebellar forward models", "text": "Hierarchical outcome and resource models predict before work runs. The later result is compared with the stored forecast. The proxy/0.2 target means an episode completed without an observed interruption or red-test/failure ending. Independent outcome_ok labels form a separate target; neither is an automatic measure of task correctness.", "input": "Task class, worker and operating context known before execution.", "output": "An outcome estimate and resource intervals tied to their evidence." }, { "title": "Basal-ganglia-inspired selection", "text": "Eligible workers are compared through posterior sampling with bounded exploration. An estimate can recommend a worker; it does not grant permission to run it.", "input": "Eligible alternatives and their outcome posteriors.", "output": "A selection recommendation with an explicit evidence basis." } ] }, { "id": "organism-design", "kind": "constraints", "title": "Purpose, resources and boundaries keep the organs coherent.", "intro": "Manna supplies the continuing organism structure: a sealed constitution, specialized cells, supervised pipelines, resource envelopes and gated effects. Synapse regulates the interaction with outside model computation.", "items": [ { "title": "A constitution outside the model", "text": "Genome laws constrain the action gate. A model’s proposed step cannot rewrite the laws that decide whether it may act." }, { "title": "Internal condition becomes control input", "text": "Interoception and attention retain resource pressure, active obligations, organ freshness and uncertain effects. Their role is coordination under finite capacity." }, { "title": "A selective intelligence boundary", "text": "Synapse records admitted source ancestry and cumulative observer exposure before transport. The membrane analogy refers to these software controls." }, { "title": "Controlled development", "text": "The organ-lifecycle design allocates only reviewed templates with a finite resource grant and lifetime. Birth, retirement and interrupted allocation have separate records.", "input": "A reviewed template and available grant.", "output": "An identified organ or a refused allocation.", "boundary": "Manna’s attention, lifecycle and continuity studies are opt-in reference implementations, not active by default. Their integration is qualified for the deployment profile." } ] }, { "id": "attention-mechanism", "kind": "mechanism", "title": "Attention has dynamics that a priority label cannot express.", "intro": "The Vagus implementation combines leaky integration with evidence freshness and surprise. An unchanged alarm and a newly deteriorating condition do not contribute in the same way.", "items": [ { "title": "Decay and charge", "text": "Potential decays with elapsed time. Active drivers add weighted level × trust; a changed fingerprint adds a new charge, while an unchanged condition adds a smaller standing contribution." }, { "title": "Firing and recovery", "text": "The potential must cross its threshold outside the refractory interval. The active cause must still support the item, so leftover potential alone cannot create a fresh issue." }, { "title": "Ranking and state remain separate", "text": "An uncapped potential informs salience ranking; a capped potential controls firing persistence. A chronic heavy condition therefore remains distinguishable from a lighter one." }, { "title": "Missing evidence is a condition", "text": "The freshness model marks stale, missing and errored inputs. The observer losing sight of a problem is different from the problem disappearing." } ] }, { "id": "prediction-selection", "kind": "comparison", "title": "Prediction, selection and modulation carry different information.", "intro": "The forward model estimates consequences; selection compares eligible alternatives; modulation changes how quickly the system explores or attends.", "items": [ { "title": "Hierarchical empirical Bayes", "text": "Outcome estimates use Beta–Binomial updates across portfolio, holder, repository, task and worker contexts. Sparse contexts borrow a bounded prior from related observations. A level’s posterior is Beta(s + k·m, f + k·(1 − m)), where m is the parent mean and k is a fitted, bounded prior strength." }, { "title": "Prequential evaluation", "text": "A prediction is scored against an outcome that arrives later. Unknown endings are excluded from the labeled outcome population. Resource medians and intervals remain separate from the outcome probability." }, { "title": "Five modulation signals", "text": "Pressure, threat, opportunity, patience and plasticity are bounded state variables. Their gains alter rates and thresholds; the API does not return an actuation verdict." } ] }, { "id": "organ-coordination-example", "kind": "narrative", "title": "A changed policy travels through the whole environment.", "intro": "In an illustrative composed episode, an institutional procedure changes while a long-running review is interrupted.", "body": [ "Temporal memory retains both the earlier instruction and the new rule. Attention notices a genuinely changed source instead of treating every repeated observation as new. The forward model estimates the next episode’s defined outcome, total token use and wall time. A continuing agent retains the objective and unresolved approval. Synapse checks the next external model task against its source and observer history.", "This is the ecosystem’s technical argument: specialized state, explicit connections and authority that remains traceable as work continues. Biological inspiration names the design lineage; the implemented algorithm and its deployment profile define the behavior." ] } ], "questions": [ { "question": "Why use biological ideas?", "answer": "Living systems offer useful design questions about specialized functions, regulation and continuity. The engineering value depends on the explicit algorithm and its measured effect in software." } ], "relationships": [ { "type": "uses", "id": "manna", "label": "Continuing organisms", "text": "Purpose, regulation and supervised subsystems.", "url": "https://naevumlabs.com/products/manna" }, { "type": "uses", "id": "vagus", "label": "Memory and attention", "text": "Temporal knowledge and attention dynamics.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "uses", "id": "synapse", "label": "Selective interaction", "text": "Context and consequence boundaries.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "research", "id": "anemone", "label": "Consequential application", "text": "The lab’s own continuing autonomous setting.", "url": "https://naevumlabs.com/lab/anemone" } ], "action": { "label": "Discuss an adaptive architecture", "href": "https://naevumlabs.com/contact?context=adaptive-intelligence" } }, "anatomy": { "representation": "An illustrative organization of cooperating functions; their roles remain distinct from biological claims.", "signature": "Distinct memory, attention, prediction and selection", "parts": [ { "id": "memory", "name": "Temporal-memory organ", "description": "Recover source-bearing context appropriate to the question." }, { "id": "attention", "name": "Attention integrator", "description": "Accumulate eligible changes and retain standing causes." }, { "id": "estimates", "name": "Forward-estimate bank", "description": "Store a defined episode target and resource estimate before execution." }, { "id": "choice", "name": "Eligible-choice selector", "description": "Compare eligible alternatives under finite resources." }, { "id": "executive", "name": "Executive and supervised work", "description": "Retain purpose and obligations while performing useful task transformations." }, { "id": "interaction", "name": "Selective interaction and return", "description": "Carry admitted external work and independent results under retained identity." } ] } }, { "id": "technology-autonomy", "name": "Autonomy levels 4 & 5", "url": "https://naevumlabs.com/technology/autonomy", "description": "Manna’s autonomy architecture retains a mandate, supervised work, causal history and unresolved effects. It separates the intelligence proposing an action from the runtime that admits it.", "applications": [ { "title": "A project that continues tomorrow", "audience": "Personal edition", "situation": "A person has a continuing objective spread across sessions, files and decisions.", "mechanism": "A genome retains the objective and envelope; heartbeats and memory resume work with its earlier context.", "impact": "The next session can begin with the objective, prior decisions and outstanding approvals already connected." }, { "title": "Operate an automated multi-market fund", "audience": "Naevum’s own fund", "situation": "Market conditions keep changing after any one decision is made.", "mechanism": "A continuing mandate connects observation, constrained action and subsequent evidence.", "impact": "The operating system can be examined across a sequence of decisions rather than a single recommendation." }, { "title": "Continue after a worker disappears", "audience": "Engineering teams", "situation": "A development attempt ends before the work is resolved.", "mechanism": "The source-bound handover retains objective, checks, findings and the next decision.", "impact": "Another attempt can continue from the durable record rather than relying on a departed session." } ], "technical_features": [ { "name": "Pipeline contract", "description": "Intent + envelope + flow + heartbeat define the supervised work unit. Cells retain specialized state; genome laws remain sealed outside model-generated proposals." }, { "name": "Admission sequence", "description": "Liveness, failed-value checks, genome laws, deny-only policy, provenance, approval tier and capability envelope precede actuation." }, { "name": "Source provenance", "description": "External values retain their untrusted origin. Transformation does not make them eligible to rewrite the intent, topology, envelope or gate conditions." }, { "name": "Effect identity and journal", "description": "An admitted effect and its observed result retain causal identity. Uncertainty remains explicit during interruption and recovery." }, { "name": "Four assurance dimensions", "description": "Structure, policy, artifact and outcome each carry PASS, FAIL or UNKNOWN with a reason and source. An unevaluated outcome does not inherit a structural pass." }, { "name": "Typed lookahead contract", "description": "The world-model research boundary separates SensorReading, BeliefState, Forecast, Grant, Proposal and OutcomeComparison. A forecast carries target, units, horizon, input cutoff and expiry." }, { "name": "Continuity criteria", "description": "Causal ancestry and critical carryover act as vetoes. Fixed behavioral probes and self-model fidelity test the successor separately; the scoring conventions require calibration." }, { "name": "Organ resource conservation", "description": "Reservation → persisted identity → activation. Retirement checks quiescence, protected templates, obligation transfer and unresolved effects." }, { "name": "Recovery certificate", "description": "A target state digest is separate from the action that restores it. Restoration is checked after execution; a ledger entry alone does not establish rollback." } ], "editorial": { "lead": "Manna’s autonomy architecture retains a mandate, supervised work, causal history and unresolved effects. It separates the intelligence proposing an action from the runtime that admits it.", "audience_job": "Level 4 and 5 autonomy is continuing work inside a written mandate, for boards and operators who delegate, so that a separate runtime checks every action before it happens.", "chapters": [ { "id": "interrupted-task", "kind": "process", "title": "A continuing operation cannot depend on a single conversation.", "intro": "A worker stops after a request may already have reached another system. Its successor must retain what was authorized, what happened and what is still uncertain. Naevum gives continuing work an explicit structure around those facts.", "items": [ { "title": "Declare a continuing purpose", "text": "Manna represents an organism through a genome and specialized cells. Each pipeline has an intent, resource/capability envelope, flow and heartbeat.", "input": "An objective and the owner’s permitted domain.", "output": "Supervised units of work with named contracts." }, { "title": "Admit the effect at runtime", "text": "Pure transformations and actuation are different operations. An action crosses a gate that checks liveness, laws, policy, provenance, approval tier and capability bounds.", "input": "A proposed effect and its current operating state.", "output": "An admitted action or recorded refusal.", "boundary": "Model output supplies no independent actuation authority." }, { "title": "Continue from the causal record", "text": "Journals preserve state transitions, effects and unresolved obligations. Recovery examines the existing identity rather than treating lost acknowledgment as permission to repeat.", "input": "The verified record and current observations.", "output": "A continuation or an explicit recovery decision." } ] }, { "id": "levels", "kind": "comparison", "title": "Level 4 defines a domain. Level 5 defines a wider research horizon.", "intro": "The levels borrow the structure of SAE J3016, the automated-driving scale: Level 4 operates inside a declared domain, Level 5 beyond it. They describe the scope of responsibility the architecture is built for, not a certification.", "items": [ { "title": "Level 4: a declared operating domain", "text": "The mandate defines tools, conditions, budgets, permitted effects and intervention points. Continuing operation is qualified within that domain.", "input": "A known environment and bounded responsibility.", "output": "Persistent work with a specified exception and recovery path." }, { "title": "Level 5: broader adaptation", "text": "The research program studies how a system can respond across a wider set of conditions while preserving purpose, evidence and authority.", "input": "Changed conditions beyond a fixed task script.", "output": "Candidate adaptation with separate acceptance.", "boundary": "A broader architecture is not a claim of unrestricted deployed autonomy." } ] }, { "id": "supervision-and-recovery", "kind": "mechanism", "title": "Supervision keeps failure local and recovery explicit.", "intro": "The runtime structure assigns failure, cancellation and compensation to the unit that owns the work.", "items": [ { "title": "Heartbeat supervision", "text": "A pipeline’s heartbeat bounds liveness. A canceled execution cannot acquire a new admitted effect merely because the underlying model is still producing text." }, { "title": "Evidence-bearing approvals", "text": "The gate distinguishes human approval, notification and bounded automatic action. Provenance, destination and the artifact under review belong to the decision." }, { "title": "Compensation and uncertainty", "text": "A partially completed workflow may require compensation. An external effect already sent cannot be undone by rewinding the local journal." }, { "title": "Outstanding obligations", "text": "A lost organ’s work remains named. A successor does not silently inherit an expanded mandate or erase an unresolved approval." } ] }, { "id": "continuity-and-development", "kind": "mechanism", "title": "A successor must retain more than a copy of memory.", "intro": "Manna’s opt-in continuity and lifecycle research implementations are not active by default. They examine causal ancestry, required carryover and policy behavior, and give organ growth a separate, constrained lifecycle.", "items": [ { "title": "Functional continuity", "text": "A valid causal chain and critical carryover are requirements, not qualities that a high score elsewhere can compensate for. Behavioral probes and capability self-description add separate checks.", "input": "Predecessor record, restored components and fixed probes.", "output": "A functional continuation judgment or a separately gated descendant.", "boundary": "This is an engineering convention for functional continuity, not a claim about conscious identity." }, { "title": "Finite organ growth", "text": "A new organ comes from a reviewed template. Grant reservation, persisted identity and activation are journalled separately so interrupted birth can be reconciled.", "input": "Reviewed template digest and available resource ceiling.", "output": "An identified allocation with a finite lifetime." }, { "title": "Verified restoration", "text": "A rollback certificate names the state that restoration must reproduce. The executor applies the inverse and checks the resulting state digest.", "input": "A target state and permitted inverse operation.", "output": "A verified restoration or an unresolved recovery requirement.", "boundary": "Irreversible external consequences require compensation, not fictional reversal." } ] }, { "id": "outcome-evaluation", "kind": "evidence", "title": "A continuing system returns to what happened.", "intro": "The assurance design keeps structural validity, policy, artifact checks and observed outcomes as four separate judgments.", "body": [ "A forecast names a target, horizon, population, calibration basis and expiry. An action names its grant and effect identity. Comparing the later outcome changes the evidence available for future work; it does not rewrite the authority that admitted the past action.", "Manna supplies Personal and Industrial editions around different responsibilities. Naevum’s own Anemone fund supplies an owner-reported setting for continuing multi-market operation. Black Box applies the same responsibility question to software delivery. Their domains, activation profiles and acceptance evidence remain specific to the system." ] } ], "questions": [ { "question": "Can an autonomous system remain under human authority?", "answer": "Yes. Continuing execution can operate inside a mandate while designated decisions, changes of scope and consequential actions remain with the responsible people." } ], "relationships": [ { "type": "uses", "id": "manna", "label": "Persistent agents", "text": "Purpose and continuing execution.", "url": "https://naevumlabs.com/products/manna" }, { "type": "research", "id": "anemone", "label": "Operating context", "text": "A cleared first-party application of consequential autonomy.", "url": "https://naevumlabs.com/lab/anemone" }, { "type": "delivery", "id": "black-box", "label": "Autonomous delivery research", "text": "A selected factory design with separate authoring and acceptance.