Just launched an open-source compliance plugin for Claude Cowork after seeing fintech teams pay £60K+ for platforms that orchestrate free public data.
UK fintech pilot (30 days, 5 analysts): • 95 minutes → 27 minutes per case • £85K annual platform cost → £240/year (Claude Pro) • Uses only free data: OFAC, UN, EU, Companies House, OpenSanctions
17 mandatory human-in-the-loop checkpoints. No auto-approvals. Deterministic risk scoring (MLR 2017 formulas). MIT licensed.
Launching today because Claude just announced Cowork plugin updates: https://www.linkedin.com/posts/claude-ai_were-introducing-up...
Testing if foundation models can replace compliance middleware for standard workflows (~70% of cases).
Demo slides: https://github.com/vyayasan/kyc-analyst/blob/main/docs/demo-... GitHub: https://github.com/vyayasan/kyc-analyst
Happy to answer questions about LLMs in regulated environments.
vyayasan•1h ago
Quick context on the pilot:
30-day test with 5 compliance analysts at UK fintech doing standard individual onboarding (Know Your Customer checks).
Before: 95 minutes per case (manual searches across OFAC, UN, EU sanctions lists, Companies House, adverse media, PEP databases)
After: 27 minutes per case (Claude orchestrates the searches, analyst reviews at 17 mandatory checkpoints)
Key architectural decision: NO auto-approvals. Every decision requires explicit analyst approval + notes.
Legal team spent 3 weeks reviewing before approving pilot. Main concern was audit trail - solved with immutable markdown logs.
Demo slides show the full workflow: https://github.com/vyayasan/kyc-analyst/blob/main/docs/demo-...
Happy to answer questions about: - The 17 stagegates - How risk scoring works (deterministic, not black box) - Regulatory requirements (FCA/MLR 2017) - What works vs what doesn't
Built this to test if foundation models can commoditize compliance middleware.