Moving an AI agent from a fragile prompt playground to a production-ready cloud deployment requires a solid architectural foundation. Microsoft Foundry changes the paradigm by introducing a managed, containerised runtime for custom agent code via Hosted Agents, alongside centralised runtime tool discovery with Foundry Toolboxes.
This 3-hour, hands-on workshop is designed for engineers, architects, and technical leaders who want to learn how to stand up, secure, and scale production-grade agent environments. Leveraging pre-built, community-vetted labs, we will step through local verification, configure centralized toolboxes via the Model Context Protocol (MCP), and ground our agents using Foundry IQ retrieval. Best of all, we will focus heavily on architecture, configuration, and control-plane deployment—meaning no deep, from-scratch coding is required.
Building enterprise agents means solving for security, managed identity, tool sprawl, and continuous quality evaluation. In this practical, lab-driven workshop, we will demystify the core components of the newly matured Microsoft Foundry ecosystem. By treating application code as a pre-built component, we focus our time entirely on the operational control plane: how agents securely communicate, how tools are discovered dynamically, and how agent reasoning loops are observed.
High-Level Workshop Outline:
Module 1: Foundations of Hosted Agents (1 Hour)
- Unpacking the architecture: Prompt-only agents vs. Hosted Agents running inside isolated secure container environments.
- Setting up the environment natively inside GitHub Codespaces.
- Running and validating a pre-built agent host locally to inspect the contract lifecycle before moving to the cloud.
Module 2: Decoupling Capabilities with Foundry Toolboxes & MCP (1 Hour)
- Managing tool sprawl: Moving away from hard-coded application tools to centralised, versioned Toolboxes.
- Configuring external integrations using Model Context Protocol (MCP) clients and pre-built connectors without changing application code.
- Setting up guardrails on tool inputs and outputs directly within the Foundry Control Plane.
Module 3: Observability, Continuous Evaluation, & Production Rollout (1 Hour)
- Implementing context-specific evaluation suites (auto-generated evaluators + test datasets) wired directly into developer workflows.
- Analyzing trace-linked data to pinpoint where an agent went off-track.
- Packaging our agent application and deploying it securely using the Azure Developer CLI (azd) workflow to the Foundry Agent Service.