Microsoft Fabric Updates Blog

What’s New for Fabric Data Agents at Ignite 2025: Unlocking Deeper Data Reasoning and Seamless AI Interoperability

From Unstructured Data and Ontology to Seamless Integration with M365 Copilot and MCP

Fabric data agents are evolving rapidly, and today we’re announcing many new capabilities that make it easier than ever to build modular, production-ready agentic solutions on your enterprise data in Fabric!

At their core, Fabric data agents are interoperable AI agents that connect your enterprise data to the agentic web. Think of them as virtual analysts, each serving as an expert on a specific domain of data. Data agents are conversational agents with the ability to reason across various data sources in OneLake and deliver insights and knowledge to end users, other agents, and applications.

In this post, we’ll walk you through new capabilities in data agents and what they mean for building agents in Microsoft Fabric, grounded in enterprise data.


Two categories of investments

The new announcements fall into two key areas:

  • Reason across all your data: Expanded data source support for reasoning across all your enterprise data.
  • Interoperability with the AI ecosystem: Ease of integration with other AI systems and applications for building intelligent AI solutions.

Reason across all your data

We recently expanded the data agent’s ability to reason across data by adding support for mirrored databases and SQL database in Fabric. This allows you to build agents in Fabric that reason over data in a variety of sources. Now we are expanding the data source support even more.

A large portion of enterprise data is unstructured, living in documents, PDFs, and other formats that don’t fit neatly into tables. We are pleased to announce that Fabric data agents now support reasoning and providing insights over unstructured data.

Azure AI Search Index in Fabric data agents

To power reasoning over unstructured data, data agents now support custom indexes through Azure AI Search. This means you can create purpose-built search indexes tailored to your specific domains, whether that’s legal contracts, technical documentation, or other types of topics. These custom indexes can then simply be added to a Fabric data agent as a source and the data can be joined with your structured data in OneLake, giving your data agents a complete view that spans both worlds.

Demo: Unstructured data support with AI Search in Fabric data agents.

Ontology as knowledge source for data agents

If you have ever built an AI agent, you probably know that AI agents can perform even better when they can access not just enterprise data, but also enterprise context. This includes semantics, business rules and other specific information for an organization.

Fabric data agents now support Ontology in Fabric as a knowledge source. Ontology is a new Fabric item that brings together rich semantics, with business rules, and organizational knowledge, all of which can be harnessed by AI agents. With this update, you can build data agents directly on top of your business Ontology, unlocking smarter, context-aware agentic experiences.

Demo: Ontology in data agents

Interoperable with the AI Ecosystem

Fabric data agents have always been designed for flexibility and integration. Whether you’re building custom Copilots in Microsoft Copilot Studio, orchestrating multi-agent solutions in Microsoft Foundry, or embedding intelligence into your own apps via the public endpoints, data agents make it easy to bring enterprise-grade insights wherever you need them. These capabilities ensure that your organization’s data expertise is available through a variety of consumer and developer experiences, always with robust security and governance. But the story doesn’t end there. This week, we’re introducing new integrations that take the interoperability of Fabric data agents to the next level.

Hosted MCP Server for data agents

Now, AI systems, including those running in VS Code, can securely connect to Fabric data agents using a managed MCP server endpoint. This means AI solutions can ‘plug in’ to enterprise knowledge on demand, accelerating AI adoption while maintaining secure, auditable access to analytics.

Example of connecting to Fabric data agents using a hosted MCP server endpoint

Data agents now integrate with Microsoft 365 Copilot

Building on the expanded interoperability, Fabric data agents now integrate seamlessly with Microsoft 365 Copilot to enhance those experiences with rich, context-aware data intelligence. By integrating data agents directly into the Microsoft 365 Copilot environment, whether in Teams, on the web, or within the desktop app, users can access the data agents they’re authorized for, right from their familiar workspace. Sharing is just as effortless since users can securely share their own data agents with colleagues across their organization, making collaboration and sharing of insights simple and efficient.

Every question sent to a data agent is securely grounded in your organization’s data and always respects the user’s permissions, including access controls like Row-Level Security (RLS) and Column-Level Security (CLS) on the underlying data sources.

These new capabilities reinforce Fabric’s commitment to open, secure, and intelligent AI experiences, making it easier than ever to unlock the full potential of your data, wherever your users work.

Get started with Fabric data agents

As Fabric continues to innovate, data agents will offer the necessary interoperability across data in OneLake and the Agentic Web. These agents will play a key role in enabling agentic, secure, context-rich analytics over enterprise data across. We encourage you to explore these exciting features and join the Fabric community to help shape the future of data-driven intelligence!

Here are some resources to explore Fabric data agents and all the new capabilities:

If you’re at Microsoft Ignite this week, be sure to check out our breakout session: BRK1739: How Fabric data agents are powering the next wave of AI.

Post Author(s):

Nellie Gustafsson – Principal Product Lead, Azure Data

Joanne Wong – Senior Product Marketing Manager

Amir Jafari – Senior Product Manager in Azure Data

Misha Desai – Principal Product Manager in Azure Data

Shreyas Canchi Radhakrishna – Product Manager in Azure Data

Entradas de blog relacionadas

What’s New for Fabric Data Agents at Ignite 2025: Unlocking Deeper Data Reasoning and Seamless AI Interoperability

enero 8, 2026 por Adi Eldar

What if generating embeddings in Eventhouse didn’t require an external endpoint, callout policies, throttling management, or per‑request costs? That’s exactly what slm_embeddings_fl() delivers: a new user-defined function (UDF) that generates text embeddings using local Small Language Models (SLMs) from within the Kusto Python sandbox, returning vectors that you can immediately use for semantic search, similarity … Continue reading “Create Embeddings in Fabric Eventhouse with built-in Small Language Models (SLMs)”

diciembre 16, 2025 por Alex Powers

As 2025 ends, we’re taking a moment to reflect on Microsoft Fabric’s second year in the market and the collective progress made alongside our community, customers, and partners. What began as a unified vision for data and AI has grown into a platform adopted by more than 28,000 organizations worldwide, anchored by OneLake and shaped … Continue reading “Microsoft Fabric 2025 holiday recap: Unified Data and AI Innovation”