AI agents are rapidly becoming the backbone of digital transformation, but connecting them to the right business knowledge often remains a challenge. Microsoft’s Foundry IQ, now available in public preview, aims to break down these barriers by creating a seamless, unified layer of enterprise knowledge—making it easier than ever for agents to access trusted information securely and efficiently.
A Unified Knowledge Layer for Every Agent
Building and maintaining custom data connections for each AI agent can be complex and resource-intensive. Foundry IQ addresses this pain point by introducing a single endpoint for enterprise knowledge access.
This centralization streamlines knowledge management and ensures that every agent operates with the most relevant, up-to-date information, all while respecting existing permissions and security policies. The foundation of this innovation is Azure AI Search, enabling:
- Reusable knowledge bases that group content by topic, simplifying how agents and apps access the right context via a unified API.
- Automatic data integration for both indexed and federated knowledge sources, handling ingestion, vectorization, and enrichment for various content types.
- Agentic retrieval engine that intelligently plans, searches, and synthesizes answers from multiple sources with configurable reasoning effort.
- Enterprise-grade security with document-level access control and flexible permissions to protect sensitive information.
Eliminating Fragmentation in Retrieval-Augmented Generation
Traditionally, each department must build its own retrieval-augmented generation (RAG) pipelines, leading to duplicated effort and inconsistent data governance. Foundry IQ streamlines this by centralizing retrieval logic in reusable knowledge bases. Teams can focus on defining what knowledge is essential—like HR policies or technical documentation—while agents consume this base without having to rebuild or reconfigure connections.
With support for data sources such as SharePoint, Fabric IQ, OneLake, Azure Blob Storage, and the web, Foundry IQ automates ingestion, chunking, and enrichment, including layout-aware processing for complex documents. This ensures structured, accurate retrieval and a consistent user experience across the organization.
Smarter, Context-Aware Retrieval
What sets Foundry IQ apart is its agentic retrieval engine. Instead of simply searching by keyword, the engine treats every user query as a reasoning task, planning multi-step searches, clarifying ambiguous questions, and delivering context-rich answers with source citations. Developers can adjust the retrieval reasoning effort, offering flexibility between quick responses and thorough, multi-source syntheses.
Security and compliance remain paramount. The system enforces document-level permissions, ensuring users only access content they’re authorized to see. This is crucial for maintaining enterprise trust and regulatory compliance.
- AT&T used this approach to accelerate customer resolution by 33% and scale to 100,000 employees with 71 AI-powered solutions.
- Ontario Power Generation unlocked four decades of institutional knowledge, streamlining onboarding and decision-making.
Enterprise-Ready Governance and Observability
Foundry IQ is designed for robust governance, leveraging Entra-ID-based security and Microsoft Purview integration for data classification and sensitivity labeling. This ensures content governance across its lifecycle, with automated access policy enforcement—no custom coding required.
Central teams gain oversight through the Foundry Control Plane, tracking agent deployment, knowledge base usage, and compliance. Real-time monitoring and auditing tools provide transparency and help maintain security standards organization-wide.
Driving the Future of Enterprise AI
By unifying access to enterprise knowledge, Foundry IQ empowers organizations to move from isolated AI experiments to scalable, mission-critical solutions. Teams can quickly ground new AI agents in trusted, permission-aware knowledge, accelerating innovation while ensuring compliance and security at every step.
Get started with Foundry IQ knowledge bases in the new Foundry portal or in Azure AI Search.
- Video: Foundry IQ for multi-source AI knowledge bases
- Docs: Read Foundry IQ documentation
- Docs: What’s new in Azure AI Search
- Blog: Knowledge base retrieval quality evaluations and results
- Private preview: Sign up for MCP Server knowledge source private preview access
- Repo: Demo AI Search Purview sensitivity labels
- Repo: Azure sample RAG app with Azure AI Search and Azure OpenAI
- Demo app: Knowledge bases and agentic RAG
Foundry IQ: Transforming Enterprise Knowledge for AI Agents