Gracie

Grace & Associates Meeting Intelligence System.

Internal platform for an 8-person federal healthcare IT consulting firm. Automates client meeting capture, AI-driven document generation, task tracking, and team alignment.

Status: planning. No production code yet. The repo currently contains specs and implementation plans only.

---

Where to start

Everything begins in docs/.

• AGENTS.md at the repo root — instructions for any AI assistant working on this project.
• docs/_context/STATE.md — live project state. Read this first if you're picking up cold.
• docs/_context/DECISIONS.md — every locked architectural decision with rationale.
• docs/_context/STACK.md — the resolved tech stack.
• docs/specs/v1-backend.md — canonical spec for backend, data, pipeline, integrations.
• docs/specs/figma-frontend.md — canonical spec for design system, UI, component anatomy.
• docs/plan/ — implementation plans, one per subsystem, executed in numeric order.

---

High-level shape (locked so far)

• Frontend: Next.js (App Router) + React 18 + TypeScript + Tailwind v4
• Backend: Fastify (separate container) + TypeScript
• Queue: BullMQ + Redis
• DB: Supabase Postgres + pgvector
• Storage: Cloudflare R2
• Auth: Logto + Microsoft SSO
• AI: Claude (claude-sonnet-4-5) for generation; embedding model TBD
• Meeting capture: Recall.ai
• Calendar: Microsoft Graph
• Email: Resend
• Automations: n8n on Coolify
• Deploy: Coolify on a VPS, Cloudflare Tunnel for ingress

Full stack with status flags is in docs/_context/STACK.md.