AI-native Healthcare Work OS

• What it is: "one product, one repo" — MISS is positioned as US healthtech platform's sole product, serving different client types as Organizations with different RBAC + config, not as separate products: clinical (De Factum — LIS+CIS+CRM+charts+BI), accreditation (IAC/CBAHI/SASE + Cortex), the patient app, and US healthtech platform itself (the first tenant, dogfooding). All other team repositories (healthcare-crm, regain-accreditation, regain-brain/Hermes/Deutsch/Popper/Bench, command-center) are in "absorption" mode → becoming MISS modules.
• Type / status / role: web-app (a huge multi-tenant monorepo) / active / contributor (major — 2nd author). Lead/founder — Anton Kim (US healthtech platform) ~4378 commits; the user (Davron+Dave93) — ~760 commits (~17%, 2nd place); plus E2E bots, Codex bot, other contributors.
• Activity window: 2026-02-09 → 2026-05-26 — 4556 commits over ~3.5 months (extreme AI-driven velocity).

• Monorepo: Turborepo + Bun; apps/ — web, admin, api (Elysia.js), queue (background/OCR), sites. Next.js 16 + React, Better Auth + deep RBAC (personas: registered nurse, physician, lab_director, admin_director_finance, transfusion nurse…), Drizzle/PostgreSQL, Eden Treaty (type-safe API), next-intl (ru/en/uz), shadcn / glassmorphism UI.
• AI ("Cortex"): a family of AI tools (cortex tool for patient-health-book and others), an OCR pipeline, IAC extraction from a PDF corpus.
• Infra/deploy: GCP (Cloud Build — cloudbuild.yaml 19 KB, cloudbuild-dev.yaml, .gcloudignore), Docker, Redis, infra/.
• AI tooling (many): .claude/ (243 sessions!), .codex/, .cursor/, .opencode/, .agents/, .deep-research/, .harness/, .mcp.json, .playwright-cli — multi-agent development.

Huge volume; PR-driven (numbers #14xx). Clinical/business modules visible:

• RCM (Revenue Cycle Management): claims, coding, denials — tab-render fixes, field-mapping, seed fixtures (#1453, #1452).
• RBAC/auth: closing persona-demo gaps, access grants (rcm→admin_director_finance, checkups→lab_director+physician) (#1451, #1450, #1447).
• Checkups: UX parity, porting 5 surfaces from healthcare-crm (#1448).
• Patient Health Book: cortex tool + admin page (#1436), sidebar + queue OCR dep (#1446).
• IAC accreditation: "ASIC Composer 2.5 evidence slice", workplan control plane, real-world PDF corpus gate for extraction (#1449, #1444, #1442).
• User contribution: ~760 commits (a separate git log --author breakdown is needed for precise modules; by volume — one of the core team members).

(Platform-wide; authorship is distributed — the user is a major contributor, not the sole one.)

• "One product = different tenants via RBAC + config" — an architectural invariant: instead of parallel codebases, clients are distinguished by roles/config/Cortex routing. → Product-architecture thinking at the platform level.
• Deep clinical RBAC — medication orders (CPOE: administer-medication), transfusion monitoring, witness controlled-substance waste — roles with fine-grained permissions for clinical workflows. → Domain-complex authorization (healthcare, regulatory).
• AI pipelines (Cortex/OCR/IAC): extraction from real PDF corpora with "gate" checks, an OCR queue, per-patient AI tools. → Production ML/AI integrations.
• Multi-tenancy + GCP deploy (Cloud Build, dev/prod), type-safe monorepo (Eden Treaty), extreme PR velocity with E2E bots.

• Sessions found: 243 .jsonl in the local Claude Code sessions directory for this project — a portfolio record, an order of magnitude more than the rest. Plus parallel tools: Codex (.codex, Codex bot in git), Cursor, opencode, .agents, .deep-research, .harness.
• What was done with AI: essentially all development is AI-native; 4556 commits over 3.5 months are only possible with heavy AI assist. CLAUDE.md (35 KB) + AGENTS.md (35 KB) — detailed strategic instructions for agents.
• AI workflow patterns (top-tier for the brand): multi-agent (Claude Code + Codex + Cursor), PR-driven with E2E bots (US healthtech platform E2E), strategic "invariants" in CLAUDE.md, Cortex as the product AI layer, .deep-research/.harness for research and verification lanes.

• 4556 commits / ~3.5 months; user ~760 (2nd author).
• A healthcare platform: LIS/CIS/CRM/RCM/accreditation/patient + AI (Cortex/OCR).
• 5 apps (web/admin/api/queue/sites), GCP deploy, deep RBAC.
• 243 CC sessions + multi-agent tooling — the portfolio's strongest evidence of AI-driven development at scale.