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Scope an autonomous system", "href": "https://naevumlabs.com/contact?context=persistent-agents" } }, "anatomy": { "representation": "An illustrative method for preserving task and request identity through action and interruption.", "signature": "Retained task, bounded action and continuity", "parts": [ { "id": "contract", "name": "Declared work-contract frame", "description": "Keep intended work and boundaries explicit through continuation." }, { "id": "work", "name": "Supervised task workbench", "description": "Transform the task while exposing cancellation and local failure." }, { "id": "admission", "name": "Runtime effect-admission interface", "description": "Compare an identified proposal with actual applicable bounds and approval." }, { "id": "consequences", "name": "Causal effect and consequence register", "description": "Retain issued effects and distinguish receipts from outcomes." }, { "id": "continuity", "name": "Successor and restoration bench", "description": "Check carryover and restoration separately from the claim that they happened." }, { "id": "assurance", "name": "Assurance and outcome comparator", "description": "Keep different acceptance questions from becoming one green status." } ] } }, { "id": "technology-security-and-privacy", "name": "Security & privacy", "url": "https://naevumlabs.com/technology/security-and-privacy", "description": "Synapse binds exact source context, the destination’s cumulative observation and the identity of a consequential request. Agency Engine protects application authority; Metahive supplies explicit compute and custody choices.", "applications": [ { "title": "Use outside intelligence on a bounded task", "audience": "Institutions and R&D teams", "situation": "A useful model needs context, while the organization holds a much larger sensitive source.", "mechanism": "Select a source-bound projection and account for its disclosure to the actual observer before dispatch.", "impact": "The organization can inspect the approved context, purpose and destination of the model work." }, { "title": "Recover an interrupted agent attempt", "audience": "Software teams", "situation": "A connection drops after a request may already have reached its provider.", "mechanism": "The durable effect remains uncertain and reuses its exact identity during reconciliation.", "impact": "Recovery can establish what happened without silently repeating an action or forgetting prior exposure." }, { "title": "Make an organizational installation", "audience": "Enterprises", "situation": "A common application must reflect local permissions and operating context.", "mechanism": "Installation and organization layers supply context while policy can only narrow inherited authority.", "impact": "Local specialization retains a comprehensible relationship to the foundation’s controls." } ], "technical_features": [ { "name": "Source identity", "description": "Immutable source snapshots and transitive ancestry bind holder, origin and context to the session. Mutable paths alone do not identify the admitted bytes." }, { "name": "Canonical observer ledger", "description": "Account aliases map to a destination observer. Cumulative source-unit exposure is checked across that observer’s requests, including unresolved reservations." }, { "name": "Combination constraints", "description": "A candidate view is checked against prior source units and prohibited combinations. Per-request permission therefore does not erase cumulative disclosure." }, { "name": "Atomic effect reservation", "description": "The broker reserves effect identity and cumulative call/byte bounds before transport. Budgeting does not imply prepaid model-token accounting." }, { "name": "Exact request binding", "description": "Holder, confidentiality, destination, model, policy, expiry and stop state contribute to admission. A mismatched retry does not inherit an earlier approval." }, { "name": "Authenticated local storage", "description": "AES-GCM protects application payloads; HKDF derives keys in the inspected broker architecture. Metadata receipts remain distinct from encrypted plaintext." }, { "name": "No uncertainty refund", "description": "Lost acknowledgment retains its exposure and effect identity. A matching receipt supports reconciliation; abandonment does not remove disclosure history." }, { "name": "Product policy restriction", "description": "The application architecture declares the governance set non-replaceable and checks its known bindings at startup. Local configuration may narrow inherited authority." }, { "name": "Scope of assurance", "description": "The named deployment profile identifies supported clients, tools, descendants and containment. The architecture does not assert universal private coverage for arbitrary software." }, { "name": "Tenant/site cost reservation", "description": "Metahive’s selected admission architecture reserves a maximum cost before dispatch and settles actual usage. Uncertain jobs retain their reservation; monetary/token budgets differ from disclosure call/byte bounds." } ], "editorial": { "lead": "Synapse binds exact source context, the destination’s cumulative observation and the identity of a consequential request. Agency Engine protects application authority; Metahive supplies explicit compute and custody choices.", "audience_job": "Naevum’s security and privacy architecture is the boundary layer for CISOs, so that a model task sees only what is permitted and an agent’s actions are checked against its authority.", "chapters": [ { "id": "bounded-request", "kind": "process", "title": "Two permitted fragments can reveal a forbidden whole.", "intro": "An institution wants an external model to review one clause without exposing the full commercial agreement. A second request can reveal the missing relationship even when each fragment looks harmless alone. Synapse accounts for the observer across requests.", "items": [ { "title": "Fix the source", "text": "Admission records an immutable source snapshot, its ancestry, holder and origin. A tool result must retain the inputs from which it was derived.", "input": "Selected bytes and their provenance.", "output": "A source identity bound to the session.", "boundary": "Registering context does not itself authorize disclosure or host execution." }, { "title": "Resolve the observer", "text": "Accounts map to a canonical observer. The ledger records which source units that observer has already seen and checks their combination with the proposed view.", "input": "Observer identity, prior units and candidate units.", "output": "A cumulative disclosure decision.", "boundary": "A second account does not create a fresh privacy history for the same observer." }, { "title": "Reserve before transport", "text": "Disclosure, effect identity and finite request bounds are reserved durably before transport. An interruption retains the exposure rather than resetting it.", "input": "The exact approved request and remaining bounds.", "output": "A reserved effect followed by a receipt or uncertain state." } ] }, { "id": "source-and-information-flow", "kind": "mechanism", "title": "The boundary follows the data that contributes to a result.", "intro": "A useful privacy architecture needs to know where a derived answer came from and which authority can release it. Synapse’s source graph and admission controls make that relationship explicit.", "items": [ { "title": "Immutable context closure", "text": "The source closure records selected files, buffers, instructions and tool-result ancestry. Session binding prevents an unrelated context record being reused as though it belonged to the current task." }, { "title": "Integrity and confidentiality", "text": "Trust in an instruction and permission to disclose information answer different questions. Source labels and provenance survive derived context; a model-generated summary does not grant itself either property." }, { "title": "Bounded projection and assembly", "text": "The selected architecture can keep a larger source local while preparing an admitted view for a defined computation. Deterministic transformations and locally retained assembly are separate from model transport.", "input": "A source graph and explicitly selected task view.", "output": "An inspectable projection and result lineage.", "boundary": "The supported profile defines which transformations and model routes are qualified." } ] }, { "id": "transactional-effects", "kind": "mechanism", "title": "An uncertain request keeps its identity.", "intro": "The broker reserves request bounds and the effect binding in a database transaction. Source, destination, model, policy, expiry and stop state are checked against the requested consequence.", "items": [ { "title": "Reserve", "text": "One durable transaction binds the effect and accounts for cumulative call/byte bounds before transport. Concurrent attempts cannot spend the same unreserved allowance." }, { "title": "Receive or retain uncertainty", "text": "A received effect can reuse its matching receipt. If dispatch may have happened but no result is established, the record stays uncertain." }, { "title": "Reconcile", "text": "Recovery compares the existing binding and receipt. Abandonment does not erase earlier exposure or create a fresh allowance for the same information." }, { "title": "Protect payloads and control state", "text": "Stored application requests and model responses use authenticated encryption. Receipts expose the metadata needed for audit without copying plaintext into the receipt." } ] }, { "id": "deployment-choices", "kind": "comparison", "title": "Application authority and compute custody work at different layers.", "intro": "Agency Engine protects the application’s decision path. Metahive gives the deployment an explicit location, tenancy and operating responsibility. These compose with Synapse’s model-task boundary.", "items": [ { "title": "Protected application core", "text": "The sealed governance set includes approval, write, model-admission, disclosure and connector/store-custody services. Product extensions specialize around the declared set; the startup guard checks known bindings rather than providing process isolation." }, { "title": "Monotone local policy", "text": "Installation, organization and workspace context can tighten inherited policy. A product’s styling or local preference cannot widen its authority." }, { "title": "Dedicated operating boundary", "text": "A deployment identifies where computation runs, who administers it, how sources are retained and which effects cross the boundary. Compute location is one part of that contract." }, { "title": "Tenant and site budget admission", "text": "Metahive’s admission design reserves a maximum cost against tenant and site budgets before dispatch. Settlement records actual usage, while uncertain execution retains the reservation for reconciliation.", "input": "Tenant/site scope, an estimated maximum and remaining monetary/token budget.", "output": "A reserved job followed by usage settlement or retained uncertainty.", "boundary": "This budget contract is distinct from Synapse’s cumulative disclosure call/byte accounting; its integration belongs to the deployment profile." } ] }, { "id": "boundary-evaluation", "kind": "constraints", "title": "The deployment profile defines the protected route.", "intro": "Context admission, provider transport, tools, descendants and recovery each contribute to the actual boundary.", "body": [ "Naevum’s architecture binds source identity, admission, human oversight and durable evidence around a supported task route. The commissioned profile specifies clients, tools, runtime identities, limits and recovery conditions. Native-client and operating-system containment need their own qualification.", "In the synthetic document example, observer A may receive either fragment A or fragment B, but their combination violates the fixed rule. The same candidate is held after prior exposure and can be reserved without it. An interrupted admitted branch is reconciled under its existing identity. This demonstrates the mechanism without using a client’s document." ] } ], "questions": [ { "question": "Does local or dedicated compute make the whole system private?", "answer": "It addresses part of custody. Sources, model access, logging, tool effects, administration and recovery still require an explicit deployment design and evaluation." } ], "relationships": [ { "type": "uses", "id": "synapse", "label": "Context and effect controls", "text": "Observer history, admission and reconciliation.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "foundation", "id": "agency-engine", "label": "Application authority", "text": "Principal, recipient and approval rules.