September 2025 - Fabric and Power BI GA features (part 2)

 This is part two in the series of all of the September announcements for both fabric and power bi. For those that haven't read it please go and have a look at part one on the following link.

This parts starts with the announcements around real-time intelligence. I'm not going to touch on the maps feature that was part of the keynotes, please go and see that separate blog for any of those announcements.

Fabric continued.

Azure monitor diagnostic logs integration via event stream

With this new feature, we can now bring metrics and logs from as your Azure sources into fabric event streams.

Creating a single location for advanced analytics on monitor diagnostic logs data.

Private link support for event stream at Workspace level

For a limited number of sources and destinations, we can now use private links at a Workspace level. If you need to use this level of security please do go and check the documentation.

External parameter support for dataflow Gen 2

Relatively small update, but allows you to access parameters from outside of a Gen 2 dataflow in order to set them for example within the UI or by the apis.

Incremental refresh to Lakehouse as a destination for dataflow Gen 2

This one is a relatively straight forward update, that allows us to right-click on a destination and set it as an increment refresh. Previously dataflows always assumed a full refresh.

Natural language to custom column and explain query with copilot in dataflow Gen 2

A couple of nice features and great use cases for CoPilot in my experience. I've always found it tricky to remember m code syntax on top of Dax, it's a great to see co-pilot take that mental load off of me rather than having to remember or Google it every time.

Fabric data factory pipelines are now pipelines

This one means just a simple rename. I kind of understand why Microsoft did this, but the new name is more confusing. Really they should have called it something like orchestration pipelines, that way it's distinct from deployment pipelines etc.

Evaluate expression experience in pipelines

A relatively small UI change, that will make it a lot easier to debug issues with expressions when dynamic variables are used. Now we have the ability to test an expression with a set of known values.

Functions activity with user data functions

Sounds fancy, but it isn't, ultimately all this means is you can call user data functions that exist in a specific workspace from pipelines.

Add up to 20 schedules for your pipeline

Whilst it's great that we can add more schedules to a pipeline, if you are getting this advanced, I would look at the native Apache airflow options. Whilst orchestrating via pipelines, is great for small tasks they definitely don't scale well - unlike the Apache alternative.

On premise and v-net data gateway support for invoke pipeline and semantic model refresh activities

For those with multi-tenant setups or on premise hybrid setups, this gives you the same functionality as was available within azure data factory. If this is applicable to you, do go and check out this month's release notes.

Invoke pipeline activity

At last the ability to call one pipeline from another has gone ga. It really shouldn't have taken this long, and this feature should have been part of MVP.

Workspace identity support in pipelines

Hidden way down at the bottom of these release notes, is another one of those features that should have happened sooner. Workspace identities can now be used to execute info pipeline semantic model and scope activity operations.

Variable library integration with pipelines

This one makes it a lot easier to reuse pipelines across multiple environments, by using a single value in a variable library and setting discrete values for Dev, test, and production for example.

This small change will make cicd easier across multiple different approaches, just remember the values are stored as plain text in your source control.

Azure Databricks jobs activity

With this feature added we can now call all relevant databricks functionality. Meaning that cross platform operations can be orchestrated using pipeline activity.

Apache airflow improvements

This month we've had a couple of updates. These are cicd support for airflow jobs, and a new user interface to make it easier to call notebooks.

Vs code extension

For those that have been waiting for the extension to go ga, this has now happened. Apparently some new features have been added, but those aren't detailed in this month's blog.

Power BI

Now we've gotten through the Fabric release notes, we turn our attention to the ga features in this month's PowerBI release notes.

Find power bi reports in m365

You can now use co-pilot to search for reports across the estate. This means that stakeholders do not have to go into the power bi service to find the report they are after. This is a setting controlled in the admin portal.

To do this it uses the report authors description, and if one doesn't exist CoPilot were automatically generate one for that report.

Performance analyzer in the Power BI service

Being able to use this desktop feature natively in edit mode in the service is one step closer to not needing PowerBI desktop.

Semantic model editing in the Power BI service

Creating an editing models in the service, instead of PowerBI desktop, is now in general availability.

Editing Direct like semantic models in power bi desktop

I know a number of developers have been frustrated at having to develop Direct Lake models in the service only, but this is no longer the case. As of today, those models can now be edited in power bi desktop.

TMDL view

Semantic models can now be developed using the tabular module definition language, allowing developers to move to a code first experience.

Now that this view is available, downloading semantic models edited through the Endpoint becomes possible for some models.

Best practice and memory analyzers

With the combination of fabric notebooks and semantic link, we now get these two analyzers available within the service. Allowing developers to see where they need to focus improvements.

Refresh data or schema options in power bi desktop

This UI change allows us to decide if we want to refresh just the schema, the data, or both. For those working on large datasets, this is going to be a big step forwards.

NFC tags and the mobile app

I always thought that the barcode option in the mobile app had a number of good use cases. For example, allowing those on the shop floor to look at stock levels by scanning the barcode of the item in question.

The barrier for that happening, was always getting the barcode data into the systems and the two linked. This feature removes that barrier. Now we can program an NFC tag from within the power bi app to be able to achieve the same thing.