๐ TripTales AI Planner
๐ Project Name:
TripTales AI Planner
โ๏ธ Team Members:
Mohammad Iqbal (Solo Developer)
Role: Full-stack Developer, Prompt Engineer, Architect
Contact: [email protected]
๐ก Inspiration
The inspiration behind TripTales came from the frustration travelers face when planning trips: juggling hotel bookings, weather forecasts, visa checks, and creating a personalized itinerary โ all from multiple sources.
"What if an AI assistant could handle all of it in one go โ intelligently and visually?"
We wanted to harness Generative AI to create a complete travel experience planner that feels like having a personal travel concierge.
๐ Problem Statement
Travel planning is time-consuming and overwhelming, especially when travelers need to gather data from multiple sources โ weather, hotels, visa, routes, and personalized preferences.
Scope: Automate the end-to-end travel planning process with GenAI โ generating itineraries, images, hotel data, packing lists, visa info, and maps, all through a single app.
๐ What It Does
TripTales is a GenAI-powered travel planner that transforms a few simple inputs into a complete, personalized travel experience โ all in minutes.
๐ง Features:
๐ Collects User Preferences
- Number of travelers (adults, children)
- Destination, duration, budget
- Start date, interests (nature, adventure, culture, etc.)
- Image style (photo, sketch, painting, etc.)
- Minimum hotel rating and search radius
๐ฆ๏ธ Fetches Real-Time Weather
- Uses Open-Meteo API (daily/hourly forecasts)
- Dynamically adjusts itinerary suggestions
๐จ Suggests Hotels + Map Integration
- Based on radius, rating, and city
- Displays location on interactive map
๐ Generates Smart Itinerary
- Day-wise plan (Morning, Afternoon, Evening)
- Based on user interests and weather
๐ผ๏ธ Generates AI Images
- Places, food visuals, packing lists using Titan
๐ฑ Suggests Local Food
- AI-generated visuals + food list
๐งณ Provides a Packing List
- Weather-aware and destination-specific
๐ Checks Visa Requirements
- Based on nationality
๐ Exports Shareable PDF
- Combines everything into one file
๐ ๏ธ How We Built It
- Amazon Bedrock:
- Claude 3 Haiku โ Itinerary, visa, food, packing list generation
- Titan Image Generator โ Visuals for destinations & food
- Streamlit for UI
- MongoDB Atlas for user/session data
- Open-Meteo API for weather forecasts
- OpenRouteService + Google Maps API for maps & routes
๐ง Challenges We Faced
- Structuring consistent output from Claude
- Generating visuals with vague location info
- Maintaining visual quality across varied prompts
- Preventing latency and API bottlenecks
- Ensuring clean session-level isolation
๐ Accomplishments
- Fully functional GenAI-powered travel planner
- Smart itinerary generation using Claude + weather API
- AI-generated travel visuals with Titan
- Seamless integration of multiple AWS services
- PDF generation and real-time map integration
๐ What We Learned
- Multi-model orchestration on Amazon Bedrock
- Effective prompt engineering for structured output
- Streamlining AI + external API workflows
- Building a real-time GenAI production app
๐ฎ What's Next
- Add login and personalized user history
- Integrate real-time transport APIs (flights, trains)
- Support multi-city and international travel
- Enable voice-based interface
- Mobile + offline access
- Enhanced route and hotel filtering via Google Maps
๐ Tech Stack
| Technology | Purpose |
|---|---|
| Claude 3 Haiku | Itinerary, food, visa, packing |
| Titan Image | Image generation (places, food) |
| Python | Backend logic |
| Streamlit | Frontend UI |
| MongoDB Atlas | Data storage |
| Open-Meteo API | Weather integration |
| OpenRouteService | Route calculations |
| Google Maps API | Geocoding |
๐งฑ Architecture Overview
Mermaid architecture diagrams available in the README
Flow:
User Input โ Prompt + Caching โ Claude / Titan / Weather APIs โ Generated Output โ PDF + Streamlit UI
๐ฐ Cost Estimates
| Service | Estimated Cost |
|---|---|
| Amazon Bedrock | ~$0.25/user |
| EC2 (t3.medium) | ~$30/month |
| MongoDB Atlas | Free tier |
| Total | ~$30 for 1,000 users |
๐ฏ Impact & Limitations
โ Impact:
- 30% faster planning vs manual process
- Reduces manual work by over 50%
- Visual-first and AI-driven experience
โ ๏ธ Limitations:
- Image resolution capped (Titan limitation)
- No automatic itinerary re-sync for live changes
- Limited hotel data (15 API call workaround)
๐ Ethical & Security Considerations
- No personal data stored beyond session
- Claude prompts tested for bias/fairness
- WCAG-compliant UI
- Clearly labeled AI-generated content
๐ Project Links
- ๐ App: http://98.81.75.200:8501/
- ๐ฅ Demo Video: https://youtu.be/91z65H9pvIY
- ๐ป Code Repository: GitHub - TripTales
๐งช Try It Locally
# Clone the repository
git clone https://github.com/miqbal303/triptales-genai-aws-bedrock
cd triptales-genai-aws-bedrock
# Create .env file from GitHub secrets
cp secrets.env .env # or manually add values to .env
# Install dependencies
pip install -r requirements.txt
# Run the app
streamlit run app.py
Built With
- amazon-web-services
- bedrock
- nominatim
- python
- streamlit