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "deployment", "id": "metahive", "label": "Infrastructure choices", "text": "Compute location and operating responsibility.", "url": "https://naevumlabs.com/products/metahive" } ], "action": { "label": "Discuss a private AI architecture", "href": "https://naevumlabs.com/contact?context=private-ai" } }, "anatomy": { "representation": "An illustrative method for comparing a selected view with observer history before an external task.", "signature": "Source projection, cumulative exposure and reconciliation", "parts": [ { "id": "source", "name": "Immutable source and ancestry", "description": "Fix selected material and its derivation before constructing an external view." }, { "id": "view", "name": "Local task-view assembly", "description": "Prepare an inspectable bounded projection under its supported profile." }, { "id": "exposure", "name": "Observer and exposure comparator", "description": "Compare new disclosure with prior and reserved exposure to the same observer." }, { "id": "reservation", "name": "Exact admission and reservation", "description": "Bind the admitted request before transport under all applicable limits." }, { "id": "transport", "name": "Supported transport and result boundary", "description": "Carry the reserved work through the specifically qualified route." }, { "id": "reconciliation", "name": "Local assembly and reconciliation", "description": "Retain the returned artifact, source lineage and uncertain effects." } ] } }, { "id": "technology-predictive-intelligence", "name": "Predictive & preventive systems", "url": "https://naevumlabs.com/technology/predictive-intelligence", "description": "Extrasensory connects survival modeling, changing operating conditions, calibrated uncertainty and constrained interventions. Vagus contributes a different prediction stack for task outcomes, duration and resources.", "applications": [ { "title": "Follow condition on an operating vessel", "audience": "Marine engineering", "situation": "Load, condition and sensor histories change while equipment remains in service.", "mechanism": "Exposure-aware diagnostics and time-to-event estimates connect the condition record to risk.", "impact": "The engineering team can investigate how operating context changes the evidence behind an assessment." }, { "title": "Start with a sparse failure history", "audience": "Industrial operators", "situation": "An asset has limited local failure data but belongs to an understood equipment class.", "mechanism": "Class priors and condition adjustments form an explicit cold-start basis.", "impact": "The operator can inspect what the estimate inherits and what the local observations contribute." }, { "title": "Compare the next useful intervention", "audience": "Asset-intensive organizations", "situation": "Inspection, continued operation and maintenance carry different practical constraints.", "mechanism": "Criticality and uncertainty inform a bounded comparison of available interventions.", "impact": "The decision can retain its assumptions and be revisited when subsequent evidence arrives." } ], "technical_features": [ { "name": "Longitudinal risk geometry", "description": "Operating exposure, censoring, delayed entry, competing events and recurrent intervals are represented with an explicit leakage policy." }, { "name": "Sparse-data prognosis", "description": "Condition-adjusted survival modeling combines equipment-class priors with applicable local observations. Maintenance countdown and inferred failure risk remain separate outputs." }, { "name": "Condition diagnostics", "description": "Multichannel inputs and load normalization preserve operating-regime context. Parameter feedback is bounded to declared tuning ranges." }, { "name": "Calibration basis", "description": "Grouped partitions and finite-sample conformal ranks retain their population, split and statistical assumptions. Coverage claims require matching evaluation evidence." }, { "name": "Intervention constraints", "description": "Criticality, uncertainty and feasible operational actions contribute to the recommendation. Prediction does not replace the operator’s authority." }, { "name": "Task-outcome posterior", "description": "Vagus uses hierarchical empirical-Bayes Beta–Binomial estimates for a named task target, with fitted and bounded prior strength." }, { "name": "Resource intervals", "description": "Log-normal token and wall-time models report medians and 80% intervals in the reference implementation. Prospective coverage is evaluated separately." }, { "name": "Prediction record", "description": "Target, input cutoff, horizon, units, scope, applicability and expiry remain distinct from the eventual outcome and from the action grant." }, { "name": "Model transfer scope", "description": "Asset, task-resource and simulated-state models retain separate labels and evaluators. Federated/representation-learning research is not presented as an operating general-purpose capability." } ], "editorial": { "lead": "Extrasensory connects survival modeling, changing operating conditions, calibrated uncertainty and constrained interventions. Vagus contributes a different prediction stack for task outcomes, duration and resources.", "audience_job": "Predictive and preventive systems are forecasts with stated uncertainty for grid and asset operators, so that they act before a peak or a failure and can check the forecast afterward.", "chapters": [ { "id": "useful-forecast", "kind": "process", "title": "A maintenance date and a failure forecast answer different questions.", "intro": "Two assets can have the same service interval and different operating histories. A signal can rise because load changed, because condition deteriorated or because the measurement is unreliable. Our predictive architecture keeps those explanations separate before relating them to an intervention.", "items": [ { "title": "Reconstruct exposure", "text": "Longitudinal histories retain operating exposure, event time, changing conditions and incomplete observations. Surviving assets contribute evidence without being labeled as failures.", "input": "Asset/context history and event observations.", "output": "An explicit population at risk over time." }, { "title": "Estimate under the current condition", "text": "Equipment-class priors provide a cold-start basis. Condition adjustments add local evidence while keeping the inherited assumptions visible.", "input": "Class information and applicable local observations.", "output": "A survival/risk estimate with its evidence basis." }, { "title": "Carry uncertainty into the decision", "text": "The model retains horizon, applicability and uncertainty when comparing inspection, maintenance and continued operation.", "input": "A forecast plus criticality and available interventions.", "output": "A bounded recommendation whose assumptions remain inspectable." }, { "title": "Return to the observed outcome", "text": "Stored forecasts precede later outcomes, preserving the basis for calibration and intervention review.", "input": "A recorded prediction and subsequent event history.", "output": "An attributable prediction/outcome comparison." } ] }, { "id": "longitudinal-model", "kind": "mechanism", "title": "The event history determines what the model is learning.", "intro": "Extrasensory’s general-purpose architecture carries explicit longitudinal training geometry. The maritime predictive/preventive application is one operating context; domain adapters retain their own equipment and observation rules.", "items": [ { "title": "Censoring and delayed entry", "text": "A history ending without a failure is not a negative event at every future horizon. An asset first observed part-way through its life enters the risk population at that later time." }, { "title": "Competing and recurrent events", "text": "Different events may terminate a period of exposure, while a repaired or repeatedly serviced asset can produce further event intervals. Their labels and risk sets must retain that distinction." }, { "title": "Leakage-aware data geometry", "text": "Predictor inputs are separated from future outcome information. Grouped partitions keep evaluation populations and calibration assumptions explicit." }, { "title": "Maintenance and inferred risk", "text": "A calendar or hours-based countdown can coexist with a survival estimate. The architecture does not collapse either into the other." } ] }, { "id": "condition-and-calibration", "kind": "comparison", "title": "Sparse data still needs an explicit evidence basis.", "intro": "The system can combine class information with operating observations while showing which part of the estimate comes from each.", "items": [ { "title": "Class priors and maturation", "text": "A cold start uses an equipment-class basis. Accumulated applicable evidence changes the local estimate and its stated maturity rather than inventing local failure labels." }, { "title": "Multichannel diagnostics", "text": "Load normalization and condition channels distinguish an operating regime change from an anomaly. Bounded feedback tuning changes declared parameters within limits." }, { "title": "Grouped conformal calibration", "text": "Finite-sample rank-based calibration retains its partition and statistical assumptions. The existence of a conformal method does not create unconditional coverage under arbitrary distribution shift." }, { "title": "Criticality-aware response", "text": "The intervention layer compares feasible actions under operational constraints. A high predicted risk does not itself grant actuation authority." } ] }, { "id": "different-predictions", "kind": "mechanism", "title": "Task outcomes have a different forward model.", "intro": "Vagus predicts whether a defined episode ends well and how many resources it is likely to consume. Its proxy/0.2 target is an episode completed without an observed interruption or red-test/failure ending; independently labeled outcome_ok is separate. Neither is an automatic measure of task correctness. It does not apply an asset-failure model to a software task.", "items": [ { "title": "Hierarchical outcome probability", "text": "A Beta–Binomial hierarchy links portfolio, holder, repository, task kind and worker. Sparse cells inherit a bounded prior from related observations without counting the same branch twice.", "input": "Task features available before execution.", "output": "A declared outcome probability and evidence basis." }, { "title": "Resource distributions", "text": "Log-normal models produce token and wall-time medians with 80% intervals in the inspected implementation. These are nominal model intervals; estimates remain distinct from budgets, observed coverage and guarantees." }, { "title": "Prequential scoring", "text": "The ledger scores a prediction made before its outcome. Unknown endings stay unlabeled; target definitions and a running-base-rate comparison remain explicit. Brier score, log loss and reliability bins are compared with a running base rate; uncertainty is clustered by root session and subgroup verdicts are Holm-adjusted." }, { "title": "Predictive planning research", "text": "Manna’s typed world-model boundary links a forecast to an objective, scope, horizon and expiry before a bounded proposal can be admitted.", "input": "A scoped belief and current grant.", "output": "A proposal whose motivating forecast can be compared later.", "boundary": "The research organ’s prediction is separate from action authority and installed adoption." } ] }, { "id": "intervention-example", "kind": "narrative", "title": "An operator can see what would change the recommendation.", "intro": "In an illustrative operating episode, a vessel’s condition signal rises during a higher-load period. The explanation follows the operating history, the applicable condition model and the uncertainty around the decision.", "body": [ "The intervention record distinguishes the load-normalized evidence from the maintenance countdown, names the forecast horizon and compares the available response. A later inspection can supply a new observation without rewriting what the system knew when it made the earlier recommendation.", "That is the common technology path from prediction to prevention. Data sources, event definitions, calibration population and intervention permissions remain specific to the commissioned domain. The general-purpose edition does not inherit every vessel-specific implementation by naming it." ] } ], "questions": [ { "question": "What makes a prediction trustworthy?", "answer": "Evidence of calibration and usefulness for the specified domain, horizon and decision. Prospective evaluation should include false alarms, missed events and whether the response was useful." } ], "relationships": [ { "type": "uses", "id": "extrasensory", "label": "Preventive decisions", "text": "Connect operating history to a useful intervention.", "url": "https://naevumlabs.com/products/extrasensory" }, { "type": "related", "id": "vagus", "label": "Outcome models", "text": "Estimate tasks and resources from retained evidence.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "research", "id": "manna", "label": "Bounded predictive organ", "text": "Synthetic scheduling research with recorded forecasts.", "url": "https://naevumlabs.com/products/manna" } ], "action": { "label": "Explore predictive operations", "href": "https://naevumlabs.com/contact?context=predictive-operations" } }, "anatomy": { "representation": "An illustrative method for retaining a forecast before comparing independently supplied later evidence.", "signature": "Separate domain forecasts and observed outcomes", "parts": [ { "id": "history", "name": "Target and eligible history", "description": "Fix the prediction question and reconstruct only eligible evidence." }, { "id": "diagnosis", "name": "Condition and context diagnosis", "description": "Separate operating-load response, persistent condition difference and unreliable sensing." }, { "id": "models", "name": "Domain estimate bank", "description": "Apply the appropriate explicit domain basis without merging unlike targets." }, { "id": "calibration", "name": "Calibration and applicability", "description": "Carry the partition, uncertainty and support conditions with the estimate." }, { "id": "responses", "name": "Feasible-response comparator", "description": "Compare informative observation, inspection or intervention proposals under real constraints." }, { "id": "outcomes", "name": "Prospective outcome register", "description": "Freeze the prediction before the decision and compare genuinely later evidence." } ] } }, { "id": "technology-memory-and-knowledge", "name": "Memory & knowledge systems", "url": "https://naevumlabs.com/technology/memory-and-knowledge", "description": "Vagus preserves the time, source and authority of knowledge. Agency Engine carries domain context and memory changes. The Vault retains reusable methods with their assumptions; Manna carries continuing purpose across work.", "applications": [ { "title": "An institution remembers its decisions", "audience": "Institutions", "situation": "A policy changes, but earlier claims and the reasons for past decisions still matter.", "mechanism": "Validity intervals, authority ancestry and contested claims preserve the history of understanding.", "impact": "A new decision can distinguish the current position from an earlier or unsupported one." }, { "title": "Find the mechanism before building again", "audience": "R&D and software teams", "situation": "A new problem may overlap earlier engineering in another project.", "mechanism": "Typed component and method registers retain contracts, scientific references and lineage.", "impact": "The team can inspect a candidate and the evidence behind its reuse." }, { "title": "Keep a small team’s commercial memory", "audience": "Founders and small teams", "situation": "A prospect moves between a public page, a question and a private conversation.", "mechanism": "Person, offer, touch and conversation records share one commercial context.", "impact": "The team can continue the relationship without repeatedly reconstructing the business or the discussion." } ], "technical_features": [ { "name": "Bitemporal claim query", "description": "Validity intervals and recorded-at cutoffs distinguish the world at a time from what the system knew at a time." }, { "name": "Authority inheritance", "description": "Derived authority is capped by contributing ancestry. Missing parents, cycles and unsupported derivations do not promote a claim." }, { "name": "Claim operation algebra", "description": "Supersede, revise, contest, suppress and retract affect applicability and recall differently. Conflicting top-ranked claims remain non-binding." }, { "name": "Holder-scoped integrity", "description": "Append-only keyed chains retain previous-record linkage and verified-prefix status. Storage boundaries remain attached to recall." }, { "name": "Deterministic retrieval", "description": "SQLite FTS5 supports a rebuildable index. Ranking includes relevance, binding authority, scope distance and ACT-R-inspired use activation." }, { "name": "Recall budget and abstention", "description": "Item and byte limits bound the returned view. Unreadable, withheld and absent results retain reasons rather than silently becoming an empty success." }, { "name": "Use-aware consolidation", "description": "Projection, failure memory, hygiene, index rebuilding, packs and examination are distinct stages. Canonical claims and their operations remain the historical basis." }, { "name": "Measured failure memory", "description": "Defined observed harms contribute to a decayed failure weight. Quarantine and release use different thresholds, and transitions retain their cause." }, { "name": "World-model scope", "description": "Connected sources, supported identity inferences, preferences, responsibilities and objectives form a scoped representation. The person or organization can inspect and correct that representation." }, { "name": "Two-layer convergent memory", "description": "An observed-remove set accumulates claim/origin elements by union. A deterministic read-time interpretation resolves the evidence without replacing the historical set." }, { "name": "Typed maintenance and