TraceFuse benchmark
Measured requirement-to-test traceability gains.
On LibEST, a dual TRIAD prior plus a small-model backward pass reached AP 48.60% and MAP 63.30% for requirement-to-test linking.
Read the TraceFuse benchmark48.60%
AP
63.30%
MAP
Find verification gapsbefore they becomeaudit problems.
Komainu AI connects your requirements, tests, code and evidence, so your V&V
team can see what is missing, what changed and what needs review before certification.
30 minutes with the founders. No preparation required.
Supported by
Startup Program
Startup campus
Google for Startups Cloud Program
Design partner program
Bring us the verification workflow that costs your team the most.
KomAInu is selecting a limited number of design partners building critical or regulated software. We want to test the product against real requirements, evidence, tools, and review constraints, and shape it around the workflows that matter in production.
A strong fit if
- Requirements, tests, and code are still reconciled manually.
- Coverage or compliance gaps surface late in reviews or audits.
- Confidentiality and integration constraints rule out a generic public SaaS.
Manual V&V does not scale with critical software.
Traditional verification bottlenecks leave coverage blind spots and burn review cycles before certification.
Spreadsheet traceability
Requirement-to-test links live in fragile sheets that drift as soon as code or the SRS changes.
Review fatigue
Teams inspect every artifact by hand instead of focusing on the findings that actually matter.
Gaps found too late
Missing coverage surfaces in audits and certification reviews, when it costs the most to fix.
Engineering evidence
Evidence across traceability and continuous review.
Two controlled engineering benchmarks examine the traceability engine and the requirement-aware reviewer on documented evaluation setups.
Copilot benchmark
Requirement-aware review against GitHub Copilot.
On the same deliberately difficult pull request, KomAInu found all 10 planted vulnerabilities while GitHub Copilot found 4 to 7.
Read the Copilot benchmark10 / 10
KomAInu
4–7 / 10
Copilot
Where will the next gap surface?
Four questions connect KomAInu's solutions to real verification work: trace every requirement with the AI Requirements Traceability Matrix, identify untested behavior with Test Coverage Gap Analysis, close validated gaps with AI Test Case Generation, and keep changes aligned with the AI Compliance Reviewer.
Product capabilities, in one console.
Map requirements and evidence, audit coverage gaps, remediate missing tests, and prevent regressions with requirement-aware pull request reviews in one connected workflow.
MAP
Traceability & Evidence Mapping
Build an AI requirements traceability matrix that maps SRS requirements to tests, code, documentation and audit evidence, so every link remains reviewable and bidirectional.
Verification workflow
Audit. Remediate.
Establish a connected verification baseline, identify the gaps that matter, then turn validated findings into concrete test and traceability updates.
One living traceability context: Requirements ↔ Tests ↔ Code ↔ Documents ↔ Decisions
01
Audit
Detect gaps and establish evidence
Analyze requirements, tests, code, documentation and requirements traceability matrices as one connected verification context. Coverage findings stay visible in the KomAInu console.
Traceability MatrixCoverage Gap Analysis
- Map requirements to tests, code and supporting evidence.
- Identify coverage and traceability gaps, then prioritize review.
- Track coverage findings live in the KomAInu application.
- Find errors earlier on complex projects.
02
Remediate
Correct validated gaps
Use any MCP-capable agent (Cursor, Claude, Codex, OpenCode) to turn reviewed coverage findings into test work, then sync results back into KomAInu after approval.
AI Test Case Generation
- Connect your AI agent through the KomAInu MCP server.
- Generate missing test cases, test code and scripts for review.
- Sync updated coverage results back into KomAInu after approval.
- Correct gaps faster with less risk.
Continuous prevention
Then keep it correct as the software evolves.
AI Compliance Reviewer is a dedicated application that carries the same requirements, standards, and evidence into every future pull request.
AI Compliance Reviewer · Continuous
Keep every code change aligned with your requirements.
KomAInu reviews every pull request in the context of your requirements and standards, so teams can catch compliance gaps and business or regulatory regressions before merge, propose corrective changes, and keep the audit trail current as the software evolves.
Every pull request
Ground each review in the same requirements, standards, evidence, and internal records used across verification and certification.
- Check each pull request against your requirements, standards, and policies.
- Link findings to the relevant source passage, code change, and review decision.
- Propose updates to traceability, risk-analysis, and certification records for validation and one-click sync.
Keep sensitive engineering data under your control.
Deployment is scoped around your project and its constraints. We define the model, infrastructure, data flow, and review gates with your team.
Model flexibility
Choose the model around the accuracy, governance, and infrastructure you need.
Private by design
Private deployment or deployment on your premises can fit your environment.
Human approval
Engineers decide what becomes part of the official project record.
Four solutions. One traceability foundation.
