Power BI Weekly

Issue 373

4th September 2026

Welcome to the 373rd edition of Power BI Weekly! First of all - a bit of a PSA. You may have noticed that I've become quite lax with the "Tuesday publication" precedent (despite best intentions), with Friday being the more common publication day recently. This just fits my week better now, so please expect the newsletter on Fridays going forward, ready for your weekend activities! Enough about me.

There's a "get your connections in order" theme running through this week's official news, starting with the ODBC to ADBC driver transition in Power BI and Microsoft Fabric, which has now reached general availability. If your reports lean on ODBC-based connectors, this is worth validating sooner rather than later. Sticking with connection management, the team has also announced Connection Recency APIs (currently in preview), which let gateway admins track and manage Fabric connections at scale, a useful governance lever for anyone trying to work out which connections are actually still in use before they clean house. Note that this doesn't yet mean tenant admins can manage all cloud connections in the tenant, but that feature is in the works (after a question on that post by yours truly, along with another like-minded individual.)

Elsewhere from the team, the Power BI sample reports have been refreshed with modern visual defaults, updated semantic models and mobile-friendly layouts, a handy reference if you want an up to date example of current best practice rather than the slightly dated defaults the samples had settled into. And on the Fabric <-> Databricks side, there's a new post on modern Power BI architecture choices for reporting on Azure Databricks, benchmarking performance across different storage modes, worth a read if Databricks is your source system and you're weighing up Import against DirectQuery or Direct Lake.

Outside of official content (though coincidentally still written by MSFT employees), Mim is back with a very interesting article digging into what constitutes a well-structured dataset that performs well with the Power BI engine: Writing Parquet That VertiPaq Likes. Finally, Phil Seamark has written up a compilation fix for calculation groups that should speed up query compilation for models with several cross referencing calculation groups, without you needing to change anything yourself.

Data Prep

Data Modeling

  • Teach Claude Your Power BI Standards, Custom Skills for the Power BI Modeling MCP Server You can teach Claude and other LLMs your Power BI modeling standards and custom skills using the Power BI Modeling MCP Server, enabling AI to generate reports and semantic models that adhere to your specific guidelines.
  • Matrix Totals – The Whiteboard #13 In this episode of The Whiteboard, experts discuss matrix totals using interactive DAX concepts, featuring a special guest from SQLBI.
  • AI Can Write DAX. So What Happens to Power BI Developers? Rob Collie explores AI's impact on Power BI development, highlighting semantic models' importance and questioning if classic skills like DAX and data modeling will stay vital for new professionals.
  • Power BI calculation groups just got faster to compile Power BI optimized calculation groups to drastically cut formula engine wait time by enhancing multi-item calculation handling, speeding up query compilation without impacting runtime.
  • Fabric Data Agent Giving Wrong Answers? Fix It with AI Agent Instructions This video explains why a Fabric Data Agent may provide incorrect answers in M365 Copilot even when the underlying Power BI report shows accurate results, and shows how to fix the issue with proper AI agent instructions.
  • Dynamic Data Dictionary with DAX DMVs The article shows how to use DMVs in DAX to automatically generate measure and column descriptions for consistent business glossary visibility across reports.
  • Translating Power BI Semantic Models with PowerShell This guide shows how to use PowerShell 7 or Power BI Desktop's XMLA endpoint in Fabric to translate a report's technical metadata like table/column names and measures into various languages without changing the data.
  • Is Power BI Dead? Is Microsoft Fabric only a Data Analytics platform? Amit Chandak builds a Microsoft Fabric App with the Rayfin SDK, including a Data Entry Sales app and an attached Power BI semantic model, to explore whether Fabric Apps point toward a future beyond traditional Power BI reporting. He frames it as a discussion rather than a verdict: the goal isn't replacing Power BI, but understanding how Fabric Apps, custom development and semantic models might work alongside it.
  • Why Your Power BI AI Answers Are Wrong (And How to Fix It) This video explains why Power BI AI answers can be wrong or slow due to semantic model issues and provides a step-by-step toolkit to fix these problems across all AI surfaces in Power BI.

Report Authoring and Interactivity

Deployment, Security and Operations

General

  • Will AI Replace Power BI Developers? AI will not fully replace Power BI developers, but the role now emphasizes validating AI-generated outputs over traditional building.
  • Is Power BI Still Worth Learning in 2026? Power BI will remain valuable in 2026 due to the growing importance of AI-generated report verification over manual creation, as shown by Gartner's recognition and real-world applications of agentic AI tools for data model auditing.

Get Power BI Weekly in your inbox every Friday

Sign up to receive the email every week and keep on top of all the announcements.

Don't forget to confirm your subscription. We have a double opt-in sign up process to prevent spam.

We will never provide your details to any third parties. We don't spam.

Power BI Weekly Logo