HealthChain is a voice-based nutrition assistant built at Cal Hacks 11.0 and later selected for the Hume AI Startup Program. It turns meal conversations into structured nutrition records, tracks calorie and protein goals, and provides meal-history views through a Next.js frontend and Flask backend.
- A user discusses a meal through Hume's empathic voice interface.
- The backend uses Gemini to parse meals and nutrition targets into structured data.
- Flask and SQLite persist meal records, goals, and history.
- The Next.js interface displays progress and prior meals.
The repository also includes exploratory ChromaDB and Gemini work for healthier-meal recommendations. That prototype is separate from the primary meal-tracking flow.
- Next.js, React, TypeScript, and Tailwind CSS
- Hume EVI for voice interaction
- Flask, SQLAlchemy, and SQLite
- Gemini for meal and target parsing
Copy .env.example to .env and provide the required Hume and Gemini credentials.
cp .env.example .envcorepack enable
pnpm install
pnpm devpython3 -m venv .venv
source .venv/bin/activate
pip install -r backend/src/requirements.txt
cd backend/src
python main.pyHealthChain was created by Addy, Derek, Joshua, and Steven.
HealthChain is a hackathon prototype. The repository documents local setup and experimentation, not a production deployment or clinical nutrition service.