--- title: Unifying Retrieval for AI Agents with Azure AI Search and Foundry IQ description: Managing multi-environment deployments in Microsoft Fabric just got dramatically easier --- [Spyglass MTG Blog ](https://www.spyglassmtg.com/blog) # [Unifying Retrieval for AI Agents with Azure AI Search and Foundry IQ](https://www.spyglassmtg.com/blog/unifying-retrieval-for-ai-agents-with-azure-ai-search-and-foundry-iq) Written by [Edgar Cuellar](https://www.spyglassmtg.com/blog/author/edgar-cuellar) | Feb 24, 2026 7:23:41 PM Azure AI Search and Foundry IQ work together to create a unified retrieval layer for AI applications. Azure AI Search provides the underlying knowledge bases and sources, while Foundry IQ builds on top of it tomake these capabilities reusable across AI agents. #### **Azure AI Search Knowledge Bases and Knowledge Sources ** Azure AI Search introduced agentic retrieval, built around two core concepts: knowledge sources and knowledge bases. These provide the building blocks for advanced retrieval-augmented generation scenarios (RAG). In summary, Azure AI Search’s knowledge base provides a unified retrieval mechanism. You load data into one or more knowledge sources then query the knowledge base which handles query understanding, searches all relevant sources and returns a compiled set of relevant information**. **Azure AI Search’s knowledge base feature can be used in any application where you need advanced semantic search or retrieval-augmented generation. #### **Foundry IQ ** Foundry IQ acts as a central knowledge layer for AI agents, leveraging Azure AI Search capabilities within the Foundry ecosystem. **What is Foundry IQ?** – Foundry IQ is the integration of Azure AI Search into Foundry’s AI agent platform. It provides a unified knowledge base service for AI agents. Foundry IQ lets you create and manage knowledge bases (the same concept as Azure AI Search knowledge bases) directly in the Foundry portal and then use those knowledge bases in AI agents or applications. - - - **Knowledge Bases in Foundry IQ:** In Foundry, you can define a knowledge base around a specific domain or topic – for example, a knowledge base for “HR Policies”, another for “Product Documentation”. Each knowledge base is reusable across any number of agents or applications. Under the hood, these are implemented using Azure AI Search’s knowledge base functionality. - **Knowledge Sources in Foundry IQ:** Just like in Azure AI Search, each Foundry IQ knowledge base is backed by one or more knowledge sources. Foundry allows you to connect to various data sources as knowledge sources. For indexed sources (like Blob storage where you want content fully indexed), Foundry IQ automates the ingestion just as Azure Search does. It will crawl the content, chunk and vectorize it, and store it in a search index managed by the Azure AI Search service. For supported remote sources (like web search), Foundry IQ invokes those systems at query time and feeds results into the retrieval pipeline. - **Agent-aware retrieval orchestration:** When an agent in Foundry sends a query to a Foundry IQ knowledge base, the knowledge base can plan, iterate, and refine the search. For instance, the knowledge base might reformulate it into several targeted sub-queries and query multiple sources accordingly. - **Separation of Concerns:** Foundry IQ allows you to update or improve a knowledge base independently of the agents. For example, if you add a new data source or re-index content, all agents connected to that knowledge base can benefit from the new data. - **Using Foundry IQ in Practice:** You can attach a knowledge base to one of your Foundry Agents. Foundry IQ knowledge bases appear to the agent as a tool it can use. When the agent gets a request, it can invoke the knowledge base to get relevant information. In summary, Foundry IQ provides a knowledge retrieval service for AI agents. #### **Relationship Between Azure AI Search KB and Foundry IQ KB** The MCP endpoint is what the Foundry agent will use when it needs to invoke the knowledge base. When the agent is running it will call the MCP endpoint (which goes to Azure Search), get the results, and then return the content back into the agent’s flow. The use of MCP is essentially abstracting the knowledge base as a tool that the agent can call. You can think of Foundry IQ integration in two phases. First, you prepare your knowledge base on Azure Search (using APIs or the Foundry portal). Second, you tell your Foundry agent about that knowledge base. **Conclusion:** In summary, Foundry IQ knowledge base points to Azure AI Search knowledge base. Azure AI Search provides the infrastructure for creating knowledge bases backed by various knowledge sources. Foundry IQbuilds on this by offering a unified knowledge layer in its AI agent platform, allowing multiple agents to efficiently share and use these knowledge bases. Foundry IQ is the layer that uses Azure AI Search’s knowledge retrieval within the Foundry ecosystem. Interested in learning more about how Spyglass MTG can help you with Azure AI Search and Foundry IQ? [Contact us ](https://www.spyglassmtg.com/contactus)today!  [View full post](https://www.spyglassmtg.com/blog/unifying-retrieval-for-ai-agents-with-azure-ai-search-and-foundry-iq) ```json { "@context" : "http://schema.org", "@type" : "BlogPosting", "author" : { "@type" : "Person", "name" : "Edgar Cuellar" }, "dateModified" : "2026-03-13T20:30:36.911Z", "datePublished" : "2026-02-24T19:23:41Z", "headline" : "Unifying Retrieval for AI Agents with Azure AI Search and Foundry IQ", "image" : { "@type" : "ImageObject", "height" : 627, "url" : "https://5237880.fs1.hubspotusercontent-na1.net/hubfs/5237880/Data2.jpg", "width" : 1096 }, "mainEntityOfPage" : "https://www.spyglassmtg.com/blog/unifying-retrieval-for-ai-agents-with-azure-ai-search-and-foundry-iq", "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject", "height" : 60, "url" : "/hs/hsstatic/content_shared_assets/static-1.4092/img/default-amp-logo.png", "width" : 60 }, "name" : "Spyglass MTG Blog" } } ```