Instead of just finding the fastest route, we built an app that learns whether you prefer scenic trams over crowded metros, quiet routes when hungover, or stations with good coffee. It mines Reddit and social media for real commuter ininsights, like which buses run hot in winter or which metro car gets you closest to the exit.

Tech

• Frontend: React Native with Expo, OTA updates for rapid iteration
• Backend: Python FastAPI, PostgreSQL, HSL Digitransit GraphQL integration
• AI-agents: OpenAI agents that understand Helsinki and translate mood into route preferences
• Architecture: Multi-repo setup with automated CI/CD through GitHub Actions

The interesting bit

We discovered commuters have incredible tribal knowledge that never makes it into official APIs. Our AI agents analyze social posts to find these insights and factor them into route suggestions. Version 1.0 shipped to real users in Helsinki who actually prefer our "5 minutes slower but you'll get a seat" routes over traditional fastest-path algorithms.

University project at University of Helsinki, 2025. Team of 5.