Power BI Update - November 2025

0:00 (bright music) (air whooshing)- Wanna know what's new for Power BI?

0:06 This one's special because it drops right alongside Microsoft Ignite.

0:10 In this video, we're diving into what's new,

0:12 what's generally available, and what's now in public preview.

0:16 For all the details, you'll see a QR code on the screen.

0:19 Scan it to jump to the monthly update blog.

0:22 I'm Adam Saxton.

0:23 This is the Microsoft Power BI update for November 2025.

0:27 Let's begin.

0:28 November is packed with Ignite-level releases that make Power BI more connected,

0:33 secure, and insight-driven than ever.

0:35 We're seeing massive strides in Copilot integration, Direct Lake reliability,

0:41 and governance enhancements powered by Microsoft Fabric.

0:44 These updates focus on performance, scalability, collaboration,

0:47 making Power BI the hub

0:49 for your business intelligence in your Fabric environment.

0:53 These month's previews are all about expanding how you build and how you share.

0:57 We're talking new authoring capabilities directly in the browser,

1:01 smarter data model management, and some early Copilot experiences that bring

1:06 generative AI directly into your Power BI workflow.

1:10 These features are your chance to shape what's coming next.

1:13 So try them out and tell us what you think.

1:15 Enough of all my talking.

1:16 Let's see it in action.

1:18 Here are the demos from the Power BI engineering

1:20 team showcasing the features that are changing how you build,

1:23 explore, and gain insights with your data.

1:26 Let's roll.

1:28 [Jacinda] In this video,

1:28 I'll walk through the new features and improvements we've made

1:30 to the new card visual as it is now generally available.

1:34 The first improvement I wanna show are the new defaults we've

1:36 updated for the card visual for a modern look and feel.

1:39 What you see here is the previous default.

1:41 But to use our new default, you can go to the View option in the ribbon,

1:45 go to Theme, and select the Power BI default theme.

1:49 And just like that, you are now using our new default.

1:52 All new card visuals that you create after this will

1:55 also be using the new default for a more modern look.

1:58 The second feature I wanna show is the new support for image hero.

2:01 Previously, you could only add an image in the Callout area like this.

2:07 But now, we've also added the support

2:09 for a standalone image hero under the Image card.

2:13 I'm gonna select the First Trail, turn on Image, and expand the Data pane.

2:17 Here, I also see that I have an additional option as well,

2:20 which is Select From Data.

2:21 So if I have a column of image URLs, I can drag that in to be displayed.

2:26 I also have different styling options as well,

2:29 such as fit, stretch, fill, and center.

2:31 And I also have different image effects, too,

2:33 that are built in, such as exposure, contrast, saturation, and blur.

2:37 We've also added new layout options, including a collage layout.

2:41 By going to the Layout group in the Cards card,

2:44 you can now select the Collage layout where the area

2:47 is split in half and one part is divided into two.

2:50 I can also choose which one is displayed in which section.

2:55 Now I can see that the image hero is taking half of the area

2:58 and the callout and the reference label are splitting the other half.

3:02 [Zoe] We're excited to announce the GA of Copilot

3:04 to write and explain DAX queries in Power BI.

3:07 So let's create a Power BI semantic model.

3:09 As announced last month, I can create Direct Lake semantic models now

3:12 on the Create page using the OneLake catalog.

3:15 And quickly add the tables and give it a name

3:18 and put it in the workspace that I want.

3:19 Now I'm in the web modeling experience, I can use DAX query view here.

3:23 And in DAX query view in the web, I have Copilot here to help me.

3:27 And now this Copilot is now GA.

3:29 So on the features, we have added these quick pull actions.

3:33 One is to suggest measures for your model,

3:35 so it can take a look at the model that you

3:37 have and it'll suggest some measures that may be helpful.

3:39 I can see these.

3:40 I can run this query, and can keep it when I'm ready.

