Power BI and AI

Power BI June 2026: what changed and what actually matters for your work

The June 2026 Power BI update brings real workflow changes: Copilot inside the semantic model, AI agents building reports from scratch, Date Picker for slicers, and several features graduating from Preview to GA. Here is what deserves your attention.

Published article

Renan Brognoli

2026-07-2110 minPower BI

Microsoft released the June 2026 Power BI monthly update on June 9th. Unlike many months where the highlights amount to cosmetic tweaks, this update brings changes that will genuinely affect how both developers and report consumers work day to day.

AI is the headline, but not in the vague, buzzword-heavy way you might expect. The AI features in this update have a clear purpose and practical application. Several capabilities that had been in Preview also reached General Availability, meaning you can safely use them in production without worrying about unpredictable behavior.

Copilot in Web Modeling: AI inside the semantic model

Until now, Copilot in Power BI was effectively a tool for report consumers. You asked something, it tried to answer with a visual. Useful in some cases, disappointing in others.

With Copilot in Web Modeling, the dynamic shifts. Now the AI works alongside whoever is building the semantic model, not just whoever uses the end result.

In practice, you can point to a measure that is returning incorrect results and ask what is wrong. You can ask Copilot to identify unnecessary relationships or redundant measures. You can ask questions about the model structure in plain language and get a comprehensible explanation back.

For anyone maintaining models with hundreds of measures and complex table relationships, this could meaningfully cut down debugging and documentation time. It is still in Preview, so do not expect perfection, but the direction is clear.

AI-Powered Reporting Skills: agents building reports from scratch

This is the feature that will divide opinions the most. AI-Powered Reporting Skills allow AI agents to handle the entire report creation workflow: planning, design, visual layout, validation, and publication, all through natural language prompts.

The idea is you describe what you need and the agent delivers a finished report.

For non-technical profiles who need to create simple reports, there is real value here. For experienced developers, the usefulness depends heavily on how much control you retain over the output. In Preview, healthy skepticism is warranted. But if Microsoft manages to refine this, the impact on the volume of routine report-building work could be substantial.

Fabric Apps for Semantic Models: the model as a reusable asset

This is the most strategically significant feature of the month, even if it is generating the least noise on social media.

Fabric Apps for Semantic Models allows you to build complete applications directly from an existing semantic model using natural language. The governance, metrics, and permissions already defined in the model remain intact. What changes is that anyone with access can now create a functional application on top of that model without touching the underlying code.

The engineering team builds and maintains the semantic model with technical rigor. The business team, working from that same model, creates the applications they need without filing a ticket to IT for every change. If it works as advertised, it resolves a long-standing friction between technical and business teams.

Date Picker for Slicers: small feature, big time saver

Among the reporting updates, the Date Picker for Slicers deserves special attention because it fixes a common problem: date filters that go stale and require manual maintenance.

With the Date Picker, you set a relative default date that automatically rolls forward as the data refreshes. Report users can still switch to a specific date when needed, but the default is always current.

For monthly, quarterly, or rolling-window reports, this eliminates a maintenance task that wastes time and introduces silent errors.

What reached GA: safe to use in production

Several features that had been in Preview reached General Availability with this update, indicating Microsoft considers them stable enough for production use.

  • Shape Map: useful for geographic maps with custom regions such as state-level breakdowns or custom sales territories.
  • Auto-expand for Matrix: removes the need to manually expand hierarchies every time a report loads.
  • Axis improvements for bar and column charts: removes padding between the axis and bars, and fixes the axis range issue when enabling data labels.
  • Card visual hover state: define a different image or style when the user hovers over the card.
  • Conditional formatting of fixed column widths in Matrix and Table: addresses a long-standing complaint from anyone working with dense information tables.

Sentence-format tooltips: context without DAX

The sentence-format tooltip option was in Preview and reached GA with this update. Instead of displaying a list of fields with their values, you write a sentence that directly references those fields.

The result is a tooltip that reads as if a person wrote it, not something generated by a table dump. Useful for dashboards where the audience is not technical and needs context, not raw data.

Scatter Chart auto-fit markers

Anyone who works with scatter charts knows the problem: bubbles or markers positioned near the edges appear clipped. The fix has always been to manually adjust the axis range or build dynamic measures.

Auto-fit Markers handles this automatically. Power BI adjusts the chart area to ensure every marker appears fully visible. Simple, unobtrusive, and most people will never even notice they had the problem before.

What to do now

If you use Power BI regularly, the June 2026 update is worth installing. The Date Picker for Slicers and the axis improvements alone justify it.

For anyone building semantic models, Copilot in Web Modeling is worth testing even in Preview. It will still make mistakes, but this is the first time AI has entered the technical side of modeling in a genuinely useful way.

And for data managers and solution architects, Fabric Apps for Semantic Models is the feature to watch closely over the coming months. If the promise holds, it fundamentally changes how business teams consume data without depending on the technical team for every deliverable.