WanderAI

A chat-based AI travel planning assistant that helps users create personalized travel itineraries using real-time data from TripAdvisor.

Try it to plan your next trip: shahdeep.me/WanderAI

Please hit star⭐ if you like this project.

Project Description

WanderAI is an intelligent travel planning web application that combines the power of Large Language Models (LLMs) with real-time TripAdvisor data to assist users in planning their trips. Through a conversational interface, users can:

• Search for attractions, hotels, and restaurants
• Get detailed information about specific locations
• Create travel itineraries
• Export their travel planning conversations

Tech Stack

Backend

• Spring Boot & Spring WebSocket
• Langchain4j & Langgraph4j for AI agent orchestration
• Google Gemini LLM
• TripAdvisor API integration
• STOMP protocol for WebSocket communication

For detailed backend setup and configuration, see the Backend README.

Frontend

• Next.js
• React
• TypeScript
• Tailwind CSS
• WebSocket client (STOMP)

For detailed frontend setup and configuration, see the Frontend README.

Capabilities

Real-time Data Access

• Location search across multiple categories (attractions, hotels, restaurants)
• Detailed information retrieval (ratings, reviews, contact info, etc.)
• Current operating hours and availability

Interactive Planning

• Natural language conversation for travel planning
• Dynamic itinerary creation
• Contextual recommendations

User Interface

• Clean, modern chat interface
• Real-time message updates
• Chat export functionality (TEXT/JSON formats)

Backend Agent Architecture

WanderAI implements an adaptive RAG (Retrieval-Augmented Generation) system with three specialized agents:

1. Supervisor Agent

• Main decision-making and routing control
• Evaluates user queries and delegates to specialized agents
• Manages conversation flow and response generation

2. Search Agent

• Specialized in location discovery
• Handles search queries across different categories
• Integrates with TripAdvisor search API

3. Details Agent

• Retrieves comprehensive location information
• Processes and formats detailed data for user consumption
• Integrates with TripAdvisor details API

Limitations

Current Limitations

• No persistent conversation storage
• No user authentication system
• API rate limits from TripAdvisor and LLM (Gemini)
• Potential LLM-related limitations:
  • Occasional inaccurate responses
  • Response generation errors
  • Model-specific constraints

Future Improvements

• Implement user authentication
• Add conversation persistence
• Expand API integration
• Enhance error handling and recovery
• Implement rate limiting safeguards