Introduction
Unit 1 of 7 · Module 10: Build knowledge-enhanced AI agents with Foundry IQ
Read this lesson
Agents need your knowledge
Picture a hill town where every house has its own tap. Each tap is an AI agent your organization has put to work. The taps can do a lot, but they only pour well when they're connected to what your organization knows: its policies, procedures, product guides, support articles, and years of hard-won expertise. On its own, an agent can't reach that private knowledge at all. It also stops at its training cutoff, like a calendar torn off months ago, and it answers in a generic voice that knows nothing about your company. And with no facts to lean on, it may simply invent an answer that sounds right.
One shared knowledge layer
You could fix that one agent at a time, but it's real work: connect the data, choose how to split it into chunks, build a vector database, and lock down who can see what. Then the next team digs the very same well all over again. Foundry IQ is Microsoft's shared knowledge layer, the town's water tower. Instead of a separate retrieval augmented generation pipeline, or RAG, for each project, you build knowledge once and pipe it to every agent. Many agents draw from the same knowledge base, and when that knowledge base improves, every agent connected to it gets better at once. In this module, you'll see how RAG answers the knowledge problem, how Foundry IQ shares knowledge between agents, how to connect sources like Azure AI Search, Blob Storage, SharePoint, and OneLake, how instructions keep an agent's citations consistent, and how to test and watch retrieval. First stop: RAG.
Facts from Microsoft's documentation: What is Foundry IQ?; Retrieval augmented generation (RAG) and indexes in Microsoft Foundry (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.