Book Restaurant Agent

This project is a hands-on assignment to design and develop a Multi-Agent System (MAS) that provides restaurant search, recommendation, and reservation services. You will use the Google ADK(Agent Development Kit) to build agent services and complete the assignment.

System Architecture

The system consists of three main components:

• Client: Provides the user interface.
• Server: Acts as a bridge between the client app and the agent.
• Agent: Uses LLMs to handle restaurant search, recommendation, and reservation requests.

Architecture Diagram

graph LR
    A[Client] -- HTTP --> B[Server]
    B -- HTTP --> C[Agent]

Loading

Client

• The client app is developed with Flutter.
• To install packages and run the client app:

   flutter pub get
   flutter run -d windows

Server

• The server is built with Python using FastAPI.
• To install packages and run the server:

   pip install fastapi uvicorn
   uvicorn server:app --host 0.0.0.0 --port 5000

• The server runs on localhost:5000.

Agent

• The agent is developed using Google ADK.
• A simple agent is provided as a starting point. You are required to develop and improve the agent to complete the assignment.
• To install packages and the agent:

   pip install google-adk
   adk web

  or

   pip install google-adk
   adk api_server

• The agent runs on localhost:8000.

Assignment Goals

• Understand the architecture of a multi-agent system.
• Implement and extend the agent to handle user requests for restaurant search, recommendation, and reservation.
• Integrate the client, server, and agent components for a complete workflow.

Additional Resources

• Sample restaurant data is provided in the yelp/restaurants.json file.