Start with the workflow that addresses your current bottleneck, then extend across the same connected requirements, tests, code, and evidence.
KomAInu solutions
Traceability foundation
AI Requirements Traceability Matrix
Connect requirements to tests and evidence in a bidirectional, audit-ready matrix.
Explore solutionCoverage analysis
Test Coverage Gap Analysis
Identify potentially uncovered requirement aspects and prioritize verification work.
Explore solutionTest remediation
AI Test Case Generation
Turn validated coverage gaps into candidate test cases and code, then update traceability under human review.
Explore solutionContinuous review
AI Compliance Reviewer
Review every pull request, propose corrective changes, and keep traceability and compliance records current.
Explore solutionBuilt by engineers who shipped critical software.
KomAInu isn’t AI built by outsiders guessing at how regulated software works. It’s built by engineers who have spent years developing critical software under the very standards our product serves, writing it, verifying it, and shipping it where a requirement gap is not an option.
Our team brings together two former Thales software engineers and an AI engineer from Arkhn, combining deep expertise in aerospace software, software verification and validation, and AI applied to highly regulated industries. We’ve worked to standards like DO-178C, ISO 27001, HDS and HL7 FHIR, so we understand your constraints, because they were ours.
STATION F
Building KomAInu from Paris.
Meet our founders
Frequently asked questions
What does KomAInu do?
KomAInu secures the consistency between requirements, tests, code, documentation, and the requirements traceability matrix across the software development lifecycle. The workflow covers three stages: audit, remediation, and prevention. First, KomAInu analyzes your requirements, tests, code, documentation, and RTM to detect coverage and traceability gaps, then justifies every finding with evidence. Second, it helps remediate the gaps by generating missing tests and scripts, suggesting corrections in your ALM, and updating traceability. Third, it helps prevent the same issues from coming back by checking every pull request, detecting business and regulatory regressions, and alerting before integration.
Is KomAInu commercially available?
Yes. KomAInu is currently available through tailored design partner engagements for selected teams building critical or regulated software. We start with a real verification bottleneck and determine together whether KomAInu fits the workflow, toolchain, deployment constraints, and review process. Pricing is scoped for each engagement rather than published as a fixed package.
What happens in the first conversation?
You speak directly with the founders for 30 minutes about one concrete V&V bottleneck, the tools and artifacts involved, and any data or deployment constraints. The goal is to decide whether KomAInu is relevant to your environment. No preparation is required, and if there is a fit, the next step and commercial scope are defined with you.
How is sensitive engineering data handled?
KomAInu works with different model strategies. During technical discovery, we scope where requirements, source code, tests, and compliance documents are processed, which models may be used, and how KomAInu connects to your existing tools. Private deployment or deployment on your premises can be designed around your infrastructure and governance constraints. Storage, access, retention, logging, and integrations are agreed for each engagement.
Which AI models does KomAInu work with?
KomAInu can work with private or open source models deployed in your own environment, including model families such as Mistral, Qwen, and similar models that you can host yourself. The choice depends on the accuracy, latency, cost, and security profile your project requires. The important point is that the model is not the product boundary. KomAInu provides the traceability workflow, retrieval context, scoring, evidence structure, and review process around the model, so the solution can be tuned to your needs instead of forcing every team into the same hosted AI setup.
Can KomAInu adapt to our workflow and standards?
Yes. KomAInu is designed to adapt to your project rather than replace your whole V&V process. We configure the solution around your standards, your requirements format, your test repositories, your ALM, and the way your team reviews and approves evidence. For example, one team may need DO-178C traceability from requirements to tests, another may need ISO 26262 coverage analysis, and another may need pull request compliance checks against internal rules. The underlying platform stays the same, but the workflow, connectors, review gates, and outputs are adapted to the certification or compliance problem you actually need to solve.
Does KomAInu integrate with ALM and existing tools?
KomAInu is built to sit on top of existing engineering and ALM workflows. It can read requirements, tests, code, documentation, and traceability artifacts, then suggest corrections, generate missing test material, and keep the traceability view current after human review. The goal is not to make teams rebuild their process from scratch. KomAInu connects the audit, remediation, and prevention loop around the tools already used by the project, so approved changes can flow back into the ALM or repository while the existing system of record remains in place.
Does KomAInu replace engineers?
No. KomAInu is built as an engineering assistant for regulated software, not as an automatic certification authority. It detects gaps, proposes tests, suggests corrections, updates traceability, and reviews pull requests, but engineers remain in control of what is accepted. That human review gate is important because safety-critical verification still depends on engineering judgement. KomAInu reduces the manual search and documentation effort, surfaces the risky inconsistencies earlier, and prepares evidence, while your team validates the output before it becomes part of the official project record.
Bring us the V&V bottleneck slowing your team down.
We will discuss your workflow, constraints and whether KomAInu is a practical fit.
Review our pricing approach