forgetting", "description": "Decayed, never-captured, unlearned and superseded remain distinct. Recurrence, retrieval spacing, promotion and retention are separate policy decisions." } ], "editorial": { "lead": "Vagus preserves the time, source and authority of knowledge. Agency Engine carries domain context and memory changes. The Vault retains reusable methods with their assumptions; Manna carries continuing purpose across work.", "audience_job": "Memory and knowledge is dated, sourced memory for compliance leads, so that any answer can be read as of the day it mattered, with later changes shown apart.", "chapters": [ { "id": "evolving-world", "kind": "narrative", "title": "An earlier decision needs its earlier world.", "intro": "An institution changes a policy on Thursday. A reviewer needs the rule that applied on Monday and what was known when Monday’s decision was made. The next task needs the current rule. A single current document or a plausible summary cannot answer all three questions.", "body": [ "Our knowledge architecture separates source records, claims, their validity and the time they were observed. A dedicated world model connects those records to identities, responsibilities, objectives and the systems a person or organization chooses to connect.", "Vagus resolves time and authority. Agency Engine maintains product and organization context through versioned memory operations. The Vault carries methods, assumptions and evidence into reuse. This gives an intelligence environment more structure than an archive of conversation text." ], "intro_role": "editorial-statement" }, { "id": "memory-distinctions", "kind": "mechanism", "title": "Time and authority are part of the data model.", "intro": "Vagus’s claim algebra separates what a statement says from who supplied it, when it applies and what operations later changed its status.", "items": [ { "title": "Two time axes", "text": "Validity time answers when a claim applied. Observation time answers when it entered the record. An as-of query can preserve the earlier state of knowledge.", "input": "Claims and operations with timestamps.", "output": "The applicable evidence at the requested time basis." }, { "title": "Authority by ancestry", "text": "The effective authority of a derived claim is bounded by the minimum authority along its contributing ancestry. A missing ancestor or cycle cannot create stronger evidence." }, { "title": "Conflict stays visible", "text": "Claims of equal scope and authority that disagree can remain contested. Supersede, revise, contest, suppress and retract carry different meanings." }, { "title": "Recall suppression is not historical deletion", "text": "Suppressing a claim from ordinary recall is distinct from changing its validity or retracting it. The canonical operation record retains the distinction." } ] }, { "id": "recall-and-integrity", "kind": "mechanism", "title": "The index can be rebuilt. The evidence remains canonical.", "intro": "The memory store uses holder-scoped, append-only records with keyed integrity chains. Its retrieval index is a derived view rather than a second editable history.", "items": [ { "title": "Rebuildable search", "text": "A SQLite FTS5 index combines claim, lesson, failure and episode pointers. Rebuild writes a temporary index and swaps it into place atomically.", "input": "Verified canonical records.", "output": "A replaceable retrieval index." }, { "title": "Bounded recall", "text": "Recall limits both items and bytes. Ranking combines textual relevance, authority, scope and use-based activation; a result retains its matched evidence and source pointer.", "input": "A query, scope and output budget.", "output": "Evidence-bearing items, withheld items or an abstention reason." }, { "title": "Integrity-aware reading", "text": "Readers use the verified prefix of the record chain and retain a finding for unverified material. A broken tail cannot silently become trusted context." }, { "title": "Partitioned use", "text": "Holder boundaries travel with storage and retrieval. A useful match does not grant permission to disclose the material or act on it." } ] }, { "id": "convergent-domain-memory", "kind": "mechanism", "title": "Distributed memory preserves the disagreement before interpreting it.", "intro": "Agency Engine separates accumulation from interpretation. Its per-origin set merge preserves the information needed to distinguish a changed fact, an incorrect belief and a disagreement between sources.", "items": [ { "title": "Convergent accumulation", "text": "The data layer merges an observed-remove set by union. The element identity contains both the claim and its origin, so two sources asserting the same fact remain two attributable elements.", "input": "Claim/origin elements and their memory operations.", "output": "A convergent accumulated set with source identity retained." }, { "title": "Deterministic interpretation", "text": "A pure read-time function interprets the canonically ordered set under scope, authority, temporal validity and conflict rules. A revised interpretation can rebuild the projection without rewriting the underlying history." }, { "title": "Typed forgetting", "text": "Decayed, never-captured, unlearned and superseded describe different events. Memory maintenance can nominate a change while preserving the historical source and its approved validity operations." }, { "title": "Generalization before promotion", "text": "A recurring cluster can justify proposing a generalization. Promotion, retrieval evidence and retention have separate contracts; repeated access inside one concentrated session does not become independent spaced evidence.", "boundary": "The inspected thresholds are declared policy choices. Future utility and retention are evaluated for the domain." } ] }, { "id": "memory-in-work", "kind": "process", "title": "Experience becomes useful through distinct operations.", "intro": "Agency Engine separates recall, proposed generalization, memory maintenance and typed forgetting. Vagus’s failure memory and the Vault’s evidence contracts retain different kinds of consequence.", "items": [ { "title": "Keep the event", "text": "An episode or attributed claim retains the source behind a decision, correction or failure. Repeated copies do not become independent supporting origins." }, { "title": "Maintain the knowledge", "text": "Duplicate detection, conflict annotations, decay nomination and approved validity changes are explicit operations. A tidy summary cannot silently replace the source history." }, { "title": "Retain a measured failure", "text": "Vagus records defined harms with decay and hysteresis. Restrictive failure memory can keep a known failing context from being treated as a fresh unexamined choice." }, { "title": "Carry assumptions into reuse", "text": "A Vault method keeps its contracts and evidence lineage. Unresolved composition assumptions remain visible to the next application." } ] }, { "id": "knowledge-and-learning", "kind": "comparison", "title": "Remembering and retraining change different things.", "intro": "Proprietary operating knowledge can improve the context for future work. Updating a model is a separate intervention with a different data and evaluation contract.", "items": [ { "title": "World-model and retrieval updates", "text": "A corrected policy or preference changes the records and context used for future tasks. The underlying foundation-model weights can remain fixed." }, { "title": "Calibration and procedure updates", "text": "Recorded outcomes can change a task estimate, selection policy or approved workflow. Their acceptance basis is specific to the recurring task." }, { "title": "Model/adaptor research", "text": "A trainable model can be studied using permitted examples and held-out future work. Retention, exposure, cost and recovery belong to that study.", "input": "An eligible model and authorized data use.", "output": "A candidate evaluated against retrieval/no-update baselines.", "boundary": "This service research direction does not assert deployed recursive weight training." } ] } ], "questions": [ { "question": "Is the world model a complete digital copy of a person?", "answer": "No. It is a scoped representation of connected sources, explicit preferences and supported inferences. Its contents and limits must remain inspectable and correctable." } ], "relationships": [ { "type": "uses", "id": "vagus", "label": "Temporal memory", "text": "History, evidence and attention.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "foundation", "id": "agency-engine", "label": "Domain context", "text": "Product and organization knowledge.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "uses", "id": "vault", "label": "Reusable knowledge", "text": "Mechanisms and their conditions.", "url": "https://naevumlabs.com/products/vault" }, { "type": "related", "id": "manna", "label": "Continuity", "text": "Purpose and state across interruptions.", "url": "https://naevumlabs.com/products/manna" } ], "action": { "label": "Discuss your knowledge environment", "href": "https://naevumlabs.com/contact?context=intelligence-environment" } }, "anatomy": { "representation": "An illustrative method for retaining canonical evidence and answering scoped historical questions.", "signature": "Source history, temporal interpretation and recall", "parts": [ { "id": "records", "name": "Attributed record store", "description": "Preserve source-connected history with two distinct time axes." }, { "id": "interpretation", "name": "Claim-operation interpreter", "description": "Apply the actual semantics of revision, contest and historical use." }, { "id": "retrieval", "name": "Bounded retrieval and indexes", "description": "Assemble relevant evidence without confusing derived lookup with canonical memory." }, { "id": "relations", "name": "Contextual relation assembly", "description": "Relate the selected evidence to the right responsibility and objective." }, { "id": "maintenance", "name": "Maintenance and promotion", "description": "Annotate, test and retain useful derived structures without rewriting history." }, { "id": "evaluation", "name": "Evaluation and reuse register", "description": "Keep actual utility, failures and subsequent-use evidence available." } ] } }, { "id": "technology-simulation-and-worlds", "name": "Simulation & living worlds", "url": "https://naevumlabs.com/technology/simulation-and-worlds", "description": "Metahive Builder couples spatial design to power, heat, inference and energy calculations. Everbloom couples generated campaigns to shared roles, durable settlement and world history. Both preserve the rules beneath the experience.", "applications": [ { "title": "Rule a kingdom together", "audience": "Players", "situation": "A King chooses a response to a threat while a Queen shapes the kingdom’s cultural life.", "mechanism": "Their decisions branch into tasks for court roles and clubs, with shared consequences.", "impact": "One decision becomes an activity people can undertake together in a changing place." }, { "title": "Leave something in the world’s history", "audience": "Players and world creators", "situation": "A successful reign should remain meaningful after its rulers change.", "mechanism": "The Chronicle connects actual contributions and decisions to persistent world records.", "impact": "Participation can become a recognizable part of the kingdom’s continuing story." }, { "title": "Compare an energy and investment scenario", "audience": "Operators and investment planners", "situation": "A denser design changes electrical demand, heat, annual energy and cost assumptions.", "mechanism": "Power and thermal propagation share a scenario basis with annual weather, supply and storage.", "impact": "The decision retains units, input quality and sensitivity instead of presenting one unexplained ROI figure." } ], "technical_features": [ { "name": "Spatial model hierarchy", "description": "Device, rack, room, building and plot have containment and placement relationships. Geometry checks include overlap, aisles, mass, floor load and land-use assumptions." }, { "name": "Provenance-carrying values", "description": "A numeric input retains value, unit, source/assumption tag, source identity and low/high basis. Derived values retain method and input lineage." }, { "name": "Distinct power semantics", "description": "Installed nameplate, provisioned peak and utilization-based average power serve different checks. Electrical losses and cooling load propagate to facility demand." }, { "name": "KV-memory equation", "description": "Per-token KV bytes = 2 × layers × KV heads × head dimension × bytes per element. Full-context concurrency is bounded by remaining accelerator memory." }, { "name": "Inference roofline", "description": "Decode time is bounded by memory transfer and compute across a batch sweep. Throughput, latency limits and joules/token retain the selected utilization and facility assumptions." }, { "name": "Hourly supply/storage model", "description": "Load, generation, imports, exports, battery charge/discharge, curtailment and shortfall are balanced at each time step. State of charge respects power, capacity and efficiency." }, { "name": "Scenario uncertainty", "description": "Coherent base/low/high input cases rerun the model. Input quality and limiting assumptions remain attached to the result." }, { "name": "Validated world generation", "description": "A reign/day-bound request passes structured output checks and bounded repair. Accepted-result persistence and authored fallback preserve a coherent generation identity." }, { "name": "Server-authoritative settlement", "description": "Role cascades separate decisions from effects. Idempotent payouts, treasury conservation and Chronicle history retain persistent consequences." } ], "editorial": { "lead": "Metahive Builder couples spatial design to power, heat, inference and energy calculations. Everbloom couples generated campaigns to shared roles, durable settlement and world history. Both preserve the rules beneath the experience.", "audience_job": "Simulation & worlds is Naevum’s work on rule-based models for planners and world builders, so that every change in a simulated site or world traces back to its rules.", "chapters": [ { "id": "two-worlds", "kind": "comparison", "title": "A layout and a living world both carry state.", "intro": "A denser rack changes memory capacity, power and cooling. A royal choice changes the work available to a kingdom. A scene is useful when those consequences follow the model’s rules rather than a visual effect.", "items": [ { "title": "A physical design model", "text": "Builder’s device → rack → room → building → plot hierarchy ties geometry to declared equipment and site assumptions. A layout change propagates into feasibility and operating calculations.", "input": "Placed equipment, workload and tagged assumptions.", "output": "Traceable design-point and hourly scenarios." }, { "title": "A persistent social world", "text": "Everbloom connects a reign and day to generated events, role-specific tasks, settlement and the Chronicle. The accepted state becomes part of later play.", "input": "Royal choices, permitted world context and player contributions.", "output": "Coherent tasks and durable world history." } ] }, { "id": "physical-calculation-graph", "kind": "mechanism", "title": "Builder follows a number back to its assumptions.", "intro": "Each calculated value names a method and its inputs. Tagged inputs carry a value, unit, source/assumption status and optional low/high bounds. Changing a scenario reruns the connected calculation graph.", "items": [ { "title": "Spatial feasibility", "text": "Containment, overlap, aisle clearance, rack mass, floor load and site footprint relate the placed objects to physical constraints. Checks distinguish a sourced limit from a recorded design assumption." }, { "title": "Three power bases", "text": "Installed nameplate supports breaker and fit checks. Provisioned peak sizes the load. Utilization-based average power supports energy estimates. Those three values remain distinct." }, { "title": "Power and thermal propagation", "text": "The electrical tree adds conversion/distribution losses. Air/liquid heat splits, airflow, coolant flow and cooling power propagate into the facility demand." }, { "title": "Uncertainty through recomputation", "text": "Base, low and high scenarios rerun coherent sets of tagged inputs. The result retains limiting assumptions and source quality instead of adding an unexplained percentage to the final