.png&w=828&q=75)
This project implements a multi-agent system (MAS) that simulates a user‑facing healthcare concierge and specialist agents (medical strategist, nutritionist, performance scientist, physiotherapist, orchestrator). The demo includes a WhatsApp‑style chat UI, a health dashboard (wearables, timeline, daily plan), and an evidence pipeline for lab reports and artifacts. Agents are instrumented with tool calls (risk calculators, memory search, wearable fetch) and produce structured outputs with evidence_refs so the frontend can show provenance. The stack is designed for demoability: a Vite + React frontend with Tailwind, a Python backend (FastAPI style) with a MemoryStore (Postgres + Redis) and pluggable LLM provider. Use cases shown: “What’s my heart risk?”, daily plan generation, and upload/interpret lab reports — all with streaming assistant replies and tool‑call badges.
24 Aug 2025