Introduction
Unit 1 of 6 · Module 13: Develop an AI agent with Microsoft Agent Framework
Read this lesson
Two workshops become one
AI agents read information, make decisions, and finish tasks with little help from people, which makes them a good fit for chores with many steps. Building them well needs a toolkit that's easy to start with and sturdy enough for real companies. For years, Microsoft had two workshops for this. Semantic Kernel had the heavy-duty enterprise tools, and AutoGen had simple, friendly ways to build agents and teams of agents. Microsoft Agent Framework is the next generation of both, built by the same teams. It keeps AutoGen's simple agents and Semantic Kernel's enterprise features, like saved conversation state, type safety, middleware, and telemetry, and it adds graph-based workflows for controlling teams of agents.
The expense claim
Here's the kind of job you'll build. Someone drops a handful of receipts on your desk: a taxi, a dinner, a hotel. You want an agent that pulls out the details, tidies them into a proper claim, and emails it to the expenses team. Agent Framework makes that straightforward, because it gives the agent tools to act with and sessions to remember the conversation. In this module, you'll connect Agent Framework to a Microsoft Foundry project, create Foundry agents with its SDK, and give them tool functions of your own.
Facts from Microsoft's documentation: Microsoft Agent Framework Overview (CC BY 4.0), rewritten in Loop's own words and examples. Narration, animation, and quiz by Loop. Microsoft, Azure, and Microsoft Foundry are trademarks of Microsoft. Loop is independent and not affiliated with or endorsed by Microsoft.