Healthcare Regulatory · AI-Assisted Compliance

SAMD-OS.

Medical device teams can't ship because their regulatory artifacts are disconnected from their codebase. Design controls live in spreadsheets. Risk files reference code that's already changed. I built a regulatory operating system that generates IEC 62304 artifacts directly from the codebase and reviews them with domain-specialist agents.

01
Specialist Reviewers
5 domain-expert agents modeled on real practitioners. Each reviews against specific standards with structured verdicts: ACCEPTABLE, NEEDS REVISION, or NOT SUBMITTABLE. Regulatory, QA, safety, cybersecurity, and clinical neonatal SpO2.
FDAISO 14971IEC 62304
02
Artifact Generators
Design controls, risk management, FHIR bundles, change impact analysis, design review packages. Output as XLSX, JSON, and Markdown. Each artifact traces back to source code and requirements — no manual reconciliation.
IEC 62304ISO 14971FHIR R4
03
Gap Analysis
Reverse-engineer compliance from existing codebases. Three skills for teams that shipped before documenting: Code → SOUP Register, Code → Design Inputs, Code → Hazard Candidates. Produces audit-ready artifacts from code that already exists.
RetrospectiveIEC 62304 §5.3.3
04
Clinical Pathway Spec
Evidence-based clinical device specification for sarcopenia resistance training with FHIR CarePlan integration. Grounded in 14 RCTs and 34 meta-analysis studies from ICFSR 2018 and EWGSOP2 consensus.
ICFSR 2018EWGSOP2FHIR
05
Validation
20/20 fixture format pass across all 5 agents. Structured verdicts validated against known-good and known-bad inputs. Full eval suite cost: ~$2.00.
TestingEvaluation
Standards
IEC 62304 ISO 14971 ISO 13485 IEC 62366-1 IEC 81001-5-1 FDA Cyber 2025
Stack
Claude Code Claude API Python XLSX JSON Markdown
Workflow
/spec /plan /build /review /ship