The AI-Assisted Data Developer: VS Code, Copilot & Claude

The AI-Assisted Data Developer: VS Code, Copilot & Claude Across Fabric, Power BI and SQL Server
AI features keep landing inside the Microsoft data platform — but the biggest productivity jump happens in your editor, not in the portal. In this demo-driven session I'll take one business request and build the solution end-to-end without leaving VS Code: exploring and refactoring T-SQL with the MSSQL extension and Copilot agent mode, scripting and deploying Fabric items with the Fabric CLI, editing semantic models as code with TMDL and PBIR, and generating DAX (including the new DAX user-defined functions) with AI assistance — all under source control with Git.

Along the way I'll show where GitHub Copilot shines, where Claude does better, the prompting patterns that actually work for T-SQL, DAX and PySpark, and the guardrails you need so AI-generated code never reaches production unreviewed. Expect honest bloopers: you'll see where the AI gets it wrong and how to catch it.

You'll leave with a setup checklist (extensions, MCP servers, repo layout), a set of reusable prompt patterns, and a practical decision guide for what to delegate to AI — and what not to.