Inspiration
Group trip planning is annoying when everyone's locations are far apart. We wanted to automate picking a meetup area.
What it does
Takes multiple map points, finds the densest cluster, finds the center, and suggests nearby activities.
How we built it
Python + DBSCAN for clustering, FastAPI backend, Gemini for activity suggestions.
Challenges we ran into
Tuning DBSCAN radius, handling coordinate formats, and getting clean JSON back from Gemini.
Accomplishments that we're proud of
Fully automated cluster detection and a clean API response the frontend can use directly.
What we learned
Distance on Earth needs haversine, DBSCAN is sensitive to radius, and LLM outputs need strict prompting.
What's next for AI-gent
Live map UI, more cities, and better activity data sources (Google Places/Yelp).

Log in or sign up for Devpost to join the conversation.