Microsoft has announced a broad set of Fabric and Azure Database changes aimed at grounding AI in company data and automating more data work. The practical shift is that Copilot can answer business questions using governed Power BI definitions, while Fabric adds tools for building agents and managing data operations.
Microsoft AI Watch analysis
What happened
At the European Microsoft Fabric + SQL Community Conference in Barcelona, Microsoft announced updates spanning Fabric, Copilot and Azure databases. The official announcement says Fabric IQ is generally available in Copilot Chat and Cowork, where answers can draw on governed Power BI semantic models, metrics and business definitions. Integration with Copilot’s Code experience is planned through the Frontier programme.
Other changes include a preview of IQ Sharing for sharing data and business context across organisations, and new agentic data engineering features. Microsoft says engineers set the outcome and boundaries, after which an agent can plan and carry out work, validate results and refine its approach. Fabric Apps adds capabilities for building applications on governed data. Database Hub is in public preview, with a central view for managing several database types.
Our top picks
- Copilot grounded in Power BI context
Answers in Copilot Chat and Cowork can use the same governed definitions and metrics as an organisation’s Power BI reporting. - Agentic data engineering
Engineers define the goal and boundaries; an agent can plan, execute and validate work within them. - Cross-organisation context sharing
IQ Sharing is in preview, with governance controls for sharing Fabric data and the context that explains it. - A broader database control plane
Database Hub is in public preview for monitoring and governing SQL Server, Azure SQL, PostgreSQL, Cosmos DB and Fabric SQL databases. - More flexible SQL deployment
SQL Server on Azure Local is generally available; disconnected operations are in preview for environments with restricted or interrupted cloud access.
Why it matters
AI answers are only useful when organisations can tell which business definitions and data sit behind them. Connecting Copilot to governed Power BI models is a concrete attempt to reduce the familiar meeting ritual of arguing over whose number is the right number.
The wider announcement also moves beyond chat: Microsoft is putting agents closer to data engineering, application development and database operations. That could make routine work easier to automate, while leaving teams to decide what the agent may do and how its work is checked.
Our read
There is substance here: the Copilot connection has a clear user-facing purpose, and the data-engineering features describe a bounded workflow rather than simply promising autonomy. The important distinction is availability. Some capabilities are generally available, while IQ Sharing and disconnected operations remain previews; the announcement does not establish how well the new tools perform in customers’ day-to-day work. Start with the governed-data connection if you already use Power BI, and treat the preview features as experiments, not settled infrastructure.
What to watch
- Whether Copilot answers consistently follow the definitions and access controls in Power BI models.
- How teams set and audit boundaries for agentic data-engineering work.
- When IQ Sharing and the specialist database agents move beyond preview.
- Whether customers report measurable benefits from these changes in real workloads.
Discussion spark: Would you trust an AI agent to carry out data-engineering work once you have set its boundaries, or should a person approve every consequential step?
Sources and evidence
- New Microsoft data innovations unlock what only your business knows – The Official Microsoft Blog (29 September 2026, 06:49 UTC)
not affiliated with or endorsed by Microsoft