Foundry IQ: Fuel agents with enterprise knowledge and agentic retrieval | BRK246

Pablo Castro presents a Microsoft Build 2026 deep dive into Foundry IQ, Microsoft’s context engineering platform for building agents that can retrieve enterprise knowledge using agentic RAG. The session covers Foundry IQ’s architecture, connecting new knowledge sources, ingestion pipeline customization, retrieval APIs, and performance/evaluation improvements.

Overview

This breakout session focuses on Foundry IQ and its integration with Azure AI Search to help teams ship higher-quality AI agents faster by grounding them in enterprise knowledge.

What the session covers

Foundry IQ architecture (layered model)

• A layered architecture is described, including:
  • Integration layer
  • Retrieval layer
  • Core retrieval engine

Serverless Foundry IQ for agents

• Demonstrates using serverless Foundry IQ with agents.
• Shows direct creation of services.

Connecting knowledge sources and ingestion

• Connecting Foundry IQ to object stores.
• Customizing ingestion pipelines.
• Example setup using Azure Blob Storage.
• Continuous ingestion is discussed as part of the setup.

Data integration context (Fabric)

• Explores Microsoft Fabric architecture and data integration layers in the broader solution context.

Agentic retrieval pipeline

• Explains an agentic retrieval pipeline, including:
  • Query planning
  • Iterative search

Retrieval performance and evals

• Covers performance improvements in agentic retrieval and contributing factors.
• Discusses retrieval performance innovations and evaluation work (noted as the third year in a row).

Retrieval API demo

• Demo setup using a retrieve API for a knowledge base.

Performance demo at scale

• Performance demo using a full English Wikipedia index.

Resources

• https://aka.ms/build26-next-steps
• https://aka.ms/build/foundrydiscord

Session metadata

• Event: Microsoft Build 2026
• Session code: BRK246
• Level: Breakout (Advanced)
• Language: English (US)
• Speaker: Pablo Castro

Chapters (from the video description)

• 0:00 - Explaining Foundry IQ’s layered architecture: integration, retrieval, and core retrieval engine
• 00:10:45 - Demonstration of using serverless Foundry IQ with agents and direct creation of services
• 00:16:09 - Connecting Foundry IQ to object stores and customizing ingestion pipelines
• 00:18:01 - Example setup using BLOB storage and continuous ingestion
• 00:23:11 - Exploring Fabric architecture and data integration layers
• 00:32:08 - Agentic retrieval pipeline explained—query planning and iterative search
• 00:33:52 - Performance improvements in agentic retrieval and contributing factors
• 00:36:09 - Demo setup: retrieve API for knowledge base
• 00:43:12 - Performance demo using full English Wikipedia index and closing remarks