answer." } ] }, { "id": "inference-and-energy", "kind": "mechanism", "title": "Inference capacity reaches all the way to the site.", "intro": "A model’s weights are only one memory demand. Context length, concurrency and the device’s memory/compute limits change what a configuration can actually serve.", "items": [ { "title": "KV cache and concurrency", "text": "KV bytes per token = 2 × layers × KV heads × head dimension × bytes per element. Context length and concurrent sequences multiply that demand. Remaining accelerator memory bounds the feasible batch.", "input": "Model shape, precision, context and available memory.", "output": "Memory placement and concurrency bounds." }, { "title": "Roofline decode bounds", "text": "The decode step uses the slower of the memory-transfer and compute times. Batch sweeps retain the limiting regime and distinguish theoretical roofs from utilization-adjusted estimates." }, { "title": "Facility energy per token", "text": "Design-point facility power divided by estimated site throughput produces joules per generated token for that scenario. A memory-infeasible configuration cannot be made feasible by an attractive energy figure." }, { "title": "Hourly energy balance", "text": "Generation, grid exchange, battery charge/discharge, export, curtailment and unmet load share one hourly balance. Battery state of charge retains capacity, power and efficiency bounds.", "input": "Load and generation series plus storage/grid assumptions.", "output": "Annual energy quantities and a traceable shortfall record.", "boundary": "Calculated scenarios are distinct from commissioned hardware measurements." } ] }, { "id": "rules-and-generation", "kind": "process", "title": "Everbloom separates generated variation from authoritative state.", "intro": "A King or Queen’s decision can become a campaign and shared work. Generative content enters through a bounded shape; server-side rules retain identity, settlement and economic consequences.", "items": [ { "title": "Generate a bounded campaign", "text": "The generation path ties the request to reign/day context and uses a defined input shape. Generated event content is converted into task structure.", "input": "The permitted campaign context.", "output": "A candidate event and role-linked tasks." }, { "title": "Validate and settle identity", "text": "Schema checks and bounded repair precede acceptance. An accepted generated result uses first-writer persistence; refusal can select an authored fallback." }, { "title": "Run the consequence cascade", "text": "The cascade runner separates decision logic from effects. Role-specific work can change world state without making generated prose the authority over the economy." }, { "title": "Retain settlement and history", "text": "Task payouts are idempotent; treasury conservation and durable Chronicle records retain what happened. Player contribution becomes part of the world’s continuing history." } ] }, { "id": "simulation-fit", "kind": "constraints", "title": "The model exposes its basis as well as its result.", "intro": "Builder reports source assumptions and method lineage. Everbloom retains the accepted world identity and settlement history. Manna’s planning research uses a bounded synthetic environment with explicit observation and action contracts.", "body": [ "These models serve different decisions. Physical planning explains why a configuration fits or fails and how a workload changes the site. A living world explains how a choice becomes coherent shared activity. Predictive-organ research examines future states before proposing an action.", "A custom simulation engagement selects the state, transitions, inputs and outcomes relevant to its real setting. The existing technology supplies a starting architecture with identifiable rules; the relationship to physical measurements or observed player outcomes is evaluated for that application." ] } ], "questions": [ { "question": "Can the same approach support training or planning?", "answer": "Yes, when the engagement defines the world’s rules, user decisions, outcome measures and the relationship between simulated behavior and the real setting." } ], "relationships": [ { "type": "uses", "id": "metahive-builder", "label": "Physical planning", "text": "Coupled configuration and workload scenarios.", "url": "https://naevumlabs.com/products/metahive/builder" }, { "type": "uses", "id": "everbloom", "label": "Participatory world", "text": "Roles, generation and persistent consequences.", "url": "https://naevumlabs.com/products/everbloom" }, { "type": "research", "id": "manna", "label": "Bounded simulation", "text": "A controlled environment for scheduling research.", "url": "https://naevumlabs.com/products/manna" } ], "action": { "label": "Discuss a simulation or living world", "href": "https://naevumlabs.com/contact?context=simulation" } }, "anatomy": { "representation": "Two illustrative methods retain separate physical calculations and completed world consequences.", "signature": "Physical scenarios and event-driven world changes", "parts": [ { "id": "state", "name": "State and assumptions", "description": "Retain the selected domain state, source basis and current identity." }, { "id": "change", "name": "Controlled change and input", "description": "Introduce one explicit physical alternative or a separately chosen world decision." }, { "id": "rules", "name": "Domain rule and transition workbench", "description": "Use different internal operations for physical calculations and world contributions." }, { "id": "comparison", "name": "Constraint and settlement comparator", "description": "Keep calculated feasibility distinct from admitted durable world change." }, { "id": "history", "name": "Scenario and history register", "description": "Retain comparisons and actual accepted consequences." } ] } } ], "applications": [ { "id": "solution-custom-ai-systems", "name": "Custom AI systems", "url": "https://naevumlabs.com/solutions/custom-ai-systems", "description": "Your systems contain much of what your organization knows. The difficult part is connecting that knowledge to the work that needs to happen next.", "systems": [ "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/products/manna", "https://naevumlabs.com/products/agency-engine", "https://naevumlabs.com/products/black-box" ], "editorial": { "lead": "Intelligence, native to your world.", "audience_job": "Custom AI systems is Naevum's build path for leaders whose work no rented tool fits, so that their folders, memory and model use become one stack they own.", "chapters": [ { "id": "your-world", "kind": "narrative", "title": "It starts with the world you already work in.", "intro": "Your systems contain much of what your organization knows. The difficult part is connecting that knowledge to the work that needs to happen next.", "body": [ "We design a dedicated environment around your sources, people, infrastructure and objectives. A versioned world model connects identity and digital footprint to memory, reasoning and permitted action.", "Hyperpersonalization comes from that inspectable context: the responsibilities the system respects, the preferences you can change and the tasks by which you judge its usefulness. Each engagement selects and evaluates the capabilities it needs." ], "intro_role": "editorial-statement" }, { "id": "understanding", "kind": "mechanism", "title": "Context that keeps its history.", "intro": "An answer needs its source and the time when it applies.", "items": [ { "title": "Connected sources", "text": "Choose systems and records within an agreed access boundary. Monitoring cadence and freshness remain visible." }, { "title": "Correctable understanding", "text": "Distinguish observations from inferences. Define correction, retention and deletion across source-connected indexes." }, { "title": "Useful action", "text": "Turn a named objective into an admitted workflow with a result, operator controls and a recovery path." } ] }, { "id": "adaptation", "kind": "evidence", "title": "The next decision has more to learn from.", "intro": "Your business has diagnostic skill, accepted decisions and failed experiments whose meaning depends on its own operating context. A dedicated learning study can examine how that experience should inform the next task.", "body": [ "We can scope a study around one recurring task. An engineering team might bring expert-corrected diagnostic drafts and the observed results of interventions; an R&D team might bring a reproducible method and the variants that failed.", "The study chooses what should change: source-connected memory, a procedure, a calibrated estimate or a trainable model. Targeted adaptation is considered when the selected model supports it and the examples are permitted for that use.", "We agree a no-update baseline and a source-grounded retrieval comparison, then evaluate new work, retained capabilities, data exposure and total cost. An accepted candidate needs its own version, promotion decision and recovery path. The evaluation determines the next operating decision." ], "items": [ { "title": "Keep the business evidence intact", "text": "Retain expert corrections, outcomes and negative findings with their source, time and permitted use." }, { "title": "Choose the layer that needs to learn", "text": "Compare memory and procedure changes with targeted model adaptation on the same task and evidence basis." }, { "title": "Judge the next work independently", "text": "Reserve unseen cases and earlier accepted tasks. Evaluate useful behavior, retained knowledge, cost and exposure before promoting a candidate." } ] }, { "id": "first-working-slice", "kind": "process", "title": "Start with one decision that keeps coming back.", "intro": "Illustrative first engagement: a procurement lead needs to explain changed supplier commitments using a bounded set of approved policy and supplier documents.", "items": [ { "title": "Bound the sources and responsibility", "text": "The procurement lead selects the documents, the recurring question and the decision that remains theirs. Record access rights, excluded material, freshness and the deployment boundary.", "input": "An approved policy set, supplier revisions and a recurring planning question.", "output": "A source/rights map and an agreed evaluation question set." }, { "title": "Follow one changed commitment", "text": "A supplier revises a delivery date. The working slice recovers the earlier answer from the earlier evidence and proposes a current planning update from the revised source.", "output": "An explainable answer with source and validity, plus a proposed next step.", "boundary": "The procurement lead retains the procurement decision." }, { "title": "Deliver and examine the slice", "text": "Deliver the bounded application, deployment/recovery instructions and acceptance evidence. Test historical answers, excluded-source refusal, correction propagation and the agreed interrupted-work case.", "output": "A useful scoped system and a documented decision about the next increment.", "boundary": "This is an engagement design, not a universal integrated installation claim." } ] } ], "questions": [ { "question": "Do we need to replace our existing systems?", "answer": "The first task is to map the systems and workflows you already use. The architecture determines which integrations are useful and where a new component is justified." }, { "question": "What will we own and operate?", "answer": "Source, data custody, deployment, intellectual-property terms, support and operating responsibilities are agreed for the engagement. Discovery makes those choices concrete." } ], "relationships": [ { "type": "uses", "id": "vagus", "label": "Memory and attention", "text": "Temporal context around actual work.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "uses", "id": "synapse", "label": "Model and action boundaries", "text": "Control what context reaches an observer.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "uses", "id": "manna", "label": "Continuing purpose", "text": "Work that can persist between instructions.", "url": "https://naevumlabs.com/products/manna" }, { "type": "foundation", "id": "agency-engine", "label": "Application experience", "text": "Domain workflows and interfaces.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "delivery", "id": "black-box", "label": "System delivery", "text": "Build and hand over the selected architecture.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Discuss your environment", "href": "https://naevumlabs.com/contact?context=intelligence-environment" } } }, { "id": "solution-enterprise-institutional", "name": "Enterprise & institutions", "url": "https://naevumlabs.com/solutions/enterprise-institutional", "description": "An institution asks why a commitment changed, which policy applied then and who owns the next decision.", "systems": [ "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/products/agency-engine" ], "editorial": { "lead": "Bring institutional knowledge into the decision.", "audience_job": "Enterprise & institutions is Naevum's Tier 3 stack for organizations that must explain their decisions, so that every answer shows the rule, the date and the person behind it.", "chapters": [ { "id": "recurring-question", "kind": "narrative", "title": "Recover the reason, not just the document.", "intro": "An institution asks why a commitment changed, which policy applied then and who owns the next decision.", "body": [ "A source-connected memory can retain current and historical claims, disagreements and the evidence behind them. An attention layer can surface new conditions that make an earlier decision worth revisiting." ], "intro_role": "editorial-statement" }, { "id": "institutional-boundaries", "kind": "mechanism", "title": "Design the responsibility around the knowledge.", "intro": "Useful institutional intelligence depends on roles, custody and the actual workflow.", "items": [ { "title": "Map access", "text": "Specify who may see which sources and how current those sources need to be." }, { "title": "Scope model context", "text": "Select the context needed for a task and account for its destination and history." }, { "title": "Return an accountable answer", "text": "Present the evidence, uncertainty and permitted next action to the responsible person." } ] } ], "questions": [ { "question": "How is value evaluated?", "answer": "Start with recurring institutional questions and decisions. Test correct source retrieval, temporal accuracy, useful attention and successful completion of the intended workflow." } ], "relationships": [ { "type": "uses", "id": "vagus", "label": "Temporal institutional memory", "text": "Claims, attention and outcome context.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "uses", "id": "synapse", "label": "Sensitive model workflows", "text": "Source, observer and effect boundaries.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "foundation", "id": "agency-engine", "label": "Organization workspace", "text": "Roles and domain-specific processes.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Discuss an institutional system", "href": "https://naevumlabs.com/contact?context=institutional-memory" } } }, { "id": "solution-autonomous-systems", "name": "Personal & industrial autonomy", "url": "https://naevumlabs.com/solutions/autonomous-systems", "description": "Continuing work becomes useful when its authority fits the setting.", "systems": [ "https://naevumlabs.com/products/manna", "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/synapse" ], "editorial": { "lead": "Keep the purpose moving through change.", "audience_job": "Personal & industrial autonomy is Manna's continuing work for individuals and plant teams, so that work moves inside limits they set and waits for them where the mandate ends.", "chapters": [ { "id": "two-operating-scenes", "kind": "comparison", "title": "A personal project and an industrial process need different mandates.", "intro": "Continuing work becomes useful when its authority fits the setting.", "items": [ { "title": "Personal", "text": "Carry an objective across sessions, observe relevant sources and prepare the next useful step for a person." }, { "title": "Industrial", "text": "Coordinate a process under specified equipment, resource, intervention and recovery responsibilities." } ] }, { "id": "continuation-design", "kind": "process", "title": "Make interruption part of the design.", "intro": "A bounded autonomous workflow must retain enough state for a responsible continuation.", "items": [ { "title": "State the objective", "text": "Define success, resources and the permitted operating conditions." }, { "title": "Separate observation and effect", "text": "A changing signal can influence a proposal; action still follows the mandate." }, { "title": "Recover deliberately", "text": "Retain effect identity, outstanding obligations and the situations requiring human intervention." } ] } ], "questions": [ { "question": "What should the first pilot cover?", "answer": "One continuing task with known inputs, bounded actions, explicit stop conditions and a recovery exercise. Evaluate task outcomes and interventions before widening the mandate." } ], "relationships": [ { "type": "uses", "id": "manna", "label": "Continuing agent architecture", "text": "Purpose, regulation and recovery.", "url": "https://naevumlabs.com/products/manna" }, { "type": "uses", "id": "vagus", "label": "Operational memory", "text": "Context and attention over time.