Inspiration
During my own health transformation, I realized consistency in hydration, sleep, food, and movement was critical but hard to sustain. I built a minimal AI journaling assistant to automate reflection, and it grew into Health Agent — a companion to help anyone track, analyze, and improve wellness rhythms.
What it does
Health Agent is an AI-powered wellness companion that integrates data, insights, and natural language interaction into a daily feedback loop. It: Connects to the Oura Ring API for sleep and activity data. Logs daily records and generates personalized insights with GPT. Embeds memories in a Chroma vector DB for recall. Syncs summaries to Google Sheets. Lets users query their health history through a GPT-4 LangChain agent. Offers a simple Streamlit UI hosted in Colab via ngrok.
How we built it
We designed the system as modular and agentic: Oura API → biometric ingestion. GPT → daily insights. Google Sheets → external sync. Chroma → persistent memory. LangChain + GPT-4 → natural language queries. Streamlit in Colab → frontend.
Challenges we ran into
Handling Chroma’s vector format limitations. Aligning embeddings with structured health logs. Avoiding agent query failures. Routing queries smoothly between Streamlit and backend.
Accomplishments that we’re proud of
Built an end-to-end demo: ingestion → memory → insights → UI. Created a scalable architecture with modular APIs. Developed a tool that improved my personal health rhythm and is generalizable for others.
What we learned
Embedding design strongly influences recall quality. User-facing memory requires balancing automation with transparency. Building a personal tool first accelerates iteration for broader use.
What’s next for Health Agent
Add more wearable and health API integrations. Build long-term trend detection and personalized recommendations. Support privacy-first, local storage. Open the framework for community health agent customization.
Log in or sign up for Devpost to join the conversation.