FabCon Europe 2026 is happening now in Barcelona (September 28 to October 1). Microsoft is placing the Power BI semantic model at the center of its analytics and AI products. Copilot answers from it. Fabric Apps build on it. Agents edit it through MCP. If your numbers are not in a governed model, these new capabilities cannot use them.
This matters for every organization still running a large Tableau estate. Here is what Microsoft is announcing this week, how it changes migration planning, and why visual-only conversions miss important AI capabilities.
What Microsoft is announcing at FabCon Europe 2026
Fabric IQ in Microsoft Copilot Chat and Cowork is generally available
Per Microsoft’s official announcement, Fabric IQ now grounds Copilot answers in the Power BI semantic models, metrics, relationships, and business definitions your organization already trusts, with existing access controls and row-level security preserved. It is on by default for Fabric and Power BI customers, with no additional AI token cost. Ask Copilot for the latest sales forecast and the answer comes from the same model that powers the sales report, not from a separate spreadsheet.
Agentic app creation comes to Power BI Desktop
The Azure Data keynote blog introduces agentic app creation in Power BI Desktop: start from a trusted semantic model, describe the app in natural language, preview it, edit it, and publish it to Fabric. These apps can accept inputs, write back data, and hold shared state. Microsoft says the preview will reach Power BI Pro and Premium Per User customers in the coming weeks, with Fabric Apps and Fabric Database capacity up to 1 GB per app included at no additional cost.
Power BI Projects (PBIP) are generally available
The Power BI September 2026 feature summary lists PBIP as generally available. Reports and semantic models now live as plain-text folders: TMDL for the model, PBIR JSON for the report. Git shows line-level diffs, teams can review and merge, and pipelines can lint and automate. Redgate’s PBIP walkthrough is a good practitioner explainer if the format is new to your team.
Agents can now build and change models through MCP
Power BI Agentic Experiences are generally available, and a remote Power BI Authoring MCP server is in preview. An agent can read and change semantic models and reports, refactor and validate DAX, and create a Direct Lake model in a Fabric workspace from a single tool call. Semantic model editors also get a new control to block read-only users from reaching a model through Copilot or MCP, which gives administrators direct control over AI access.
Also in the release
- Microsoft 365 Copilot inside Power BI and Fabric (preview)
- Create and edit DirectQuery semantic models in the Power BI service
- Approximate distinct count for Import and Direct Lake, string indexing, and full-text search in semantic models (preview)
- IQ Sharing for governed cross-organization context, a Fabric data engineering agent for migrations and lakehouse modernization, and Database Hub in Fabric (preview)
Why FabCon Europe 2026 changes the Tableau conversation
Organizations often move from Tableau for lower costs, Microsoft 365 alignment, and platform consolidation. Those reasons still apply. This week adds another: the semantic model is now the interface for AI across Copilot, Fabric Apps, and agents. A Tableau workbook is not a semantic model. A Power BI report recreated screen by screen from a Tableau workbook is not one either.
Copilot does not read a chart. It reads measures, relationships, descriptions, and business definitions in the model. If a migration recreates visuals on a flat, undocumented table, Fabric IQ has no reliable business context. The dashboards render, but Copilot cannot provide reliable answers from them.
What a Tableau migration now needs to include
Because these tools depend on the semantic model, a Tableau migration must cover the logic and governance behind each workbook, not just its dashboard layout. That requires four deliverables:
- Business logic in DAX. Translate Tableau calculated fields and LODs into documented measures that can be tested and reused across reports.
- A structured semantic model. Resolve Tableau blends and custom SQL into clear relationships, dimensions, and a dedicated date table.
- Security parity. Recreate row-level security in the semantic model so reports, Fabric IQ, and Copilot apply the same access rules.
- Versioned project files. Export PBIP and TMDL so your team can review changes in Git and use agent-based development tools safely.
These are core migration requirements, not optional AI preparation. They keep calculations and access rules consistent while giving Copilot reliable business context. Our step-by-step Tableau to Power BI guide covers the full workflow. For a closer comparison of the underlying data structures, see published Tableau data sources versus Power BI semantic models.
What this means if you are mid-migration
- Assess for model readiness, not just chart count. Inventory calculated fields, LODs, blends, and security rules. Those are the parts that become the semantic model.
- Export to PBIP from day one. PBIP is GA. Treat your migrated output as code you can diff, review, and hand to an agent.
- Write descriptions. Copilot and Fabric IQ use them. A migrated measure with no description is a measure AI cannot explain.
- Validate numbers before visuals. Check data parity, calculation parity, and security parity before UAT. Copilot uses the same definitions, including any errors.
BIChart converts Tableau workbooks into Power BI semantic model structure plus visuals, translates calculated fields to DAX, and exports .PBIP for Power BI Desktop or publishes to Fabric. About 80% of each dashboard is typically ready to go, with a 50-70% effort reduction versus a manual rebuild. Speed matters, but model quality matters more. The output must be a semantic model Copilot can use.
See the semantic model your Tableau estate becomes
If the FabCon Europe 2026 announcements make Copilot a higher priority, start with a Tableau migration assessment to see which workbooks carry the logic worth modeling, or Try Migrating Now on a workbook you know. Product overview: Tableau to Power BI migration and Tableau to Microsoft Fabric.