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "uses", "id": "synapse", "label": "Bounded effects", "text": "Control external computation and recovery.", "url": "https://naevumlabs.com/products/synapse" } ], "action": { "label": "Scope a persistent agent", "href": "https://naevumlabs.com/contact?context=persistent-agents" } } }, { "id": "solution-commercial-intelligence", "name": "Commercial intelligence", "url": "https://naevumlabs.com/solutions/commercial-intelligence", "description": "A person asks whether an offer fits their circumstances. The business needs an answer grounded in its current knowledge.", "systems": [ "https://naevumlabs.com/products/supersell", "https://naevumlabs.com/products/agency-engine" ], "editorial": { "lead": "Make the next conversation more useful.", "audience_job": "Commercial intelligence is Naevum's sales cluster for customer-facing teams, so that a customer gets a sourced answer at any hour and the business decides what it commits to.", "chapters": [ { "id": "invited-assistance", "kind": "narrative", "title": "Start with a need the person has expressed.", "intro": "A person asks whether an offer fits their circumstances. The business needs an answer grounded in its current knowledge.", "body": [ "Connect offers, supporting material and conversation history to an assistance journey. Let the person choose a next step and give the business a clear review path for consequential communication." ], "intro_role": "editorial-statement" }, { "id": "commercial-workflow", "kind": "process", "title": "Keep the relationship intact across the workflow.", "intro": "A relevant response depends on accurate context and appropriate permission.", "items": [ { "title": "Ground the offer", "text": "Retain the source and effective conditions of business assertions." }, { "title": "Prepare the response", "text": "Use the person’s question and permitted relationship context." }, { "title": "Respect the next decision", "text": "Apply recipient, consent and approval conditions before an external communication." } ] } ], "questions": [ { "question": "Which outcomes matter?", "answer": "Resolved needs, useful follow-up, accepted actions and retained consent. A pilot should examine both the person’s experience and the business’s ability to supervise it." } ], "relationships": [ { "type": "product-expression", "id": "supersell", "label": "Invitation-led experience", "text": "Connect a person’s needs to useful business assistance.", "url": "https://naevumlabs.com/products/supersell" }, { "type": "foundation", "id": "agency-engine", "label": "Product and workflow foundation", "text": "Context, services and organizational policy.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Discuss commercial assistance", "href": "https://naevumlabs.com/contact?context=commercial-intelligence" } } }, { "id": "solution-research-and-invention", "name": "Research & invention", "url": "https://naevumlabs.com/solutions/research-and-invention", "description": "A proposed mechanism looks promising. The next step is to define what it should change and what evidence could disprove it.", "systems": [ "https://naevumlabs.com/products/vault", "https://naevumlabs.com/products/black-box" ], "editorial": { "lead": "Turn a difficult question into reusable engineering.", "audience_job": "Research & invention is Naevum's test-first research service for R&D teams with a promising idea, so that they learn whether its effect holds before they fund building it.", "chapters": [ { "id": "research-brief", "kind": "narrative", "title": "Make the technical uncertainty explicit.", "intro": "A proposed mechanism looks promising. The next step is to define what it should change and what evidence could disprove it.", "body": [ "We connect source research, alternative explanations and experiments to a specific technical effect. The resulting artifact retains its assumptions and the conditions for adoption." ], "intro_role": "editorial-statement" }, { "id": "from-finding-to-use", "kind": "process", "title": "Keep the evidence through composition.", "intro": "A reusable result needs more than a successful demonstration.", "items": [ { "title": "Challenge the mechanism", "text": "Compare a defined baseline and investigate failure cases." }, { "title": "Expose the contract", "text": "Specify interfaces, environmental assumptions and unresolved obligations." }, { "title": "Evaluate adoption", "text": "Reproduce the effect in the intended system and retain the outcome, including negative findings." } ] } ], "questions": [ { "question": "Does every engagement aim at a patent?", "answer": "No. The output may be a component, method, feasibility decision or invention investigation. Patentability, filing and granted rights are separate questions." } ], "relationships": [ { "type": "uses", "id": "vault", "label": "Contracts and evidence", "text": "Reusable knowledge with assumptions attached.", "url": "https://naevumlabs.com/products/vault" }, { "type": "delivery", "id": "black-box", "label": "Applied engineering", "text": "Integrate and evaluate a selected result.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Discuss a research problem", "href": "https://naevumlabs.com/contact?context=research-and-invention" } } }, { "id": "solution-living-worlds", "name": "Living worlds", "url": "https://naevumlabs.com/solutions/living-worlds", "description": "A shared world becomes interesting when choices reach other people and leave consequences they can recognize.", "systems": [ "https://naevumlabs.com/products/everbloom", "https://naevumlabs.com/products/vault" ], "editorial": { "lead": "Build a world people can change together.", "audience_job": "Living worlds is Naevum's design work for studios building shared game worlds, so that one player's choice changes what others find and generated events stay inside the world's rules.", "chapters": [ { "id": "meaningful-roles", "kind": "narrative", "title": "Give a decision somewhere to go.", "intro": "A shared world becomes interesting when choices reach other people and leave consequences they can recognize.", "body": [ "Design roles, responsibilities and persistent state together. Generated events can add variation inside a world grammar; players need to understand what changed and how their participation mattered." ], "intro_role": "editorial-statement" }, { "id": "world-continuity", "kind": "mechanism", "title": "The world needs continuity beneath the story.", "intro": "Generation, identity and durable state have distinct jobs.", "items": [ { "title": "Validate the event", "text": "Check the campaign structure, permitted effects and economy constraints." }, { "title": "Commit the consequence", "text": "Keep server-owned identity and durable settlement coherent across retries." }, { "title": "Retell from evidence", "text": "A shared history should derive from the recorded decisions of that world." } ] } ], "questions": [ { "question": "How do we evaluate a world prototype?", "answer": "Observe whether participants understand their roles and consequences, whether state remains coherent and whether the intended interaction is compelling. Public retention claims need real player evidence." } ], "relationships": [ { "type": "product-expression", "id": "everbloom", "label": "Shared kingdom", "text": "A living-world expression of roles and persistent consequences.", "url": "https://naevumlabs.com/products/everbloom" }, { "type": "research", "id": "vault", "label": "Reusable mechanisms", "text": "Contracts and evidence for simulation systems.", "url": "https://naevumlabs.com/products/vault" } ], "action": { "label": "Discuss a living-world collaboration", "href": "https://naevumlabs.com/contact?context=living-worlds" } } }, { "id": "solution-data-infrastructure", "name": "Data & infrastructure", "url": "https://naevumlabs.com/solutions/data-infrastructure", "description": "Compute procurement, a new site and a private deployment begin with different constraints.", "systems": [ "https://naevumlabs.com/products/metahive", "https://naevumlabs.com/products/metahive/builder", "https://naevumlabs.com/products/synapse" ], "editorial": { "lead": "Bring the workload into the infrastructure plan.", "audience_job": "Data & infrastructure is Metahive's compute layer for organizations sizing AI workloads, so that they know where their work runs, who operates it and what energy it draws.", "chapters": [ { "id": "service-requirement", "kind": "comparison", "title": "Capacity is useful only in an operating context.", "intro": "Compute procurement, a new site and a private deployment begin with different constraints.", "items": [ { "title": "Compute buyer", "text": "Define workload, service units, location and operational responsibilities." }, { "title": "Site planner", "text": "Compare fit, power, cooling, network and physical capacity." }, { "title": "Institution", "text": "Connect custody, model access and operating control to a deployment design." } ] }, { "id": "coupled-feasibility", "kind": "process", "title": "Compare the site and the workload together.", "intro": "A larger equipment list can still fail the intended service.", "items": [ { "title": "Model the physical system", "text": "Follow equipment through the rack, room, building and plot." }, { "title": "Model the workload", "text": "Check inference memory, placement, concurrency and compute limits." }, { "title": "Examine operation", "text": "Compare energy and capacity scenarios with declared assumptions, then plan the measurements required at deployment." } ] } ], "questions": [ { "question": "Can we start with feasibility only?", "answer": "Yes. A bounded engagement can deliver a source-linked configuration, workload scenarios, critical constraints and the evidence needed for an investment or build decision." } ], "relationships": [ { "type": "deployment", "id": "metahive", "label": "Compute and infrastructure", "text": "Service, site and operating responsibilities.", "url": "https://naevumlabs.com/products/metahive" }, { "type": "uses", "id": "metahive-builder", "label": "Planning and scenarios", "text": "Physical, energy and inference constraints.", "url": "https://naevumlabs.com/products/metahive/builder" }, { "type": "uses", "id": "synapse", "label": "AI deployment boundaries", "text": "Context, model access and effects.", "url": "https://naevumlabs.com/products/synapse" } ], "action": { "label": "Discuss infrastructure requirements", "href": "https://naevumlabs.com/contact?context=private-ai" } } } ], "technical_pages": [ { "entity": "eu-ai-act-native", "name": "Built for the EU AI Act", "canonical_path": "/technology/eu-ai-act-native", "scene": "synapse", "technical_features": [ { "name": "Authority in the execution path", "description": "Source-bound admission, operating envelopes and human checkpoints make responsibility part of the system’s behavior." }, { "name": "Records connected to actual effects", "description": "An attempt’s inputs, permitted scope, decisions and resulting evidence remain associated through interruption and recovery." }, { "name": "Documentation from the system boundary", "description": "Risk, intended purpose, data boundaries and oversight are specified for a deployment. Technical architecture supports this assessment; it does not confer a universal conformity status." }, { "name": "Controlled change", "description": "Versioned source, acceptance evidence and bounded policy changes preserve the basis on which an operating system was assessed." } ], "applications": [ { "title": "Use outside intelligence on a bounded task", "audience": "Institutions and R&D teams", "situation": "A useful model needs context, while the organization holds a much larger sensitive source.", "mechanism": "Select a source-bound projection and account for its disclosure to the actual observer before dispatch.", "impact": "The organization can inspect the approved context, purpose and destination of the model work." }, { "title": "A process with a safe stopping point", "audience": "Industrial edition", "situation": "An operational process must react to observations while respecting equipment and responsibility boundaries.", "mechanism": "A pipeline separates perception, proposal and gated action; the journal records the resulting state.", "impact": "Operators can inspect why a step was admitted, stop the process and recover from its record." }, { "title": "Turn architectural direction into delivery", "audience": "Organizations commissioning custom AI", "situation": "A difficult brief spans data, software, interfaces and operating responsibilities.", "mechanism": "The Planner defines a mandate; implementation, review and QA preserve its acceptance conditions.", "impact": "The customer can examine how the delivered system answers the original brief." } ] }, { "entity": "invention-practice", "name": "Naevum’s invention practice", "canonical_path": "/lab/research", "scene": "vault", "technical_features": [ { "name": "Scientific evidence travels with the method", "description": "A mechanism retains its scientific references or the explicit search that found none. Failed searches, counterexamples and refutations remain useful research outcomes." }, { "name": "Components and methods are different registers", "description": "One register describes what a system does; another describes how work is performed. Typed identities, contracts and lineage connect candidates without confusing a workflow with executable capability." }, { "name": "Evidence has a promotion ladder", "description": "Assertion, available artifact, functional evidence, independent reproduction and replication carry different grades. A negative control tests that the proposed verification can actually fail." }, { "name": "Composition preserves the argument", "description": "Candidate discovery, compatibility checks and discriminating experiments connect a question to a reusable result. The reason a component was selected remains available to the next project." }, { "name": "Invention is a technical claim", "description": "Mechanism, technical effect, prior art and a deliberate disclosure decision form the invention path. Candidate claims, filed applications and granted rights remain distinct records." } ], "applications": [ { "title": "Find the mechanism before building again", "audience": "R&D and software teams", "situation": "A new problem may overlap earlier engineering in another project.", "mechanism": "Typed component and method registers retain contracts, scientific references and lineage.", "impact": "The team can inspect a candidate and the evidence behind its reuse." }, { "title": "Challenge an invention early", "audience": "Research organizations", "situation": "A plausible idea may reproduce known prior art or fail under a simple counterexample.", "mechanism": "Prior-art search, negative controls and evidence grades stay attached to the technical claim.", "impact": "Research effort can move toward a discriminating experiment before the idea becomes a product promise." }, { "title": "Carry knowledge between projects", "audience": "Institutions", "situation": "A useful engineering result should survive its original author and project.", "mechanism": "The selected method retains inputs, contract, limitations and its experimental record.", "impact": "A later team receives a technical argument it can inspect and reproduce." } ] } ], "stories": [ { "name": "From purpose to a continuing agent", "url": "https://naevumlabs.com/stories/purpose-to-agent", "description": "An illustrative journey through a continuing research task.", "editorial": { "lead": "A project waits overnight. Its purpose does not disappear.", "chapters": [ { "id": "situation", "kind": "process", "title": "A purpose with an unfinished obligation.", "intro": "An illustrative journey through a continuing research task.", "items": [ { "title": "The objective", "text": "A person asks for a comparison of two system designs. Useful completion means a reviewable recommendation grounded in selected sources, with unresolved questions visible." }, { "title": "The interruption", "text": "An observation step finishes, but the working session ends before the comparison is complete. The objective, source state and outstanding obligation remain in the journal." } ] }, { "id": "consequence", "kind": "process", "title": "Continue into the changed conditions.", "intro": "Available resources and deadlines have changed. A regulated process can adjust attention and scheduling within its envelope while preserving the due work.", "items": [ { "title": "The next condition", "text": "Available resources and deadlines have changed. A regulated process can adjust attention and scheduling within its envelope while preserving the due work." }, { "title": "The accepted result", "text": "The agent prepares the comparison and returns the decision to the person. The outcome is judged against the original objective, separately from whether the process ran successfully." