Pathora

Pathora is an inclusive AI tour planner that designs personalized city experiences — tailored by mood, time, and physical needs.

Comment

Inspiration

Travel planning can be overwhelming — especially when time is limited and your energy or mood changes day to day. We dreamed of a tool that understands your vibe and your constraints, and instantly maps the perfect route for you. That’s how Pathora was born — an AI-powered tour planner that personalizes short urban explorations, in seconds.

What it does

Pathora generates customized walking tours based on:

  • Your current mood, energy level, interests
  • Your available time and city
  • Filters like cost, accessibility, or type of experience It returns a markdown itinerary with real Google Maps places, estimated visit durations, transitions between stops, and direct links. You can save and revisit your favorite tours on the web platform. 🎯 Future users could say: “I’m in Kyoto, I love indie bookstores and only have 2 hours” — and Pathora will do the rest.

How I built it

  • Frontend: HTML/CSS + Vanilla JS, hosted on Firebase
  • Backend: FastAPI (Python), hosted on Render
  • AI: OpenAI’s GPT-3.5-turbo for itinerary generation
  • Maps: Google Maps Platform (Places API, geolocation, search)
  • Storage: Firebase Firestore for saved tours The backend fetches real data from Google Maps, structures it, and prompts OpenAI to write a human-like tour with times and links.

Challenges I ran into

Prompt engineering was key. I iterated to:

  • Ensure OpenAI used only real Google Places data
  • Avoid invented locations or unrealistic time blocks
  • Balance personalization with tour structure (timing, cost, accessibility) I also debugged CORS issues, cross-platform deployment, and formatting the output for markdown + front-end compatibility.

Accomplishments that I'm proud of

  • Built and deployed the full AI-driven stack in under 24 hours
  • Delivered a functional, emotionally-aware travel planner with real-time data
  • Created an architecture ready for scale, and UX-ready for mainstream use

What I learned

  • Integrating Google Maps data into natural language tasks is a powerful pattern
  • Accessibility, personalization, and real-world grounding can coexist
  • With the right prompt design, AI can assist in emotionally meaningful tasks like trip planning

What's next for Pathora

📍Embed as a plugin or layer inside Google Maps: an assistant that lives inside the map UI ♿ Add accessibility modes, such as: - Wheelchair-friendly routing - Sign-language integrated UI (for guided instructions or museum-style tour reading) 🧳 Expand tour modes: slow travel, museum-only paths, café hopping 🌐 Add support for more languages and international cities

Share this project:

Updates