Garmin Guru

Historic Volume Data visualization
AI Critique
Homepage
Run Location Heatmap
Basic Profile Info

Inspiration

We wanted to create a more accessible source of analytics and direction for runners in need of coaching, as other apps like Strava and Runna provide AI coaching and analytics, but only with a subscription.

What it does

It provides advanced analytical information about past running activities, analyzing performance and exertion data to provide AI recommendations about performance improvement and progression.

How we built it

Frontend: Streamlit LLM Orchestration: LangChain AI Models: Google Gemini (Generative AI) Vector Store: FAISS (for coaching document retrieval) Maps: Folium & Streamlit-Folium Data: GarminConnect Python API

Challenges we ran into

Learning how to use a new API wrapper to streamline analytics, and figuring out the most optimal way to display the resulting graphs and statistics.

Incorporating the correct tools for the RAG running coaching agent to be able to use.

What we learned

We learned how to integrate AI with a streamlit app, and how to use APIs. We also learned how to create new visualizations using libraries like folio.