Travel Planner

Paste any travel video link → AI extracts places → confirm your spots → get a full optimized itinerary with a live route map. Travel planning in under a minute.

Comment

Inspiration

Video travel content like TikTok, Instagram Reels, and YouTube Shorts is how people discover places now... But there's no bridge between watching a video and actually going. That gap between inspiration and a real plan is where rips die. We wanted to close it in one tap.

What it does

Paste a Video URL, and our AI extracts every place mentioned from captions, transcripts, and hashtags. You confirm the spots you want, set your dates, and we generate a day-by-day itinerary (morning/afternoon/evening blocks) with an optimized driving route drawn live on a map using Google's Routes API.

How we built it

  • Frontend: Expo (React Native) with Expo Router, Apple Maps on iOS via expo-maps, Google Maps on Android/web via react-native-maps
  • Backend: FastAPI with async ingestion jobs - yt-dlp pulls video metadata, AI extracts, and geocodes places
  • AI: Two pipelines - place extraction with confidence scoring, and itinerary generation
  • Database: Supabase for auth, storage, and realtime

Challenges We Ran Into

Extracting place names accurately from geocoding was harder than expected. A mention of "that café in Lisbon" needs to be resolved to real coordinates. We also navigated a tricky dual-map library set up to preserve native Apple Maps UX while still supporting route optimization, and hit breaking API changes migrating from the deprecated Directions API to Google's new Routes API mid-hackathon.

Accomplishments We're Proud Of

Getting the full pipeline working end-to-end: a raw Video URL becomes a geocoded, optimized, day-by-day itinerary on a live map in under a minute. The route optimization, where the map redraws the optimal stop order in real time, feels like magic.

What We Learned

Confidence scoring for AI extraction matters more than we expected. Without it, every uncertain guess becomes user-visible noise. We also learned that shipping a clean async architecture early (job queue → polling) saved us constantly during the crunch when features changed.

What's Next For Travel Planner

  • Collaborative trips - share a trip link and let friends add their own saved places
  • Budget tracking and booking integrations
  • A more personalized itinerary-making process that suggests places similar to other places the user has been.
  • Share extension
  • Trending searches
  • Badges

Built With

Share this project:

Updates