Orchestrate Agents with Microsoft Foundry Workflows

Agent orchestration is challenging. Goals, tools, and context must remain aligned while multiple agents hand off tasks without losing state or quality. Fragmented observability, uneven tool access, and model drift can quickly turn even simple automations into brittle chains.

Microsoft Foundry addresses these challenges by unifying the control plane. It provides shared state management, standardized tool access, and robust routing logic to transform fragile chains into resilient systems. The workflow canvas allows you to coordinate decisions, retries, and fallbacks visually, ensuring agents reliably work together.

In this lab, you will create a Microsoft Foundry workflow that requires multiple AI agents to collaborate in handling customer support tickets. To further strengthen the workflow results, you will integrate with the Model Context Protocol (MCP). MCP exposes external systems in a structured, discoverable way. Agents can access real-world tools via MCP for greater success rates.

Learning Objectives

Upon completion of this intermediate-level lab, you will be able to:

• Coordinate multiple agents within Microsoft Foundry workflows
• Delegate tasks effectively among different agents
• Leverage Agent tools, such as model context protocol (MCP) servers, to enhance workflow capabilities

Intended Audience

• AI Engineers
• Data Scientists
• Machine Learning Engineers
• IT Professionals

Prerequisites

Familiarity with the following will be beneficial for this lab:

• AI Agents in Microsoft Foundry

The following content can be used to fulfill the prerequisite:

• Observability & Monitoring for Azure AI Agents Using Application Insights and Log Analytics
• Enhancing Agents with Azure AI Services