3:43 Not only that, but in DAX query view,

3:45 once you've created measures in a DAX query,

3:48 you can actually add these to the model by either using the text above each one

3:52 of them or using that button at the top

3:55 to update the model with all of the changes.

3:57 I can also have it explain DAX query.

4:00 So I have a quick action here that I can utilize to quickly bring up an example.

4:05 Or I can ask questions of my own, like,

4:07 "What is the difference between SUM and SUMX?" And what's cool,

4:09 it'll give you examples from the model you're in.

4:12 Now it's not only available in the web, it's also available in Power BI Desktop.

4:16 So let's go ahead and live edit this semantic model in Power BI Desktop.

4:20 One of the other things, of course,

4:21 you can do is you can just ask it for any query

4:24 you want and it can generate that here in DAX query view.

4:27 Again, you can run it and keep it.

4:29 But now I noticed that I actually don't want

4:32 the total quantities delivered here to compare to my target.

4:34 I need to know the delivery count.

4:37 And so I can quickly select that, add that change,

4:40 see everything that changed in that nice diff view,

4:43 run it, and keep this query when I'm ready.

4:47 [Rui] The Power BI modeling MCP server brings

4:49 Power BI semantic modeling to your AI agents.

4:52 The easiest way to get started is opening Visual

4:55 Studio Code side by side with Power BI Desktop, open the GitHub Copilot chat,

5:00 and ask the agent to connect to the Power BI Desktop file I have opened.

5:04 So in my case, I have a file named Sales, so I can just type,

5:09 "Connect to Power BI Desktop Sales," and the agent will connect to the semantic

5:13 model running in the local Analysis Services instance of Power BI Desktop.

5:17 And when the connection is successful, I can now use natural language and ask

5:21 the agent to do any semantic modeling change.

5:24 For example, let's start asking the agent

5:27 to analyze the current naming convention of my model and make sure that it

5:32 will follow a consistent and business-friendly one.

5:35 And we can see the agent going through the metadata of the model,

5:39 reason about the current naming, and also proposing some changes.

5:42 Let's tell the agent to go ahead but also set some personal rules,

5:46 such as counting measures should include the number sign,

5:50 measures should be uppercase,

5:52 and no underscores or technical terms should be used.

5:57 And we can see the agent applying these changes

6:00 to the model in Power BI Desktop in real time, just like an external tool would.

6:04 And just like that, I was able to deeply analyze my model names,

6:08 identify inconsistencies,

6:10 and bulk rename everything in seconds using my own custom rules.

6:15 This MCP server is also capable of connecting to a semantic model in Fabric.

6:19 All we need to do is ask the agent

6:22 to connect with a workspace and the semantic model name.

6:25 And the agent will connect to the semantic model in the Fabric workspace.

6:29 And of course, because it's in Fabric, it will prompt you to authenticate.

6:33 And when the connection is done, you can ask the agent to perform any modeling

6:37 change just like you do in Power BI Desktop.

6:40 And similar to what we did before,

6:41 let's open the semantic model in the web side by side with Visual Studio Code

6:45 and ask the GitHub Copilot to create new

6:48 measures following an existing pattern of already existing measures.

6:52 Notice that I don't even need to be exact with the measure names.

6:55 The agent will now analyze the DAX code of the existing 12-month rolling

7:00 average and it will create the two new variations following the same pattern.

7:04 If I refresh the browser, the two new measures will be there.

7:07 And I can switch to Edit mode and confirm that they

7:11 are following the same DAX code pattern of the 12-month rolling average.

7:15 In the previous example, we worked against live Analysis Services databases,

7:19 but the MCP server is also capable

7:21 of working against offline Power BI projects files.

7:25 So in this case, I have my PBIP in here

7:27 and I have my semantic model represented as a TMDL files.

7:31 So what I can do now is I can ask the agent to load or connect to this PBIP.

7:37 This time, we are going

7:38 to use and inbuilt prompt called ConnectToPowerBIProject.