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "manna", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/manna" }, { "type": "related", "id": "vagus", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "related", "id": "synapse", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/synapse" } ], "action": { "label": "Discuss a continuing agent", "href": "https://naevumlabs.com/contact?context=persistent-agents" } } }, { "name": "From private knowledge to institutional intelligence", "url": "https://naevumlabs.com/stories/private-institution", "description": "An illustrative institutional question that needs both history and controlled model context.", "editorial": { "lead": "The policy changed. Yesterday’s decision still needs an answer.", "chapters": [ { "id": "situation", "kind": "process", "title": "Recover the decision in its own time.", "intro": "An illustrative institutional question that needs both history and controlled model context.", "items": [ { "title": "Ask at the right time", "text": "An operations team asks why a prior commitment used an earlier policy. A temporal memory retrieves the claim valid at that point, with its evidence." }, { "title": "Preserve the disagreement", "text": "A newer document contradicts one part of the record. The answer retains the conflict and its source instead of merging both into an invented certainty." } ] }, { "id": "consequence", "kind": "process", "title": "Bring a bounded view into the workflow.", "intro": "Only the source-bound context required for the task is proposed for disclosure. Admission considers the destination and what it has previously received.", "items": [ { "title": "Choose the model context", "text": "Only the source-bound context required for the task is proposed for disclosure. Admission considers the destination and what it has previously received." }, { "title": "Return responsibility", "text": "The result carries its source context and remaining uncertainty to the responsible person. Any external action follows a separate approval and recovery path." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "vagus", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "related", "id": "synapse", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/synapse" }, { "type": "related", "id": "agency-engine", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Discuss institutional memory", "href": "https://naevumlabs.com/contact?context=institutional-memory" } } }, { "name": "From a weak signal to a preventive decision", "url": "https://naevumlabs.com/stories/signals-to-prevention", "description": "A synthetic asset scenario illustrating predictive operations; no avoided failure is claimed.", "editorial": { "lead": "A rising signal creates a decision before it creates an incident.", "chapters": [ { "id": "situation", "kind": "process", "title": "Interpret the changing condition.", "intro": "A synthetic asset scenario illustrating predictive operations; no avoided failure is claimed.", "items": [ { "title": "Read the history", "text": "A condition signal has increased. The operating history also shows a change in load, so the signal needs interpretation in context." }, { "title": "Ask what is known", "text": "The estimate retains its evidence basis, forecast horizon and uncertainty. Sparse observations may require a class prior rather than a fitted asset-specific model." } ] }, { "id": "consequence", "kind": "process", "title": "Choose a response, then examine the outcome.", "intro": "The operator compares continued operation, an inspection and an intervention. The most useful next step may be the one that resolves a decision-critical unknown.", "items": [ { "title": "Choose the next observation", "text": "The operator compares continued operation, an inspection and an intervention. The most useful next step may be the one that resolves a decision-critical unknown." }, { "title": "Evaluate later", "text": "The record separates forecast, proposed response and applied action. Later evaluation examines calibration, false alarms and the actual outcome against a declared comparison." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "extrasensory", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/extrasensory" }, { "type": "related", "id": "vagus", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/vagus" } ], "action": { "label": "Discuss predictive operations", "href": "https://naevumlabs.com/contact?context=predictive-operations" } } }, { "name": "From a hard problem to an accountable system", "url": "https://naevumlabs.com/stories/idea-to-system", "description": "An illustrative custom-development journey from a recurring operational problem to handover.", "editorial": { "lead": "A useful brief should still be visible in the finished system.", "chapters": [ { "id": "situation", "kind": "process", "title": "Define the useful change and prove the difficult part.", "intro": "An illustrative custom-development journey from a recurring operational problem to handover.", "items": [ { "title": "State what needs to change", "text": "A team repeatedly reconstructs information before approving work. The brief defines the users, sources, decision, baseline and conditions for a useful result." }, { "title": "Prove the difficult connection", "text": "A representative slice connects the relevant sources to one reviewable decision. It exposes integration and authority problems before the scope expands." } ] }, { "id": "consequence", "kind": "process", "title": "Deliver an artifact with its operating record.", "intro": "Implementation delivers an exact source revision. Separate review and QA examine the acceptance conditions and retain unresolved findings.", "items": [ { "title": "Build and challenge", "text": "Implementation delivers an exact source revision. Separate review and QA examine the acceptance conditions and retain unresolved findings." }, { "title": "Hand over the operation", "text": "The accepted artifact arrives with integration contracts, controls, recovery procedures and agreed operating responsibility. Later changes retain that acceptance basis." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "black-box", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/black-box" }, { "type": "related", "id": "agency-engine", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "related", "id": "vault", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/vault" }, { "type": "related", "id": "conductr", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/conductr" } ], "action": { "label": "Discuss your system brief", "href": "https://naevumlabs.com/contact?context=ai-products" } } }, { "name": "From a device to a useful compute site", "url": "https://naevumlabs.com/stories/compute-to-capacity", "description": "A proposed compute-site scenario showing the relationship between equipment, energy and inference.", "editorial": { "lead": "A rack is only useful if the whole site can support its workload.", "chapters": [ { "id": "situation", "kind": "process", "title": "Place the equipment. Follow its physical demand.", "intro": "A proposed compute-site scenario showing the relationship between equipment, energy and inference.", "items": [ { "title": "Place the equipment", "text": "A planner assembles devices into racks and checks fit, clearance and floor loading in the intended room." }, { "title": "Follow the power and heat", "text": "Component demand, PSU losses, redundancy and cooling contribute to facility requirements. The calculation keeps its distinct power bases." } ] }, { "id": "consequence", "kind": "process", "title": "Test the workload against the whole site.", "intro": "Model weights and KV cache must fit the proposed memory and placement. Batch, concurrency and compute bounds determine the scenario’s service limits.", "items": [ { "title": "Bring in the workload", "text": "Model weights and KV cache must fit the proposed memory and placement. Batch, concurrency and compute bounds determine the scenario’s service limits." }, { "title": "Compare the operating case", "text": "An hourly energy scenario adds grid limits, generation and storage assumptions. The planner compares alternatives with the same inputs and opens the methods behind the limiting factor." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "metahive-builder", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/metahive/builder" }, { "type": "related", "id": "metahive", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/metahive" }, { "type": "related", "id": "synapse", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/synapse" } ], "action": { "label": "Discuss a capacity scenario", "href": "https://naevumlabs.com/contact?context=private-ai" } } }, { "name": "From a changing crown to a living world", "url": "https://naevumlabs.com/stories/decision-to-world", "description": "An illustrative day in Everbloom’s shared-world design.", "editorial": { "lead": "The crowns decide. The kingdom carries the consequence.", "chapters": [ { "id": "situation", "kind": "process", "title": "A royal choice reaches the roles.", "intro": "An illustrative day in Everbloom’s shared-world design.", "items": [ { "title": "A royal choice", "text": "The King and Queen choose to prepare a festival. The campaign gives the decision a form that can be validated within the world’s rules." }, { "title": "Work reaches the roles", "text": "The decision cascades through responsibilities. Players take part in the work; wicker NPCs keep unclaimed roles present." } ] }, { "id": "consequence", "kind": "process", "title": "A shared consequence becomes history.", "intro": "Completed work becomes a durable consequence in the kingdom. The shared state distinguishes pending tasks from settled changes across servers.", "items": [ { "title": "The world changes", "text": "Completed work becomes a durable consequence in the kingdom. The shared state distinguishes pending tasks from settled changes across servers." }, { "title": "The Chronicle remembers", "text": "Recorded decisions become the basis of the day’s account. The next reign inherits both the state and the history of the kingdom it will govern." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "everbloom", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/everbloom" } ], "action": { "label": "Discuss a living world", "href": "https://naevumlabs.com/contact?context=living-worlds" } } }, { "name": "From a research question to a technical invention", "url": "https://naevumlabs.com/stories/research-to-invention", "description": "An illustrative engineering investigation into whether two components can be composed.", "editorial": { "lead": "A promising component arrives with a question attached.", "chapters": [ { "id": "situation", "kind": "process", "title": "Find the assumption the connection leaves open.", "intro": "An illustrative engineering investigation into whether two components can be composed.", "items": [ { "title": "Name the proposed effect", "text": "A team expects a memory component to improve a workflow’s use of historical decisions. It states the task, baseline and failure condition before testing." }, { "title": "Inspect the assumptions", "text": "Compatible ports still leave an unanswered requirement about update ordering. The composition carries that obligation into the assembled design." } ] }, { "id": "consequence", "kind": "process", "title": "Make the result useful to another system.", "intro": "A test introduces delayed and contradictory updates. The observed behavior determines whether the proposed combination preserves the intended decision history.", "items": [ { "title": "Run the discriminating experiment", "text": "A test introduces delayed and contradictory updates. The observed behavior determines whether the proposed combination preserves the intended decision history." }, { "title": "Make the result reusable", "text": "The artifact retains its contract, evidence and negative findings. A subsequent adoption tests whether the same effect survives in the receiving system." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "vault", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/vault" }, { "type": "related", "id": "black-box", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/black-box" }, { "type": "related", "id": "agency-engine", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Discuss a technical question", "href": "https://naevumlabs.com/contact?context=research-and-invention" } } }, { "name": "From business knowledge to a coherent conversation", "url": "https://naevumlabs.com/stories/business-to-conversation", "description": "An illustrative SuperSell assistance journey, describing the current experience direction.", "editorial": { "lead": "An invitation begins with something the person wants to know.", "chapters": [ { "id": "situation", "kind": "process", "title": "An expressed need meets the business context.", "intro": "An illustrative SuperSell assistance journey, describing the current experience direction.", "items": [ { "title": "The person asks", "text": "A visitor wants to understand whether an offer fits a particular need. They choose to bring the business into the conversation." }, { "title": "The business supplies context", "text": "Current offer information and supporting material ground the answer. Source and effective conditions remain available when the offer changes." } ] }, { "id": "consequence", "kind": "process", "title": "A useful response leaves the person in control.", "intro": "The assistant explains the relevant terms and prepares a next step. A consequential communication follows recipient, consent and approval conditions.", "items": [ { "title": "A useful response takes shape", "text": "The assistant explains the relevant terms and prepares a next step. A consequential communication follows recipient, consent and approval conditions." }, { "title": "The person decides", "text": "The person can continue, involve a human or stop. Evaluation asks whether the need was resolved and the next step was useful to both sides." } ] } ], "questions": [], "relationships": [ { "type": "related", "id": "supersell", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/supersell" }, { "type": "related", "id": "agency-engine", "label": "Explore the system", "text": "The product page explains its specific mechanisms and operating scope.", "url": "https://naevumlabs.com/products/agency-engine" } ], "action": { "label": "Explore commercial assistance", "href": "https://naevumlabs.com/contact?context=commercial-intelligence" } } } ], "engagements": [ { "id": "intelligence-environment", "title": "Dedicated intelligence environment", "problem": "Knowledge and action are scattered across private systems.", "deliverables": [ "Source and identity map", "Versioned context model with memory and monitoring", "Admitted workflows and operator controls", "Deployment and recovery design" ], "acceptance": "Named tasks completed correctly with current context, appropriate permissions and recoverable results.", "systems": [ "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/products/manna", "https://naevumlabs.com/products/agency-engine" ], "action": { "label": "Discuss your