Trip Planner using Hands-Off Agents Pattern - OpenAI Agents SDK

🤖 Multi-Agent Trip Planning System

Our trip planning system leverages OpenAI's Agents SDK to create an intelligent, multi-agent workflow that handles everything from initial preferences to final booking. The system consists of several specialized agents working in sequence:

1. User Preferences Agent: The frontline agent that collects all necessary information about the trip
2. Destination Research Agent: Researches activities and attractions using web search
3. Itinerary Agent: Creates detailed day-by-day plans
4. Booking Agent: Handles accommodation and activity reservations
5. Summary Agent: Provides final confirmation and trip overview

🛠️ Key Components

Web Search Integration

The system utilizes OpenAI's WebSearchTool to gather real-time information about destinations, including:

• Local attractions and activities
• Weather forecasts
• Cultural events
• Restaurant recommendations
• Transportation options

Agent Flow

1. User Preferences Collection

   • Destination
   • Travel dates
   • Budget constraints
   • Travel style preferences
   • Group size

2. Destination Research

   • Web search for relevant information
   • Activity recommendations
   • Local insights

3. Itinerary Creation

   • Day-by-day planning
   • Time optimization
   • Activity sequencing

4. Booking Management

   • Accommodation options
   • Activity reservations
   • Transportation arrangements

5. Final Summary

   • Complete itinerary
   • Booking confirmations
   • Important reminders

[Insert screenshot of the agent flow diagram]

🎯 Key Features

Backend Capabilities

• Asynchronous agent communication
• Real-time web search integration
• Context-aware conversation handling
• Seamless agent handoffs

Frontend Experience

Our Gradio-based interface provides:

• Real-time chat interaction
• Clear conversation history
• Progress tracking
• Easy input handling

[Insert screenshot of the chat interface]

🚀 Getting Started

Check out our GitHub repository for:

• Complete setup instructions
• Code documentation
• Example conversations
• Contribution guidelines

📝 Demo

On launching the app, we provided input like "France trip plan". This activated the user preferences agent which returned follow-up questions to the user asking their preferences. Once the preferences were collected, the agent handsoff the control to another agent for researching the top activities to do in that destination.

📈 Future Enhancements

Possible future enhancements:

• Integration with booking APIs
• Enhanced natural language understanding
• More destination-specific features
• Advanced budget optimization
• Weather-aware planning

---

Interested in contributing? Feel free to submit a pull request or open an issue!