Microsoft Fabric Blog authors Nellie Gustafsson, Joanne Wong, Amir Jafari, Misha Desai, and Shreyas Canchi present new capabilities for Fabric data agents unveiled at Ignite 2025, emphasizing AI interoperability and modular enterprise data analytics.

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

Introduction

Microsoft Fabric data agents have received significant updates, enhancing their ability to provide modular, production-ready agentic solutions on enterprise data. These agents operate as interoperable AI modules, serving as domain experts and enabling reasoning across structured and unstructured data sources within Fabric.

Expanded Reasoning and Data Source Integration

New capabilities allow Fabric data agents to reason over increasingly broad data types:

  • Support for Unstructured Data: Agents now analyze documents, PDFs, and other unstructured formats, extending insight beyond tables.
  • Mirrored Database and SQL Integration: Full reasoning support over mirrored and SQL databases in Fabric.
  • Ontology Integration: Agents leverage Fabric Ontology for richer semantic context, including business rules and organizational knowledge, enabling smarter, context-aware operations.
  • Azure AI Search Index: Purpose-built search indexes from Azure AI Search empower agents to bridge structured and unstructured data for comprehensive insights.

Interoperability with the AI Ecosystem

Fabric data agents connect seamlessly across the tech landscape:

  • Integration with Microsoft Copilot Studio and Foundry: Build custom Copilots and orchestrate multi-agent workflows.
  • Hosted MCP Server Endpoints: Securely connect agents to remote AI systems (including VS Code) for real-time, auditable enterprise analytics.

Microsoft 365 Copilot Integration

Agents now fully integrate with Microsoft 365 Copilot:

  • Context-Aware Data: Users in Teams, web, or desktop environments can interact with authorized agents directly, supported by robust Row-Level (RLS) and Column-Level Security (CLS).
  • Collaborative Sharing: Secure sharing and collaboration features allow users to disseminate insights organization-wide.

Security and Governance

Data agents respect organizational security boundaries, ensuring:

  • Permission-based Access: All insights remain governed by user permissions and enterprise security models.
  • Auditability: Real-time access tracking through MCP endpoints.

Getting Started

Visit these official resources to learn more and begin using Fabric data agents:

Community and Events

Attendees at Microsoft Ignite can join breakout session BRK1739 to discover how Fabric data agents power the next wave of AI-driven analytics.

Authors:

  • 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

This post appeared first on “Microsoft Fabric Blog”. Read the entire article here