New AI capabilities in Microsoft Fabric
From Copilot and Data Agents to Fabric IQ and AI functions: which AI capabilities Fabric is adding and what you can do with them in practice.
Jesper de Groot
Data Engineer
AI
New AI functionalities in Microsoft Fabric: what does this mean for your organization?
Microsoft Fabric is increasingly evolving from a data platform to an intelligent platform for analytics, automation, and AI agents. In this blog, we list the most important AI functionalities and explain what they practically mean for businesses.
From data platform to AI platform
Microsoft Fabric started as an integrated platform for data integration, data engineering, data warehousing, real-time analytics, and Power BI. Meanwhile, attention is shifting more and more towards AI. Not as a separate addition, but as a layer on top of the existing data architecture.
This development is logical. AI only becomes valuable when answers are based on reliable, up-to-date, and well-managed corporate data. Fabric aims to deliver exactly that combination: centralized data in OneLake, semantic models in Power BI, governance through Microsoft Purview, and AI experiences such as Copilot and Data Agents.
1. Copilot in Microsoft Fabric
Copilot in Fabric helps users analyze, transform, and visualize data. Instead of writing everything manually, data engineers, analysts, and business users can get support with code, reports, data processing, and explanation of existing components.
The biggest change is that AI is moving closer to daily data tasks. A data engineer can set up notebooks faster, an analyst can create visuals quicker, and a Power BI user can move from question to insight faster.
Area | What Copilot can support | Practical value |
Power BI | Reports, visualizations, DAX assistance, and summaries | Faster from question to dashboard |
Notebooks | Generate code, provide explanations, and help fix errors | Less time spent on boilerplate code |
Data Factory | Assistance with data streams and transformations | Build and understand pipelines faster |
Data Warehouse | Support with SQL and analytical questions | Lower threshold for warehouse usage |
2. Fabric Data Agents: asking questions of corporate data
One of the most interesting developments is Fabric Data Agents. This allows organizations to build conversational AI experiences on top of their own data. Users ask questions in natural language and the agent searches for the answer in, for example, a Lakehouse, Warehouse, Power BI semantic model, KQL database, ontology, or Microsoft Graph.
The difference with a general chatbot is important. A Fabric Data Agent is not intended to provide general knowledge, but to answer questions based on controlled corporate data and user permissions. This makes AI more usable for real business processes.
· A sales manager asks: “Which region is lagging behind target and why?”.
· A CFO asks: “Which cost items deviate the most from the budget this month?”.
· An operations manager asks: “Which suppliers cause the most delays?”.
· An HR team asks: “How is absenteeism developing per department?”.
3. Integration with Copilot Studio
Microsoft is working on integrations between Fabric Data Agents and Microsoft Copilot Studio. This allows organizations to build their own agents that utilize Fabric data, which are then made available in, for example, Teams, internal portals, or other workflows.
This makes the step from dashboard to action smaller. A user doesn't just have to open a report, but can ask a question, have a follow-up action suggested, or combine information with other business processes. This is precisely where the business value of agentic AI lies.
Traditional BI | AI agent on Fabric data |
User searches for the right report themselves | User asks a question in plain language |
Insight often remains in dashboard environment | Answer can become part of a workflow |
Analysis requires knowledge of filters and definitions | Agent can include context and definitions |
Action follows outside the report | Agent can help with next steps |
4. Fabric IQ: semantics for AI and decision making
Fabric IQ is a new direction within Microsoft Fabric where business meaning, models, and systems come closer together. The goal is for AI agents to not only see tables but also understand what concepts like revenue, customer, margin, inventory, or forecast mean within the organization.
This is essential, because without semantic context, AI can easily draw wrong conclusions. If two departments define "revenue" differently, an agent needs to know which definition is relevant. Fabric IQ focuses on that layer of shared business meaning.
5. AI functions in Fabric Data Warehouse
Microsoft is expanding Fabric Data Warehouse with AI functions for unstructured text. Think of summarizing, classifying, sentiment analysis, translating, correcting grammar, and executing prompts directly from T-SQL. This opens up interesting scenarios for organizations with large amounts of text data.
Examples include customer feedback, service tickets, survey responses, contract notes, or email categories. Where this previously often required separate AI pipelines, part of this analysis is now moving closer to the data warehouse environment.
AI application | Example in practice |
Summarizing | Condensing long customer feedback into short management insights |
Classifying | Automatically categorizing tickets by issue type |
Sentiment analysis | Tracking customer satisfaction based on text responses |
Translating | Uniformly analyzing international feedback |
6. Data Factory MCP and AI-driven automation
AI-oriented features are also appearing on the data engineering side. Data Factory MCP is in preview and allows AI assistants to create, test, and deploy Dataflow Gen2 components via natural language. This fits into a broader trend where AI not only analyzes but also helps build and manage data solutions.
