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.
Retain the question, relevant prior work and explicit negative search results.
Proposed reuse and intended environment.
Retain the question, relevant prior work and explicit negative search results.
Claim and intended environment.
References, counterexamples and failed searches.
Negative searches and counterexamples remain part of the research record.
Discover typed candidates while distinguishing capability from working method.
Source-connected component/method candidates.
Discover typed candidates while distinguishing capability from working method.
Mechanism identity, ports and contracts.
Workflow identity, evidence and lineage.
A capability entry and a working method are different kinds of evidence.
Grade the specific claim and test whether the proposed verification can fail.
Specific claim, artifacts and existing evidence.
Actual harness/canary or reproduction result.
Actual result, including failure/refutation.
Grade the specific claim and test whether the proposed verification can fail.
Assertion/artifact/function/reproduction/replication remain distinct.
Refutations remain useful retained records.
A proposed verification must be able to expose a failure.
Compare typed interfaces, QoS and environmental assumptions.
Producer guarantees and consumer assumptions.
Evidence grade and refutations, not automatic compatibility.
Explicit age/absence behavior and remaining environment obligation.
Compare typed interfaces, QoS and environmental assumptions.
Type, cadence, absence and ordering.
Undischarged assumptions remain attached to the assembly.
Type compatibility alone does not discharge timing, ordering or environment assumptions.
Test the gap in the intended setting and retain the actual result for reuse.
A concrete gap to test in the target.
Independent behavior under the intended delayed/absent-input case.
Test the gap in the intended setting and retain the actual result for reuse.
Delayed/absent inputs and intended conditions.
Actual outcome and identified downstream use where available.
A target-setting evaluation is evidence for that setting; it is not general proof.
Proposed reuse and intended environment.
Retain the question, relevant prior work and explicit negative search results.
Claim and intended environment.
References, counterexamples and failed searches.
Negative searches and counterexamples remain part of the research record.
Source-connected component/method candidates.
Discover typed candidates while distinguishing capability from working method.
Mechanism identity, ports and contracts.
Workflow identity, evidence and lineage.
A capability entry and a working method are different kinds of evidence.
Specific claim, artifacts and existing evidence.
Actual harness/canary or reproduction result.
Actual result, including failure/refutation.
Grade the specific claim and test whether the proposed verification can fail.
Assertion/artifact/function/reproduction/replication remain distinct.
Refutations remain useful retained records.
A proposed verification must be able to expose a failure.
Producer guarantees and consumer assumptions.
Evidence grade and refutations, not automatic compatibility.
Explicit age/absence behavior and remaining environment obligation.
Compare typed interfaces, QoS and environmental assumptions.
Type, cadence, absence and ordering.
Undischarged assumptions remain attached to the assembly.
Type compatibility alone does not discharge timing, ordering or environment assumptions.
A concrete gap to test in the target.
Independent behavior under the intended delayed/absent-input case.
Test the gap in the intended setting and retain the actual result for reuse.
Delayed/absent inputs and intended conditions.
Actual outcome and identified downstream use where available.
A target-setting evaluation is evidence for that setting; it is not general proof.
An illustrative process for investigating claims, comparing contracts and retaining unresolved obligations.
Additional mechanisms are described in the technical reference.
What the research says
A preregistered study tested tools sold as hallucination-free and found the claim did not hold. A test built to break a claim costs far less before funding than after launch, so each engagement starts with the test that could disprove the idea and keeps its negative results in the Vault.
A proposed mechanism looks promising. The next step is to define what it should change and what evidence could disprove it.
We connect source research, alternative explanations and experiments to a specific technical effect. The resulting artifact retains its assumptions and the conditions for adoption.
A reusable result needs more than a successful demonstration.
Compare a defined baseline and investigate failure cases.
Specify interfaces, environmental assumptions and unresolved obligations.
Reproduce the effect in the intended system and retain the outcome, including negative findings.
A head of innovation decides whether an idea earns its next budget; each test result joins a record that later teams can check. Illustrative.
Systems behind it
Evidence-led research & invention
The Vault supplies existing mechanisms with their assumptions, and stores each result, failures included, for the next reuse.
AI systems & software development
Black Box Factory builds the test setup and the first working version, with a separate reviewer checking both.
No. The output may be a component, method, feasibility decision or invention investigation. Patentability, filing and granted rights are separate questions.
Challenge the mechanism. Compare a defined baseline and investigate failure cases.
Expose the contract. Specify interfaces, environmental assumptions and unresolved obligations.
Evaluate adoption. Reproduce the effect in the intended system and retain the outcome, including negative findings.
An illustrative engineering investigation into whether two components can be composed.
Follow the storyWhich idea are you about to fund that nobody has yet tried to disprove?
Discuss a research problem