Alan AI Hack

A health-aware food recommendation app that reads your real-time bio-signals and suggests nearby restaurants and dishes tailored to your current body state.

Team

The Problem

Most food apps recommend based on taste preferences or ratings — but what you should eat depends on how your body is doing right now. Low HRV after a poor night's sleep calls for different nutrition than a post-workout recovery window. There's no mainstream tool that bridges real-time biometric data with actionable, personalised food choices.

What It Does

Alan AI Hack connects to the Thryve health data platform to pull your live bio-signals (heart rate, HRV, sleep score, activity levels, stress index, etc.) and builds a structured bio-state context. That context is passed to Mistral AI along with your personal goals to generate conversational, health-aware food recommendations.

Users can:

• Chat with the AI assistant about what to eat based on how they feel
• Browse nearby restaurants (via Google Places) filtered by their current bio-state
• View epoch-level charts of HR, HRV, and other vitals over a 24-hour rolling window
• Set personal health goals that the AI factors into every recommendation

The Next.js frontend provides a clean chat interface with bio-state panels, sleep/activity cards, and a day picker for historical views. The FastAPI backend handles bio-context assembly, LLM calls, and restaurant discovery.

Tech Stack

• FastAPI + uvicorn — Python backend
• Mistral AI 🚀 — LLM for personalised recommendations
• Thryve 🚀 — Real-time health & bio-signal data platform
• Next.js + React — Web frontend
• Recharts — Bio-signal data visualisations
• Google Places API — Nearby restaurant discovery
• Limeat API — Dish nutritional analysis
• httpx — Async HTTP client

Special Track

Are you submitting to a special track? If so, which one?

• [ ] Alan Play

What We'd Do Next

• Deeper personalisation — factor in micronutrient deficits and long-term trend data, not just same-day signals
• Meal logging — close the feedback loop by tracking what was eaten and correlating it with subsequent bio-signal changes
• Push notifications — proactive nudges when bio-state enters a recovery or performance window
• Wearable onboarding — direct integrations with Garmin, Apple Health, and Whoop beyond Thryve
• Restaurant menus with dish-level scoring — rank every item on a menu by how well it fits the user's current bio-state

Built With

• python