Fabric now spans lakehouses, warehouses, real-time intelligence, SQL databases, Power BI and a growing family of data agents. The question is no longer what Fabric can do, but what is ready for production and how to run it well. SQLBits is where practitioners compare notes.
Capacity sizing and cost control, Direct Lake, CI/CD, governance, migrating from Synapse and Data Factory, and building agents on your own data. Sessions are demo-led and honest about what is GA and what is still preview.
Microsoft is a headline sponsor of SQLBits, bringing members of the Fabric product, engineering and Customer Advisory teams. Ask them how to make Fabric work in your organisation.
If any of these sound familiar, you will find people at SQLBits who have already been there.
OneLake gives every workload one copy of data under one security model, and agents inherit it. OneLake security (row and column level) is moving towards GA.
Data agents plug into Copilot Studio, Azure AI Foundry, Teams and Microsoft 365 Copilot, and can be called by external orchestrators.
Fabric IQ ontology (preview) defines entities, relationships and rules once, so agents and apps use the same vocabulary. It can be generated from a Power BI semantic model.
Graph in Fabric adds a native graph engine for relationship-heavy questions. Check its current status before quoting.
DiskANN vector indexes on Fabric SQL and Azure SQL (GA) put semantic search next to operational data.
Mirroring lands source data in OneLake as Delta. BigQuery mirroring is GA, and Snowflake security-role mirroring is in preview.
Eventstreams, Eventhouse and Activator feed live signals to operations agents.
AI functions (summarise, classify, transform) run in notebooks, with Warehouse AI functions in preview.
Eight places agents are already earning their keep in Microsoft Fabric, and where SQLBits sessions will show you how.
Fabric data agents (GA) answer natural-language questions over lakehouses, warehouses, semantic models, KQL databases, mirrored databases and ontologies, read-only and under your security model.
Copilot generates and fixes SQL, notebook code and Dataflow and pipeline logic. It is generally available in Data Factory, SQL database, Power BI and Real-Time Intelligence, and in preview in notebooks and the Data Warehouse.
The Data Engineering Agent (preview, announced at FabCon Europe, September 2026) plans and executes migrations, ETL and performance tuning inside guardrails you set.
The operations agent in Real-Time Intelligence monitors live data and triggers alerts or playbooks. Microsoft Learn lists it as GA.
Observability and operations agents (preview) centralise telemetry, alerts and root-cause analysis, building on workspace monitoring.
Copilot writes semantic model descriptions (preview, April 2026), and the OneLake catalog search API, MCP and CLI tools (March 2026) help agents and people find the right data.
The local Fabric MCP server (GA, March 2026) and the remote server (preview) let GitHub Copilot, Claude and others scaffold items, query data and deploy. A Data Factory MCP server is also available.
Copilot in Power BI (GA) writes DAX and measure descriptions. Agentic app creation in Power BI Desktop (preview, announced September 2026) builds apps from a description.
All of these speakers have appeared on the SQLBits stage – check our 2027 agenda when it’s announced for details of our 2027 Fabric speakers.
Author and blogger (crossjoin.co.uk) and a leading authority on Analysis Services, Power Query and Dataflows.
Works on Fabric mirroring and Data Warehouse migrations.
Author of five SQL Server books, former MVP and co-host of the Guy in a Cube channel.
Works across the tools and experiences that power SQL Server, Azure SQL and SQL database in Microsoft Fabric. Host of Data Exposed, connecting the people building Microsoft’s data technologies with the people putting them to work.
