Azure-Samples/azure-ai-travel-agents

What it solves

This project addresses the complexity of managing specialized tasks in a travel agency setting. It demonstrates how to use multiple AI agents to handle distinct customer needs—such as understanding preferences, recommending destinations, and planning detailed itineraries—rather than relying on a single, general-purpose model.

How it works

The application uses a modular architecture where specialized agents are orchestrated by frameworks like LangChain.js, LlamaIndex.TS, or the Microsoft Agent Framework. These agents communicate with various tools via the Model Context Protocol (MCP). The MCP servers, which can be written in Python, Node.js, Java, or.NET, provide the necessary data and services to the agents. The entire system is containerized with Docker for easy deployment to Azure Container Apps.

Who it’s for

It is designed for developers and architects looking for a reference implementation of multi-agent orchestration using modern frameworks and the Model Context Protocol (MCP).

Highlights

• Orchestration support for LangChain.js, LlamaIndex.TS, and Microsoft Agent Framework
• Uses Model Context Protocol (MCP) to connect agents to diverse tools
• Modular design with MCP servers implemented in Python, Node.js, Java, and.NET
• Containerized with Docker for deployment to Azure Container Apps
• Includes OpenTelemetry integration for monitoring agent performance