Pontabi

Turn saved travel inspiration into a personal map and an actionable trip.

Comment

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Pontabi

Inspiration

People discover their next trip on Instagram, TikTok, blogs, screenshots, and messages, but those ideas remain trapped in disconnected apps. When it is time to travel, finding and organizing them becomes work all over again.

What it does

Pontabi turns travel inspiration into something usable. Share an Instagram or TikTok post through the native iOS Share Extension, paste a URL or note, or import a screenshot. Pontabi extracts candidate place names, resolves them against real place data, lets the traveler confirm the result, and saves the destination to a photo-backed personal map. Saved places can then become trips and day-by-day plans.

The key experience is continuity: the place and map pin appear immediately while durable photo storage and cloud synchronization continue safely in the background.

How we built it

Pontabi is an Expo and React Native iOS application with a native Share Extension. Supabase provides authentication, Postgres, Storage, and Edge Functions. Google Places resolves destinations, and server-side AI services assist with extraction and itinerary generation.

During OpenAI Build Week, I used Codex with GPT-5.6 to meaningfully extend and harden the existing product. We redesigned photo persistence as an idempotent, checkpointed background job: save the place first, keep the temporary image visible, persist each unfinished stage before execution, resume after relaunch, and replace the image only after durable storage succeeds.

Challenges

The hardest problem was making results durable without slowing the save experience. A database refresh with an empty durable-photo field could overwrite a valid temporary image, and app termination could happen before failed work reached a retry queue. The final design separates image download, Storage upload, and database update into resumable checkpoints, refreshes authentication on demand, and merges remote data without degrading better local state.

Accomplishments

  • Native share-to-map workflow
  • Multi-source place import
  • Immediate saved-place and map-pin feedback
  • Resumable background photo persistence
  • Relaunch, authentication, network, and partial-failure recovery
  • Bilingual first-run guidance
  • Verified native build and TestFlight distribution

What we learned

AI-assisted engineering is most useful when it helps reason across product behavior and failure states, not only when it generates code. The strongest Codex collaboration came from translating user-visible problems into explicit invariants and testing those invariants against real interruption scenarios.

What's next

Better itinerary constraints, collaborative trips, and recommendations that respect the places a traveler already chose to save.

Built With

Share this project:

Updates