For organizations, this does not mean that data engineers will become obsolete. It mainly means that repetitive configuration and standard work can go faster, while experts focus on architecture, quality, security, and performance.
7. Governance and security for AI in Fabric
AI on corporate data requires extra attention to governance. Microsoft therefore emphasizes that Copilot and Data Agents must take security, permissions, Purview controls, and responsible AI into account. This is crucial: an AI agent must not display sensitive data to someone who normally does not have access to it.
New features like DSPM for AI in Fabric focus on monitoring AI interactions, detecting sensitive information in prompts and answers, and supporting audit and eDiscovery through Microsoft Purview.
· Start with clear data classification and ownership.
· Ensure permissions in Fabric and Power BI are correctly configured.
· Test AI responses with real business questions and edge cases.
· Define which AI scenarios are allowed and which are not.
What does this mean for businesses?
The new AI capabilities make Fabric more attractive to organizations wanting to move beyond traditional dashboards. The question shifts from “can we report?” to “can employees get reliable answers directly and take action faster?”.
At the same time, the foundation becomes more important than ever. AI amplifies good data foundations, but also magnifies the problems of poor data quality, unclear definitions, and messy permission structures. Anyone wanting to apply AI in Fabric must therefore first ensure a mature data foundation.
Prerequisite | Why important for AI |
Data quality | AI answers are only as good as the underlying data |
Semantic model | Ensures that concepts and KPIs are unambiguous |
Governance | Prevents sensitive data from being misused |
Adoption | Users must know when AI answers are reliable |
Monitoring | Needed to track usage, costs, errors, and risks |
How do you get started with AI in Microsoft Fabric?
The best approach is to start small with a concrete business problem. Don't choose a broad AI transformation right away, but pick a scenario where the data is available, the value is clear, and users regularly ask the same questions.
· Choose one domain, such as finance, sales, operations, or customer service.
· Determine which questions users currently answer manually.
· Check if the required data is reliable and well secured.
· Create a pilot with clear success criteria.
A good pilot proves not only that the technology works, but also whether users trust the answers. That trust is decisive for adoption.
How QUIKK helps with Fabric and AI
QUIKK helps organizations not only set up Microsoft Fabric technically, but also make it practically applicable. This means: a good data layer, clear KPI definitions, reliable Power BI models, and a governance approach that is ready for AI.
For AI scenarios, we first look at the basics. Which data is suitable? Which definitions need to be established? Which permissions apply? And which AI scenario delivers demonstrable value to the business? Only then do we build Copilot, Data Agent, or Power BI scenarios.
· Microsoft Fabric architecture and implementation.
· Power BI semantic models and dashboarding.
· Data governance and security for AI scenarios.
· Pilots with Fabric Data Agents and Copilot applications.
Conclusion
The AI functionalities in Microsoft Fabric make the platform increasingly interesting for organizations that want to turn data into action faster. Copilot accelerates daily tasks, Fabric Data Agents enable conversational analytics, Fabric IQ brings business meaning closer to AI, and new governance features help manage risks.
But the most important lesson remains the same: AI only works well on a reliable data foundation. Want to discover which Fabric AI scenarios are interesting for your organization? QUIKK helps you with a practical roadmap, from data foundation to your first AI pilot.
Sources
· Microsoft Learn - Overview of Copilot in Fabric: https://learn.microsoft.com/en-us/fabric/fundamentals/copilot-fabric-overview
· Microsoft Learn - Release status of AI and Copilot experiences in Fabric: https://learn.microsoft.com/en-us/fabric/fundamentals/copilot-ai-feature-state
· Microsoft Learn - Fabric data agent creation: https://learn.microsoft.com/en-us/fabric/data-science/concept-data-agent
· Microsoft Learn - Create a Fabric data agent: https://learn.microsoft.com/en-us/fabric/data-science/how-to-create-data-agent
· Microsoft Learn - Fabric Data Agent in Microsoft Copilot Studio: https://learn.microsoft.com/en-us/fabric/data-science/data-agent-microsoft-copilot-studio
· Microsoft Learn - What’s new in Microsoft Fabric: https://learn.microsoft.com/en-us/fabric/fundamentals/whats-new
· Azure Blog - Microsoft Build 2026: agentic apps with Microsoft Fabric and Microsoft databases: https://azure.microsoft.com/en-us/blog/microsoft-build-2026-building-agentic-apps-with-microsoft-fabric-and-microsoft-databases/