7:43 And we just need to provide the PBIP path into the parameter.

7:47 And notice that as you integrate a prompt for me instructing

7:49 the agent to load the PBIP folder but it does also going

7:53 to include the resource file that is in the MCP server

7:57 that is going to teach the agent how to work with the PBIP,

8:00 where to read, where to write, and some basic instructions on, for example,

8:04 what to do when you wanna create a new semantic model from scratch.

8:08 So let's run this.

8:09 The agent is going to load the TMDL files from the PBIP

8:13 folder and load that into a offline Tabular Object Model instance.

8:17 And just like before, I can ask to perform any kind of modeling changes.

8:21 But for example, let's ask the agent to set descriptions on all my measures.

8:26 The agent will go through each measure and analyze

8:29 its DAX code and generate business-friendly descriptions for each one.

8:34 And when it's done, it will save the TMDL back to the PBIP.

8:37 And because I'm using Git,

8:38 I can also easily review all the changes that the agent did with TMDL this.

8:44 [Stephen] The remote model context protocol server

8:46 for Power BI include a set of agentic

8:49 tools that enables you to chat with your data

8:51 from your chosen MCP client or agent.

8:53 Imagine I'm a data analyst at Zava Smart Materials and I

8:57 want to learn which of my materials my customers prefer.

9:00 Here, I've loaded my semantic model ID into the context of the agent.

9:03 Make sure I'm in Agent mode.

9:05 And then also choose my LLM provider.

9:07 I'll choose Sonnet 4.5.

9:09 And now I'm ready to ask my question

9:11 about which materials have the best feedback.

9:13 First, the agent retrieves the semantic model schema

9:15 to understand the context of the semantic model.

9:18 So we'll see that it passes in the artifact

9:20 ID or the semantic model ID into the tool,

9:22 and then it retrieves the schema structure as the output of that tool.

9:26 Next, it calls a generate query tool,

9:29 which leverages the same DAX query generation logic

9:32 as the Power BI Copilot in the standalone experience.

9:35 The agent formats the input and waits

9:37 for the DAX query to be returned from Power BI.

9:39 And now that the agent has the optimized DAX query,

9:42 it is ready to invoke the execute query tool

9:45 to get the data and then summarize the results for us.

9:48 So here, we see our top five materials

9:51 by feedback all broken down with the sentiment score,

9:54 the praise rate, and the complaint rate.

9:56 [Tori] The standalone Copilot in Power BI is

9:58 your go-to space to ask questions about any report,

10:01 model, or data agent you have access to.

10:03 Unlike the Copilot pane,

10:04 which only lets you ask questions about the report you currently have open,

10:07 the standalone Copilot lets you start broad

10:09 and ask without specifying what data to use

10:12 and will start looking for the right data to provide you with an answer.

10:15 Copilot searches for items based on their content,

10:17 how recently you used them, and more.

10:19 Since the topic of my question was clear,

10:21 Copilot was able to find the right artifact and tell

10:24 me that the social media campaign was the most successful.

10:27 Alternatively, if your question is less specific,

10:29 like this one about travel insights,

10:31 Copilot will work with you to narrow down the topic before generating an answer.

10:35 In the background, it found some reports about tourism,

10:37 airports, and hotel bookings,

10:39 so it's asking a clarifying question to zero in on the right one.

10:42 Once it's gotten the info it needs from me,

10:44 it confidently selects the tourism report

10:47 that contained a verified answer matching my question.

10:49 If we make another request related to tourism in Hawaii,

10:52 Copilot will intuitively know to use the same report.

10:55 But if you know a specific report you'd like to ask about,

10:57 you can easily attach it.

10:59 Our updated menu has all item types in one

11:01 list organized by recency to speed things up.

11:03 I wanna use our data about climbs on Mount Rainier,

11:06 so I'll just select it here and ask my question.

