TravelGenie Home - Where user can chat with TravelGenie to get trip itinerary
Chat Interface- User Interaction with TravelGenie
Once the trip is validated, TravelGenie Core intelligently coordinates with other agents to get trip itinerary
Trip Planning- UI (Weather, Route & Flight Details)
Popular places & Popular restaurants at destination city
List of upcoming events at destination city
GIF
Architecture

TravelGenie: AI-Powered Multi-Agent Travel Planner

Global AI Agents League

Fetch.ai Innovation Lab Hackathon 2025

Team Members:

Rohit Kosamkar
Sapna Chavan

✨ Inspiration

While using various AI chatbots to plan trips, we noticed most platforms provide long, static, text-heavy responses. These are often hard to follow and lack the real-time data that modern travelers need. We imagined a platform that could function like a smart travel buddy—engaging, responsive, and visually interactive. Thus, TravelGenie was born—to make trip planning fun, intuitive, and intelligent.

What it does

TravelGenie is an AI-powered multi-agent system that helps users plan their trips interactively. Here's what it offers:

Accepts natural language input from the user via chat (e.g., "Plan my trip from Boston to NYC for next weekend").
Extracts trip details and validates them.
Coordinates with registered agents to fetch:
- Route Info (via Google Maps API)
- Weather Forecast (via OpenWeatherMap API)
- Flight Options (via Amadeus API)
- Popular Attractions (via Google Places)
- Top Restaurants (via Google Maps)
- Local Events (via Ticketmaster)
Returns the complete itinerary on a beautifully styled dashboard with visual cards.
Uses Langchain + Gemini 2.0 Flash for LLM-powered reasoning and orchestration.

How we built it

Frontend: React with Tailwind CSS for responsive UI.
Backend: FastAPI for API management and agent coordination.
Agents: Built using uAgents framework, each agent is modular and live on Agentverse.
APIs Integrated: Amadeus, Google Maps, Google Places, Ticketmaster, OpenWeather.
Agentverse: All agents deployed on Agentverse for public use.
LLM Orchestration: Gemini 2.0 Flash + Langchain (ReAct) for parsing, planning, and intent classification.
Dashboard UI: A visual transition from chat to interactive cards for itinerary breakdown.

Architecture

TravelGenie AI-Agent Architecture

Web UI → User chat input
Supervisor Agent → Intent classifier, trip detail extractor, memory tracker
TravelGenie Core → Orchestrates API agents
Agents:
- Weather Agent: Fetches weather forecasts using OpenWeatherMap API.
- Route Agent: Computes route and transport suggestions between cities.
- Flight Agent: Searches top flight options using the Amadeus API.
- Restaurant Agent: Retrieves top-rated restaurants using Google Maps API.
- Places Agent: Recommends popular places via Google Places API.
- Event Agent: Lists upcoming events using the Ticketmaster API.
- Summary Agent: Summarizes the itinerary with useful tips and prep advice using Gemini.
Gemini 2.0 Flash → Intent classification & reasoning
Langchain → Task planning and orchestration
UI Output → Dynamic dashboard

Challenges we ran into

Agent Coordination: Syncing multiple agents with message sequencing and async logic.
Selenium Flight Scraper: Initially tried scraping Kayak using Selenium but response time was too slow. Pivoted to the Amadeus API for faster, reliable flight data.
Real-time Validation: Ensuring that trip inputs (source ≠ destination, dates are valid, duration ≤ 14 days).
Agent Registration: Handling model serialization and decoding issues during registration on Agentverse.

Accomplishments that we're proud of

Deployed 6+ intelligent agents publicly on Agentverse.
Designed a chat → dashboard experience that feels seamless and delightful.
Integrated multiple real-time data sources and handled them modularly via agents.
Built reusable agent codebases (Weather, Flights, Food, Events, etc.) that can be adapted for other ecosystems.

What we learned

Agent-oriented system design with uAgents.
Fine-tuned usage of Langchain ReAct prompting and Gemini 2.0 Flash.
Best practices in handling multiple APIs securely and efficiently.
How to deploy, monitor, and manage autonomous agents via Agentverse.

What's next for TravelGenie Ai-Agent

User Login & Itinerary Save: Personalize and revisit past trips.
Hotel Booking Agent: Integrate with lodging APIs.
Budget Estimation Agent: Track costs and suggest optimizations.
Voice Input Support: Make trip planning hands-free.
Co-Planning with Friends: Share and collaborate on itineraries.
Mobile App Version: Full native support for Android/iOS.

Conclusion

Building TravelGenie during this hackathon has been a transformative learning experience. From juggling agent communication and orchestrating real-time data to deploying a highly interactive frontend, we stretched our boundaries and embraced the power of agent-based design. More than a hackathon project, TravelGenie reflects our passion to simplify and elevate how people plan their journeys.