Power BI Weekly

Issue 380

23^rd October 2026

No official announcements from the Power BI team this week, so straight into the community content.

Last week's Feature Summary brought us the native Gantt chart (in preview), and Bas Dohmen (and numerous others, as you will see later in the newsletter) has wasted no time building a full project Gantt chart with it, complete with milestones, owners, grouped phases and a zoom slider. Also for report authors, Juliana Smith has put the "Have tab order match visual order" button to the test in Tab Order in Power BI: Sorted by Position, Not Logic, showing that a few pixels' difference in height is enough to jumble the keyboard order of a page, which matters for anyone navigating your reports with a keyboard or screen reader.

Over on the modeling side, Patrick LeBlanc has walked through Direct Lake calculated columns, covering where they work, how user context comes into play, and why pushing the logic back to the source is usually still the faster option. Finally, Marco Russo has announced the DAX Lib MCP server, which lets your AI assistant search DAX Lib and install user-defined function packages straight into a model open in Power BI Desktop.

Data Prep

Looking for Patterns Instead of Exceptions: A Better Way to Clean Dirty Data in Power Query by Jason Cockington · 1 min read This article introduces a methodology for using Power Query to identify and standardize common data patterns, improving cleaning efficiency over traditional exception handling approaches.

Data Modeling

Install DAX libraries into Power BI Desktop by asking your AI assistant by Marco Russo · 3 min read The new DAX Lib MCP server enables AI assistants to directly install DAX libraries into Power BI Desktop models, automating function downloads while maintaining original code.
How to Use Direct Lake Calculated Columns in Power BI by Patrick LeBlanc Patrick LeBlanc walks through Direct Lake calculated columns (preview) using a real scenario where changing the source warehouse wasn't an option. He explains why they work on Direct Lake on OneLake but not on SQL, the TMDL change, user-context functions such as user culture, how they differ from measures, and why source-side columns usually remain the better-performing choice.
Microsoft Wrote Power BI Rules for AI. I Disagree With 2 by Fernan Espejo The video critiques two specific Power BI AI rules from Microsoft’s “Skills for Fabric” guide, offering alternative best practices for calculations, naming, and data hygiene.
Dynamic DAX query to squash many-to-many relationships by Michael Steedle · 3 min read A portable DAX query, built on INFO.VIEW.RELATIONSHIPS() and COLUMNSTATISTICS(), that flags relationships in import models where the data supports a more efficient cardinality (many-to-one or one-to-one) than is currently set. It includes an optional check for blank keys, which block the "One" side.
Monitoring The Impact Of Full-Text Indexing On Semantic Model Refresh In Power BI by Chris Webb · 2 min read Chris Webb uses a Profiler trace to measure how long building a full-text index adds to semantic model refresh. On a street-name column with 73,206 distinct values, the build took about 1.4 seconds (484ms of CPU). He argues the feature is worth exploring for "chat with your data" scenarios, even though reports can't use it yet.
PBIXRay for macOS: Open and Inspect PBIX Files on a Mac by Igor Cotruta · 1 min read Igor Cotruta, creator of the open-source pbixray Python library, introduces a native, read-only Mac app for inspecting PBIX and .abf files without Power BI Desktop or Windows. It shows the model diagram, DAX and Power Query, VertiPaq-style column metrics, stored data, RLS and report-page wireframes, plus a Finder Quick Look preview, all processed locally. The viewer is free; a one-time in-app purchase unlocks other features.
5 DAX Mistakes That Quietly Break Your Power BI Reports by Alan Conlon · 1 min read Five DAX mistakes that validate and refresh fine but return the wrong number, each with a defensive fix. Prefix variables (vStatus) to avoid name clashes, and use IN or CONTAINSROW instead of chained = comparisons, which treat BLANK as 0 or "". Handle blanks first in SWITCH(TRUE()), avoid SELECTEDVALUE in calculated columns, and use DIVIDE() rather than / except when dividing by a constant.
Where is This Business Logic? by Teo Lachev · 2 min read The article recommends placing materialized calculations and data quality issues in ETL, context-dependent analytics in the semantic model, and presentation-specific features in reports.