11:08 Insights are always within reach.

11:11 [Michal] We are excited to introduce

11:12 the standalone Copilot in Power BI mobile apps.

11:15 To start using Copilot, just open the Power BI mobile app and look

11:19 for the Copilot button on the bottom navigation.

11:22 Tap the Copilot button in the app and ask a question,

11:24 such as, "What is total spending by island in tourism?" Copilot will

11:29 find a list of relevant items to select from to provide an answer.

11:32 In the list, I can see that the report I want to use,

11:35 "Tourism Analysis Overview," is also matched as a verified answer.

11:39 I selected the item using the suggestion pill.

11:42 Copilot comes back with an answer.

11:44 In this case, it provides both a textual answer and a visual.

11:48 To take a closer look on the data, I tap on the visual.

11:51 This will open a drawer with the live

11:53 version of the visual that can be interacted with.

11:55 I can also share this visual as a screenshot with colleagues via chat or email.

11:59 Every response includes an option to copy the response

12:02 to clipboard so it can be easily paste and share.

12:05 Next, I will ask Copilot for a summary focused on spending analysis.

12:09 Copilot generates a summary based on the report,

12:12 including citation that opens a drawer displaying

12:15 the relevant visuals in an interactive graphical view,

12:18 making it easier to interpret the data.

12:20 From the drawer, I can open the visual

12:23 spotlighted directly in the report for deeper analysis.

12:25 Using the back button will take me directly back to the chat.

12:29 When using the Power BI mobile app from an iPhone or iPad,

12:32 it is also possible to dictate a question using the mic in the chat box.

12:37 For example, "Show me humidity trend for climb on Rainier."

12:41 Copilot will find a list of relevant item to select from.

12:44 After selecting the first option, I notice that Copilot output looks different.

12:48 Copilot displays a warning to review the answer carefully since this data

12:52 hasn't been marked as prepped for use with AI by the model author.

12:56 I can still view the answer by tapping

12:58 on View Answer to explore and interact with it.

13:01 Now I want to use Copilot to answer a question

13:04 from a data agent I have in my organization.

13:06 First, I ask Copilot to find it by saying,

13:09 "Find movies analysis data agent." I can see that the data

13:13 agent I need is numbered two in the list, so I ask, "Use number two.

13:18 Which five movies have earned the highest gross revenue?"

13:21 The agent responds right within the Power BI mobile app,

13:24 bringing Fabric data and Power BI together in one seamless mobile experience.

13:30 [Jacinda] In this video,

13:30 I'm gonna show you the visual enhancements we've made to our image visual.

13:34 We've also added an entry point in the Build

13:36 pane that's accessible through Surface and Desktop.

13:39 When I click on it, I see a placeholder.

13:41 And I'm going to expand the Style card and the Image group.

13:44 Under the Image Source dropdown, I see that I have additional options now,

13:47 including Enter URL and Select From Data.

13:50 I'm gonna choose Select From Data.

13:52 And I can drag in a column of image URLs into the field well.

13:55 I can also make this interactive by adding a button slicer.

13:59 So depending on the value that I select, the image will change.

14:03 For example, when I click on Germany, I see an image of the Alps.

14:06 When I click on USA, I see an image of Mount Rainier.

14:08 Now back to just the image visual, I also have additional customization options.

14:13 I can choose the different image fits, such as stretch and fill.

14:17 I can also make it more transparent as well.

14:20 And I have additional image effects option,

14:22 including exposure, contrast, saturation, and blur.

14:27 I can also add a shape and border around my image as well.

14:31 I can adjust the roundness of the border, too.

14:34 I can also add a background behind my image.

14:38 For example, I can add this light-blue background.

14:41 There's also one more addition that we've added

14:43 which is the different states that you can customize.

14:46 We have default, hover, and pressed.

14:48 And depending on which state you select, you can add different customizations.

14:51 For example, if I select Hover, I can even select a different image source.

