Risk management
ISO 14971 · ISO/TR 24971Hazards, hazardous situations, risk controls, residual risk and benefit-risk, checked for completeness and internal consistency.
FDA reviewers now work with AI that reads an entire submission in minutes and cross-checks every document against every other. Anna runs that depth of analysis on your Design History File first, so gaps in risk, design controls, testing and validation surface while they are still cheap to fix.
Inconsistencies a human reviewer might skim past are now found systematically, across every document in the file.
A 510(k) Additional Information request puts the submission on hold, with up to 180 days to respond. The review clock stops while you do.
A risk control with no test. A harm scored two ways. A supplier change with no design change. Anna reads the connections.
Hazardous situation HS-22, battery overheating during charge, is reduced to "acceptable" by RC-22. No protocol or report in the file verifies RC-22, and the trace matrix row is blank.
RMF-003 Risk Analysis rev D, row 22 · TM-001 Trace Matrix, row 118
Link test report TR-118 (thermal cutoff, 60 °C) to RC-22 if it covers the control; otherwise add a verification protocol and re-evaluate residual risk and benefit-risk.
The same harm carries different severity ratings across two risk documents, which changes the risk priority of 6 failure modes.
RMF-003 rev D, harms list H-07 · dFMEA-002 rev B, rows 31–36
Reconcile both documents to the harms list in the risk management plan and re-score the affected failure modes.
DI-41 appears in the design input specification but is not referenced by any drawing, specification or test. DR-3 minutes list it as resolved without evidence.
DIS-001 rev F, §4.6 · DR-003 minutes, action 12
Identify the enclosure output that satisfies DI-41, add IEC 60529 test evidence, and amend the DR-3 action record.
The protocol cites 95% confidence / 95% reliability. With zero failures that requires n=59 under the success-run method; no alternative rationale is provided.
VP-092 Protocol §6.2 · TR-092 Report, Table 3
Add a statistical rationale justifying n=10 or extend the sample to 59 and report the result against the stated criterion.
CT-5 is identified as critical in the use-related risk analysis but is not among the tasks in the summative evaluation protocol.
URRA-001 rev C, CT-5 · HF-SUM-001 Protocol, Table 2
Add CT-5 to the summative study, or document why it is validated elsewhere, per IEC 62366-1 and FDA human factors guidance.
The software design specification lists v3.0.8. The SBOM in the cybersecurity section lists v3.0.13.
SDS-004 rev B, Appendix A · CYB-002 SBOM, line 41
Confirm the shipped version, align both documents, and update the vulnerability assessment for that version.
The new supplier uses a different resin grade. No design change, risk review or biocompatibility re-assessment is linked to the CAPA.
CAPA-2025-14 · Supplier file SUP-031 · BIO-001 ISO 10993 report
Open a design change, assess biocompatibility impact under ISO 10993-1, and update the risk file and DMR.
Illustrative findings from a synthetic file. Select a row to see evidence and mitigation.
Hazards, hazardous situations, risk controls, residual risk and benefit-risk, checked for completeness and internal consistency.
One harms list, scored the same way everywhere it appears.
Inputs, outputs, reviews, transfer and changes, with closures backed by evidence.
Acceptance criteria, sample-size rationale, test-to-requirement links and unresolved deviations.
Critical tasks, user groups and use environments validated against the risk file.
Safety class, SOUP, anomalies, threat model and SBOM consistency.
Process validation, supplier changes and CAPAs that should have reached the design file.
Every link in the chain, end to end, with broken and orphaned rows listed.
Add the file as it exists today: PDFs, Word, Excel trace matrices and exports from your eQMS. No reformatting or templates required.
Anna indexes every requirement, hazard, control, test and result, then checks each against the others and against the standards and guidance that apply to your device and pathway.
Each finding is ranked by severity and cited to the document, revision and row, so your team can confirm it in seconds.
Work through suggested mitigations, update the documents, and re-run the analysis until the report is clean enough to submit.
Each customer’s documents are processed in an isolated workspace, are never used to train models, and are deleted on request.
Join the early access cohort for upcoming submissions.
Request early access