Copilot Across Fabric, Power BI and Microsoft 365

What's inside

1. Level 1: The Copilot Family and What Each One NeedsFree
   Describes each Copilot experience across Power BI, Fabric workloads, Fabric data agents and Microsoft 365 Copilot, and which licence or capacity each one needs, using Microsoft Learn read on 2026-10-20 (a paid F2+ Fabric capacity or P1+ Premium capacity; trial capacities not supported; standalone Power BI Copilot is in preview).
2. Level 2: Tenant and Admin Settings for CopilotFree
   Sets the Fabric admin portal tenant settings that Copilot and data agents depend on: enabling Copilot, cross-geo processing and storage for AI, the standalone cross-item Copilot setting (preview) and the option to show only approved items.
3. Level 3: Model Hygiene: Names, Descriptions, Hidden Fields
   Gets a semantic model ready for AI by giving tables, columns and measures clear business names, writing descriptions and hiding helper and key fields so Copilot reasons over the right objects.
4. Level 4: Synonyms and AI Data Schemas
   Adds synonyms through the Q&A linguistic schema and uses an AI data schema in Prep data for AI (preview) to limit Copilot to the subset of the model it should use.
5. Level 5: Verified Answers, AI Instructions and Approval
   Adds verified answers with trigger phrases and AI instructions (both part of Prep data for AI, preview), tests them with the skill picker and 'How Copilot arrived at this', then marks the model Approved for Copilot.
6. Level 6: Copilot in the Report Pane
   Uses the Copilot pane on an open report to summarise pages, answer questions about visuals and data, and create report pages in edit mode, then checks each answer against the report.
7. Level 7: Standalone Copilot in Power BI (Preview)
   Uses the full-screen standalone Copilot (preview) to find reports, semantic models and apps you can access, ask questions across items, and attach a specific item when the right source matters.
8. Level 8: Writing and Explaining DAX with Copilot
   Uses Copilot to write, explain and improve DAX measures and queries, then checks the results with DAX query view, known totals and edge cases before adding them to the model.
9. Level 9: Copilot in Fabric Notebooks
   Uses Copilot in Fabric notebooks to generate, explain and fix PySpark or pandas code over lakehouse data, and reviews the generated code before running it on real tables.
10. Level 10: Copilot in Data Factory Pipelines and Dataflows
    Uses Copilot in Data Factory to build and explain Dataflow Gen2 transformations and pipeline activities, then checks the steps it generates before scheduling them.
11. Level 11: Copilot for SQL in Warehouses and SQL Endpoints
    Uses Copilot in the Fabric warehouse and SQL analytics endpoint to write, explain and fix T-SQL queries, and tests them against expected results.
12. Level 12: Building a Fabric Data Agent
    Creates a Fabric data agent (generally available as of 2026-10-20) over a lakehouse or warehouse, choosing up to five data sources and only the tables that matter, within its read-only limits.
13. Level 13: Tuning Data Agent Instructions and Examples
    Improves data agent accuracy with agent instructions that route questions and define terms, plus example question-query pairs (up to 100 per source; not supported for semantic model sources), then publishes the agent.
14. Level 14: Using Data Agents from Copilot and Copilot Studio
    Makes a published data agent available outside Fabric, through Microsoft 365 Copilot and as a connected agent in Copilot Studio (preview as of 2026-10-20), and checks each feature's current status in the documentation.
15. Level 15: Power BI Data in Excel, Teams and PowerPoint
    Brings Power BI semantic model data into Excel, Teams and PowerPoint with Copilot and Microsoft 365 Copilot, checking current feature status and making sure figures trace back to the governed model.
16. Level 16: Security: Permissions, RLS and Sensitivity Labels
    Confirms that every Copilot experience and data agent answers under the user's own permissions, respects row-level and column-level security, and follows sensitivity labels and Purview policies.
17. Level 17: Evaluating Answers and Catching Made-Up Numbers
    Uses a repeatable evaluation method of test question sets, expected answers, source tracing and repeat runs to catch hallucinated or wrongly scoped numbers in Copilot and data agent output.
18. Level 18: Governance, Adoption and Next Steps
    Plans governance and adoption for Copilot over Power BI and Fabric (approval workflow, capacity monitoring, audit, Git and deployment pipelines, user training) and maps next steps to PL-300 and DP-600.
19. Level 19: Final Challenge: Trusted Answers End to End Exam
    Brings the whole course together in one scenario: confirm licences and settings, make a model AI-ready, build and tune a data agent, use it in Microsoft 365, check security and evaluate the answers.

Access

The first 2 levels are free with a free account. Every level, the podcast edition and the AI tutor come with All Access at £4.99/month or any Creator plan — see pricing.

Start free →