14:56 For example, if I wanted the Alps to always show when it's hover,

15:00 I can go ahead and paste an image URL.

15:02 And when I hover over the image visual, it will show a picture of the Alps.

15:08 [Jeroen] Ever build a matrix in Power BI

15:10 only to find your layout looking cramped and unreadable?

15:13 That's where the new Grow To Fit feature comes in.

15:16 Here's a matrix visual showing sales by region and product category.

15:19 Notice how there's all this white space to the right that goes unused.

15:22 Let's fix that.

15:23 Select the matrix, go the Format pane, expand Column Headers, Options,

15:29 make sure the Auto-size Width is on, and set this to Grow To Fit.

15:33 Instantly, the matrix adjusts and the column

15:36 widths expand to fill the available space.

15:40 [Zoe] Here I have my Power BI report that I've published to the web.

15:43 I can go through each page one by one to see how they're loading,

15:47 but what I really want is a performance analyzer from Power BI Desktop,

15:51 so we've added it in.

15:52 To access Performance Analyzer, first you must be editing the report in the web.

15:57 Once editing, select View and choose Performance Analyzer.

16:01 This will open up the same Performance Analyzer

16:04 pane that you see in Power BI desktop.

16:06 And just like in Desktop, click Start Recording,

16:09 then Refresh Visuals to see all the load times for the visuals on this page.

16:14 I can expand each one and see a breakdown of each

16:17 duration and access a DAX query that loaded data into this visual.

16:22 [Carly] We're excited to share both authoring updates and filtering updates

16:25 for verified answers that lead to better exploration for business users.

16:28 Let's first take a look at a report that I used

16:30 that relies heavily on both slicers as well as field parameters.

16:33 Previously, when you created verified answers,

16:36 they would inherit any filters that were applied on the visual,

16:38 but not other things, like field parameters,

16:41 slicer selections, or report interactions,

16:43 like cross-filtering and cross-highlighting.

16:45 With this new update,

16:46 you can see here that when I select slice values or change things,

16:49 like which measure is displayed via my field parameters,

16:52 I can actually inherit those states into the verified answer definition and I

16:56 can see them here reflected in the Applied To This Visual section.

16:59 From here, I can do the rest of the setup as normal,

17:02 creating my trigger phrases,

17:03 either by selecting the suggestions or writing in custom prompts,

17:07 and of course select any filter permutations in additions

17:10 to the ones that are already applied there.

17:11 Now let's take a look at another verified answer that I wanna set up.

17:14 I'll add my normal trigger phrases using

17:17 terminology that I know my users leverage,

17:19 like funnel analysis and things that relate to campaign funnel health.

17:23 And then, of course, I'll go and set up my filter permutations.

17:26 Now one thing you'll notice here is that there are several choices here.

17:29 Previously, I could only select three, but now I can select up to 10,

17:33 which drastically increases the flexibility and exploration that I

17:37 have and can enable for my end users.

17:39 This is what it looks like for those end

17:41 users when they go to do their analysis.

17:42 So I'm gonna go ahead and say,

17:44 "Show me the funnel health for my B2B and preferred customers who use Bing." Now

17:48 this is a combination of a couple different

17:50 filters that you saw me set up there,

17:51 but Copilot is gonna do its magic and go ahead and apply

17:54 those accurately with two filters applied

17:56 to customer type and one to traffic source.

17:59 Wow, those were some amazing demos.

18:01 What are you most excited about?

18:03 What else do you wanna see?

18:04 Drop 'em in the comments below.

18:05 We'd love to hear what you think.

18:07 For all the details, scan that QR code on the screen or hit the link

18:10 in the description below to check out the monthly update blog.

18:14 And don't forget to join the Microsoft Fabric Community to connect,

18:18 learn, and share your own Fabric journey and Power BI journey.

18:22 Thanks for watching, keep being awesome, and we'll see you in the next video.