Prototype implementation

The repository now contains a working prototype of the Beat Suite platform:

• backend/ – FastAPI + LangChain + Prisma (SQLite) service that ingests sensor/nurse data, runs an adaptive control engine, and exposes REST endpoints under /api. The service includes a background loop that continuously evaluates every registered patient and logs environment recommendations with LangChain narration plus optional Google Agent Developer Kit scoring.
• frontend/ – React + Vite + Tailwind + shadcn-inspired UI library that renders the storytelling site and streams live room telemetry from the backend. A useBeatSuiteState hook bootstraps a demo patient, streams synthetic smartwatch packets, and surfaces the FastAPI data (or safe fallback copy when offline). The public landing page focuses on the hero + patient room vignette, while the /command route hosts the full Nurse Command Center and Adaptive Control Center for logging notes or sending overrides straight to the FastAPI backend.

Running locally

Backend:

cd backend
python -m venv .venv && source .venv/bin/activate
pip install -e .
prisma generate && prisma migrate deploy  # optional: to use SQLite persistence
uvicorn app.main:app --reload

Frontend:

cd frontend
npm install
npm run dev

The Vite dev server proxies /api requests to the FastAPI service so the landing page automatically showcases the live data loop. Navigate to http://localhost:5173/command for the clinician console, or stay on / for the public-facing experience. Use .env variables (prefixed with BEAT_SUITE_) to point the backend at real Google Agent Developer Kit credentials or adjust control-loop intervals.