The Power BI June 2026 update (version 2.155.756.0) landed on June 9 and brought two genuinely significant changes for data professionals: DAX User-Defined Functions are now generally available, and a new Desktop Bridge puts AI agents inside a live Power BI session for the first time.

If you spend time building reports and semantic models, both of these matter more than the usual monthly changelog.

DAX UDFs: The Feature the Community Has Wanted for Years

DAX User-Defined Functions let you define a typed, reusable function once and call it across measures, calculated columns, and visual calculations anywhere in your model.

Before UDFs, if you had complex logic you needed in ten places, you had to write that logic ten times, or build workarounds using measure branching or disconnected lookup tables. It worked, but it produced models that were hard to maintain and harder to hand over to someone else.

With UDFs, you write the logic once with a proper function signature, and reference it everywhere. The June update also ships optional parameters — you can define default expressions for parameters, so callers can omit arguments they don’t need. This makes UDFs substantially more flexible than the initial preview.

For teams building enterprise-grade Power BI models, this changes the calculus on what belongs in the model versus a supporting layer. Complex date-intelligence patterns, custom aggregation logic, and organisation-specific business rules can now live as named, versioned functions rather than scattered formula fragments.

UDFs are on by default in Power BI Desktop from this release forward.

AI Agents That Edit Reports Live

The second major change is the Desktop Bridge, currently in preview. It opens a live connection between a running Power BI Desktop session and an external AI agent or developer tool.

The practical effect: an agent can submit a change to your report, the Desktop reloads, the agent checks the result, and the cycle repeats — without anyone sitting in a terminal manually refreshing. For anyone experimenting with AI-assisted report generation, this closes the loop that previously required human intervention at each iteration.

Combined with the Report Authoring Agent Skills announced at Build 2026, which let agents handle the full workflow from model design to published report, the Desktop Bridge adds an interactive feedback layer that makes automated iteration realistic rather than theoretical.

What Else Shipped

The June update rounds out with a few other notable items:

Copilot in Web Modelling is now in preview. You can ask Copilot to rename tables and columns, create relationships, and generate DAX measures directly in the Power BI Service, without opening Desktop. Useful for quick fixes and for teams that manage models collaboratively in the browser.

Fabric IQ Expansion extends data answering capabilities across Fabric workloads and into Microsoft 365 Copilot surfaces, meaning users can query their Power BI data from inside Teams and Outlook.

Shape Map visuals are now generally available, moving out of preview after a long incubation period.

What This Means for Business

For organisations running Power BI as their primary analytics platform, June 2026 is a meaningful release in two ways.

The near-term payoff is DAX UDFs. If your team has been managing complex models built on repeated measure logic, this is the right moment to refactor. Models with UDFs are easier to audit, easier to document, and easier to onboard new analysts onto. The time investment pays back quickly on any model of reasonable complexity.

The longer-term signal is where Microsoft is taking the agentic layer. Agent Skills at Build, Desktop Bridge in June, Copilot in web modelling alongside it: the direction is clear. AI agents are moving from generating content about your data to directly modifying how your data models are built. Report authoring, semantic model design, and schema management are all on the path.

For data teams, that means the work shifts toward model governance and business logic definition — the decisions about what the data should mean — while agents handle more of the mechanical construction. That is a better use of skilled analyst time, but it requires the underlying model to be governed well enough that agents have reliable ground to work on.

Enterprise DNA covers Power BI and Microsoft Fabric in depth. If your team is working through what DAX UDFs mean for your model architecture, the Power BI learning track is the fastest way to build the foundation.