TravelBuddy

• 

  Architecture

• 

• 

🚀 What it does

TravelBuddy allows users to:

• ✈️ Generate smart travel itineraries based on source, destination, and dates
  
• 🚗 Get a route summary that estimates travel time by road (if in-country) or flight (if international)
  
• 📍 Discover attractions and restaurants near destinations (via the Foursquare API)
  
• 🤖 All powered by LLMs like Gemini, integrated into a smooth React frontend + FastAPI backend
  

🛠 How we built it

• Frontend:

  • React + Vite for blazing-fast SPA development
    
  • Tailwind CSS for sleek, responsive styling
    

• Backend:

  • FastAPI (Python) with modular route/logic separation
    
  • Asynchronous handling and LLM prompt management
    

• AI Integration:

  • Google Gemini Pro/Flash for natural language content generation
    
  • Custom prompt engineering for itinerary and route summaries
    

• APIs Used:

  • 🗺 Foursquare Places API – for attractions & restaurants
    
  • 🧠 Google Gemini API – for AI-generated trip content
    

🧩 Challenges we ran into

• 🧾 LLM response parsing
  Gemini returns natural text, not JSON — formatting the response reliably was tricky

• 🛫 Multiple airports per city
  Estimating flights is tough when cities have several airports (e.g. New York, London)

• 🌦 Weather forecast limits
  Most APIs cap forecast data to 14 days — which limits trip planning beyond that window

• 🔄 Frontend-backend sync
  Async API calls + Gemini latency required careful loading state management in React

🏆 Accomplishments that we're proud of

• 🚀 Built a full-stack AI travel planner from scratch in under 36 hours
  
• 🧠 Integrated a production-grade LLM into a real-time planning tool
  
• 🧱 Designed a clean, scalable backend with reusable prompt modules
  
• 🎨 Built a beautiful, responsive frontend with interactive trip planning flow
  

📚 What we learned

• 📝 How to structure prompts for consistent LLM outputs
  
• 🕸 FastAPI + async best practices
  
• 🔁 Designing scalable frontend-backend data flows
  
• 🧩 Implementing fallback logic when APIs are rate-limited or fail
  

🔮 What's next for TravelBuddy

• 🗺 Show maps and directions between locations
  
• 📱 Package as a mobile app or Chrome extension
  
• ⛅ Integrate live weather and local events into itineraries
  
• 🧭 Let users save + share travel plans

Built With

• amazon-web-services
• fastapi
• fetch.ai
• gemini
• google-maps
• llm
• python
• react-native
• tailwind
• weatherapi