Skip to content

BRK360 Breakout Tue 17 Nov 14:00 PST

The semantic layer for the AI era with Power BI and Microsoft Fabric

Eun Hee Kim, Christian Wade

Summary

From the recording / Microsoft AI summary

Microsoft supports Apache Ozzy, an open semantic standard, with DAX as an official dialect. Semantic views in Fabric, in public preview, let teams define shared metrics once in OneLake for use in reports and agents.

Session description

Microsoft's catalogue description

AI is changing how people interact with data, but trusted answers still depend on shared business meaning. Explore how Power BI and Microsoft Fabric provide a trusted semantic foundation that connects data and business context across analytics and AI experiences. Learn how to create consistent definitions, reuse trusted semantics, and help everyone move from questions to insights and decisions faster.

Demo timestamps

Microsoft's player does not support links to a specific time, so open the recording and skip to the time shown.

  1. Skip to 00:06:48

    Creating a semantic model with AI in GitHub Copilot app

    A semantic model is created with AI starting in the GitHub Copilot app; the experience is said to be essentially the same in Visual Studio Code. Fabric Skills plugins are installed.

  2. Skip to 00:14:06

    Validating AI-created model in the UI

    The UI is used to validate what the AI created, including native CI/CD.

  3. Skip to 00:24:24

    Taking context into Cowork for a meeting and deck

    The same context is taken into Cowork to create a meeting and a PowerPoint deck, with invitees inferred from email rather than typed in.

  4. Skip to 00:32:48

    Semantic views in a lakehouse with DAX metrics

    Semantic views are defined in a lakehouse, or by dragging in files, with metrics in DAX.

  5. Skip to 00:36:00

    Copilot surfaces an existing finance team artefact

    A later GitHub Copilot pass has Copilot point out that the finance team already has a relevant artefact.

Watch on-demand (opens in a new tab)

Resources named in the session

← All sessions