environment", "href": "https://naevumlabs.com/contact?context=intelligence-environment" } }, { "id": "institutional-memory", "title": "Institutional memory and decision systems", "problem": "Teams repeatedly reconstruct decisions and lose the reasons behind them.", "deliverables": [ "Source-connected knowledge model", "Temporal recall and contradiction handling", "Attention rules and workflow integration", "A fixed question set for evaluating useful answers" ], "acceptance": "Correct evidence recovered and useful decisions supported on the agreed task set.", "systems": [ "https://naevumlabs.com/products/vagus", "https://naevumlabs.com/products/agency-engine", "https://naevumlabs.com/products/vault" ], "action": { "label": "Discuss your institutional knowledge", "href": "https://naevumlabs.com/contact?context=institutional-memory" } }, { "id": "persistent-agents", "title": "Persistent agents and operational autonomy", "problem": "Work needs to continue across events, tools and interruptions.", "deliverables": [ "Objective and mandate model", "Supervised processes and resource envelopes", "Intervention and recovery paths", "Task-level operating evaluation" ], "acceptance": "Accepted completion, recovery and intervention rate under stated conditions.", "systems": [ "https://naevumlabs.com/products/manna", "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/products/vagus" ], "action": { "label": "Scope a continuing agent", "href": "https://naevumlabs.com/contact?context=persistent-agents" } }, { "id": "predictive-operations", "title": "Predictive operations and prevention", "problem": "Early signals are available but do not become timely decisions.", "deliverables": [ "History and condition model", "Uncertainty and decision scenarios", "Intervention workflow", "Prospective evaluation design" ], "acceptance": "Calibration, lead time, false alarms and adjudicated intervention outcomes.", "systems": [ "https://naevumlabs.com/products/extrasensory" ], "action": { "label": "Discuss a predictive operation", "href": "https://naevumlabs.com/contact?context=predictive-operations" } }, { "id": "ai-products", "title": "AI products and software platforms", "problem": "A business needs a complete product with its own domain workflows.", "deliverables": [ "Product architecture and interface", "Service composition and integrations", "Roles, approvals and operational controls", "Tested release, documentation and handover" ], "acceptance": "Accepted user journeys, upgrade and recovery behavior, and maintainable integrations.", "systems": [ "https://naevumlabs.com/products/agency-engine", "https://naevumlabs.com/products/black-box", "https://naevumlabs.com/products/conductr", "https://naevumlabs.com/products/vault" ], "action": { "label": "Discuss a product to build", "href": "https://naevumlabs.com/contact?context=ai-products" } }, { "id": "private-ai", "title": "Private AI and compute-system design", "problem": "AI use depends on custody, location, capacity, energy or inference constraints.", "deliverables": [ "Deployment and custody design", "Compute and energy scenarios", "Integration prototype", "Operating model and measured capacity plan" ], "acceptance": "Named boundary tests, workload measurements and independently checked physical assumptions.", "systems": [ "https://naevumlabs.com/products/synapse", "https://naevumlabs.com/products/metahive", "https://naevumlabs.com/products/metahive/builder" ], "action": { "label": "Discuss deployment and compute", "href": "https://naevumlabs.com/contact?context=private-ai" } } ], "delivery": [ { "id": "discover", "title": "Discover and define", "text": "Map the workflow, sources, people, constraints and current baseline. Decide what a useful result would change.", "outputs": [ "System brief", "Outcome definition", "Data and rights boundary", "First architecture" ] }, { "id": "prove", "title": "Prove the critical path", "text": "Build one representative slice through the intended integrations or an agreed test environment. Examine failure cases before expanding the scope.", "outputs": [ "Working vertical slice", "Evaluation results", "Failure cases", "Decisions for the full system" ] }, { "id": "build", "title": "Build and integrate", "text": "Implement the selected scope with review, acceptance and operational responsibilities attached to the exact release.", "outputs": [ "Reviewed software", "Integration contracts", "Operator controls", "Acceptance record" ] }, { "id": "operate", "title": "Operate and improve", "text": "Agree monitoring, support, recovery and change control. Evaluate improvements against actual tasks and retained baselines.", "outputs": [ "Operating handbook", "Recovery procedures", "Support responsibilities", "Review cadence" ] } ], "vision": { "lead": "Intelligence should expand what people can build together.", "chapters": [ { "id": "collective-capability", "kind": "narrative", "title": "Develop the organization as carefully as the technology.", "intro": "We are building a lab where specialized people and systems can work toward a shared purpose while retaining their independence.", "body": [ "Knowledge should travel. Responsibility should be clear. A useful system should make it easier for the next person to build on what came before.", "Our research extends to the organization itself: how a team remembers, notices what matters, makes decisions and hands over work. People determine purpose and authority; software can make their coordination more inspectable." ], "intro_role": "editorial-statement" }, { "id": "organisational-organs", "kind": "mechanism", "title": "Give each essential function a working structure.", "intro": "We call these functions organizational organs.", "items": [ { "title": "Sensing and memory", "text": "Notice changes and retain the evidence, decisions and lessons that make them intelligible." }, { "title": "Deliberation and responsibility", "text": "Create places to examine alternatives, disagree and assign a bounded mandate." }, { "title": "Delivery and review", "text": "Make work visible, judge the result and preserve a path to correction." }, { "title": "Learning and continuity", "text": "Turn useful experience into shared knowledge that can survive a handover." } ] }, { "id": "horizontal-coordination", "kind": "process", "title": "A team can act with a clear mandate.", "intro": "A bounded team receives a purpose, the authority to carry it out and a way to account for the result.", "items": [ { "title": "Agree the responsibility", "text": "Define the outcome, resources and decisions that stay with others." }, { "title": "Make the work inspectable", "text": "Retain decisions, evidence and unresolved questions as the work proceeds." }, { "title": "Return knowledge", "text": "Share the accepted result and useful failures so another team can continue from them." } ] }, { "id": "model-that-travels", "kind": "process", "title": "Help other structures take shape.", "intro": "We want to share adaptable models for horizontal organization and support creation and adoption programs.", "items": [ { "title": "Map the current organization", "text": "Understand responsibilities, knowledge flows and the decisions people repeatedly struggle to make." }, { "title": "Design a bounded pilot", "text": "Choose organs and mandates appropriate to the people, work and legal setting." }, { "title": "Evaluate and adapt", "text": "Observe the result, revise the model and document what another organization could reuse." } ] }, { "id": "contested-mandate", "kind": "narrative", "title": "A disagreement should change the mandate visibly.", "intro": "Proposed adoption-pilot example: a delivery team asks to expand a mandate because a shared dependency has changed.", "body": [ "The initiating team records the proposed scope, resources and consequence. The team responsible for the affected dependency can challenge the assumptions and suggest a bounded alternative.", "The person or group holding the agreed decision right resolves the scoped question. The revised mandate becomes effective through that recorded decision; the earlier version and unresolved objections remain inspectable.", "This is a worked program proposal. Each organization agrees its own decision rights, stewardship and conflict-resolution process in its actual legal and operating setting." ], "intro_role": "editorial-statement" } ], "questions": [ { "question": "Is there one Naevum governance model everyone must adopt?", "answer": "The mission is to develop and share adaptable models. An organization’s people, work and legal setting determine how a model should be designed and tested." }, { "question": "Who is this invitation for?", "answer": "People commissioning intelligent systems, researchers, builders and organizations interested in a practical program for horizontal coordination." } ], "relationships": [ { "type": "related", "id": "vagus", "label": "Institutional memory", "text": "Software supporting knowledge and attention.", "url": "https://naevumlabs.com/products/vagus" }, { "type": "related", "id": "agency-engine", "label": "Organization context", "text": "Domain work, responsibility and shared methods.", "url": "https://naevumlabs.com/products/agency-engine" }, { "type": "delivery", "id": "black-box", "label": "Practical delivery", "text": "Turn a defined purpose into inspectable work.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Build with Naevum", "href": "https://naevumlabs.com/contact?context=vision" }, "people": [ "Theo Georgiev", "Hris Dimitrov", "Stan Sarbinski", "Lyubo Boyhinov", "Yo Dimitrov", "Steph Angelov" ], "prior_team_experience": [ "BMW", "Deutsche Telekom", "Infineon", "Nexo", "ProSiebenSat.1", "HWA", "Lidl", "IBM", "Poker Stars", "Cisco", "FXCM", "IKEA", "Bayer04 Leverkusen", "Microsoft", "AOK", "Cashpoint Sportwetten", "MYX Robotics" ], "partners": [ "Delta Engineering", "WeAreJerry", "NeuralAI", "Fleetcore" ] }, "philosophy": { "lead": "Shared purpose. Distributed responsibility.", "chapters": [ { "id": "purpose-and-independence", "kind": "narrative", "title": "Coherence can grow from explicit relationships.", "intro": "A horizontal organization needs practical ways for people to act independently and still contribute to a shared purpose.", "body": [ "Responsibility should come with the context, resources and authority needed to carry it. A mandate should explain what is expected, what can be decided and when the work returns to others.", "Disagreement is useful when it can be recorded, examined and resolved at the right place. Shared memory preserves the reasons behind a decision, including the alternatives that were considered." ], "intro_role": "editorial-statement" }, { "id": "continuity-and-learning", "kind": "mechanism", "title": "Make handover a capability of the organization.", "intro": "People and systems change. A coherent organization retains enough knowledge for work to continue.", "items": [ { "title": "Remember the reason", "text": "Keep decisions with their evidence and effective conditions." }, { "title": "Review the consequence", "text": "Separate a completed activity from an outcome that served the purpose." }, { "title": "Share an adaptable method", "text": "Document what worked, what failed and which assumptions another organization would need to revisit." } ] } ], "questions": [ { "question": "How can an organization explore this model?", "answer": "Begin with a bounded coordination problem and a pilot mandate. Evaluate the result with the people doing the work, then adapt the structure before expanding it." } ], "relationships": [], "action": { "label": "Discuss organizational adoption", "href": "https://naevumlabs.com/contact?context=horizontal-organisation" } }, "research": { "lead": "A technical idea becomes useful when its effect survives scrutiny.", "chapters": [ { "id": "research-sequence", "kind": "process", "title": "Carry the question all the way into adoption.", "intro": "Naevum’s research practice connects mechanisms, experiments and reusable engineering.", "items": [ { "title": "Mechanism and hypothesis", "text": "Describe the interaction that should cause the proposed effect and the conditions under which it applies." }, { "title": "Experiment and contradiction", "text": "Choose a baseline and a test that could show the idea is wrong. Preserve negative findings." }, { "title": "Artifact and contract", "text": "Package the result with interfaces, environmental assumptions, evidence and unresolved obligations." }, { "title": "Adoption and later utility", "text": "Examine the component in a receiving system. Reuse is valuable when it helps the next task under its actual conditions." } ] }, { "id": "cleared-directions", "kind": "mechanism", "title": "Research questions across the lab.", "intro": "The lab explores mechanisms whose effects can be examined in a concrete operating setting.", "items": [ { "title": "Authority-preserving memory", "text": "How can temporal recall retain evidence and uncertainty without promoting observations into instructions?" }, { "title": "Observer-aware model work", "text": "How can a disclosure decision account for what an observer already knows across several effects?" }, { "title": "Evidence-conditioned prediction", "text": "How should sparse histories, priors and uncertainty inform a useful intervention?" }, { "title": "Provenance-carrying simulation", "text": "How can a result retain the method, units and input lineage needed to challenge it?" } ] }, { "id": "protection-and-contribution", "kind": "constraints", "title": "Choose the right form for the result.", "intro": "Research output can become a component, method, publication, invention investigation or a decision to stop.", "body": [ "A documented idea, a tested technical effect, a filed application and a granted patent are different milestones. Rights, disclosure and publication are handled for the particular work." ] }, { "id": "learning-from-operating-experience", "kind": "mechanism", "title": "Learning needs a record of what changed.", "intro": "Proprietary operating knowledge can support several different improvement paths. The first research decision is which layer needs to change and how its contribution will be tested.", "body": [ "A source-connected memory can change what an existing model receives. Calibration and selection can change a system’s estimates or choices. Updating model weights or an adapter is a separate intervention with its own data and model-access requirements.", "Research such as [SEAL](https://arxiv.org/abs/2506.10943) and [Language Models Need Sleep](https://arxiv.org/abs/2606.03979v2) studies model adaptation and consolidation. [FunSearch](https://deepmind.google/blog/funsearch-making-new-discoveries-in-mathematical-sciences-using-large-language-models/) pairs generated candidates with an evaluator. Each engagement selects its update mechanism, evaluation task and operating boundary before work begins." ], "items": [ { "title": "Preserve the original experience", "text": "Keep observations, expert decisions, generated descendants and later outcomes distinguishable. Source access and permission to create training derivatives are separate questions." }, { "title": "Protect the comparison", "text": "Compare a fixed baseline, evidence retrieval and any proposed adaptation using held-out work. Generated variants retain ancestry so they cannot silently cross the evaluation boundary." }, { "title": "Measure useful change and retained capability", "text": "Score future task utility, earlier accepted behavior, calibration where relevant, data exposure and full cost across repeated updates." }, { "title": "Promote a versioned candidate", "text": "Keep the model, data, method and evaluator identities with the result. Accept or retain the incumbent through an explicit decision, with a tested recovery route." } ] } ], "questions": [ { "question": "Can we commission a feasibility investigation?", "answer": "Yes. A bounded investigation can establish what is technically plausible, identify the decisive unknowns and deliver an experiment or artifact that supports the next decision." } ], "relationships": [ { "type": "uses", "id": "vault", "label": "Research knowledge", "text": "Methods, components, contracts and evidence.", "url": "https://naevumlabs.com/products/vault" }, { "type": "delivery", "id": "black-box", "label": "Engineering adoption", "text": "Bring an examined result into a useful system.", "url": "https://naevumlabs.com/products/black-box" } ], "action": { "label": "Discuss a research collaboration", "href": "https://naevumlabs.com/contact?context=research-and-invention" } } }