Report Authoring and Interactivity

The Dropdown Slicer We Waited 10 Years For (Power BI) by Bas Dohmen Power BI's list slicer gains a dropdown mode that shows your selected items at the top, supports search, and offers a "Show selected first" option so you no longer have to scroll to see what's filtered.
Power BI September 2026 Update! 🤯 Small Multiples & Slicer Sliders by Amit Chandak Amit Chandak walks through two features from the September 2026 update: visual calculations now working with small multiples (including the ROWPAGES axis), and new colour and handle-style options for the slicer slider.
Tab Order in Power BI: Sorted by Position, Not Logic by Juliana Smith · 5 min read The "Have tab order match visual order" button in Power BI sorts elements by their screen coordinates rather than logical layout, which can lead to unexpected tab sequences for stacked or closely positioned visuals.
Power BI Finally Has a Real Gantt Chart by Bas Dohmen Power BI now includes a native Gantt chart with built-in features like milestones, status tracking, owners attribution, phase grouping, and an interactive zoom slider for detailed project management.
Let’s Go Gantt: My 10-Minute Experiment. by Parul Rani Sagar · 1 min read The Power BI Gantt visual displays course timelines, enrollment counts, completion rates, and customizable time axes with progress bars and milestones for effective project scheduling analysis.
New Gantt Chart in Power BI (October 2026 Preview) by Ashish kumar Jha A first look at the native Gantt visual in preview with the October 2026 update: how to enable it under Preview features, and how phases, task durations and colour-coded timeline bars come together for a sample project plan.
Power BI Gantt Visual (Preview): First Look and Workarounds by Juliana Smith · 6 min read A project-controls view of the new native Gantt visual (preview): it beats the AppSource options for stakeholder timelines, but has no baselines, no today line and no conditional formatting on columns. DAX workarounds cover milestones stored with equal start and end dates, matrix-style column repetition, and a dynamic subtitle.
The new Gantt chart in Power BI by Valerie Junk · 1 min read Power BI’s preview native Gantt visual lets users easily track project timelines, format tasks, and highlight overdue items in red for better resource management.
Continuous axis for any Date table in Power BI by Marco Russo · 1 min read The auto date/time table keeps a continuous X-axis at month level because of specific metadata, and you can apply the same to your own Date table. Use a user hierarchy that starts at a Year level (dataCategory: Years) and ends at a key Date column (the undocumented dataCategory: PaddedDateTableDates). These are set in TMDL, Tabular Editor or via the Power BI Modeling MCP server, with notes on tooltip redundancy.
Autoselect the Latest Month in a Slicer (Another Attempt) by Dane Belarmino This Power BI video explores a new method aiming to automatically select the latest month in a slicer without altering the displayed month and year, addressing limitations of previous workarounds.
ChatGPT 6.0 Codex + Power BI | Part 6 by Pankaj Namekar Part 6 of Pankaj Namekar's ChatGPT + Power BI series brings the earlier episodes together: Codex builds a complete Power BI report from a brief, including pages, visuals, navigation and the DAX measures, with guidance on correcting it and checking the result.
ChatGPT Codex + Power BI | Part 7 by Pankaj Namekar Part 7 of Pankaj Namekar's ChatGPT + Power BI series uses Codex to generate a Power BI theme JSON file from prompts — brand colours, fonts and visual defaults — then imports it into Power BI Desktop and iterates on it.
Botei o Claude pra fazer meu Dashboard no Power BI e ficou ABSURDO [Portuguese] by Alon Pinheiro Claude, connected to Power BI Desktop, builds the calendar table, relationships and measures for a sales model, then generates HTML Content visuals for KPI cards, trend and ranking charts that still respond to slicers. The video closes with a reminder to audit the numbers. [Portuguese]

Deployment, Security and Operations

ChatGPT Codex + Power BI | Row Level Security Setup and Testing 🔥 | Part 8 by Pankaj Namekar Part 8 of Pankaj Namekar's ChatGPT + Power BI series has Codex generate and test 24 row-level security roles (by region and store) for a sample retail model. It then verifies them with View as Roles and covers assigning users in the Power BI service.
Why To Move All Fabric Connections to Service Principal Authentication by InsightsByV · 3 min read Moving fabric connections to service principal authentication eliminates dependency on personal accounts, simplifies permission management, and enhances security and business continuity for reporting solutions.

General

5 core concepts every Power BI developer must know by Bas Dohmen · 1 min read Power BI developers improve report consistency and performance by prioritizing explicit measures, custom date tables, star schemas, and context-aware DAX calculations.
Power BI + Claude: Solving a Real Interview Case Study in 30 Minutes by Okeh Efasa The video shows how to use Power BI and Claude AI within 30 minutes to turn a cosmetics brand's interview case study into an executive deck, highlighting data analysis, strategic planning, and storytelling.
Mapa AI nástrojů kolem datové platformy by Jiří Neoral · 6 min read Jiří Neoral maps the AI tools around the Microsoft data platform into three tiers: asking questions (Copilot in Power BI, Microsoft 365 Copilot with Fabric IQ, Fabric data agent), building agents (Copilot Studio vs Foundry) and developing with agents (MCP servers, GitHub Copilot, Claude Code). Prep data for AI on the semantic model is the shared foundation, and a summary table sets out each tool's limits. [Czech]