Edit

Share via

Write DAX queries with Copilot

  • Article

You can use Copilot to write and explain Data Analysis Expressions (DAX) queries in DAX query view of a Power BI semantic model.

Enable Copilot

Check to make sure you meet the Fabric Copilot requirements.

Power BI Desktop

In Power BI Desktop, in the Preview features section of the Options, enable DAX query view with Copilot.

  1. Open Power BI Desktop.
  2. Open an existing Power BI report or connect to published semantic model as live edit or live connection. Alternatively, you can add data and create a new semantic model and report.
  3. Go to DAX query view, the fourth view available on the left side of Power BI Desktop.
  4. Create a new query tab.
  5. Open Copilot by clicking on the Copilot button or using CTRL + I.

Screenshot showing Power BI Desktop with Fabric Copilot in DAX query view.

Power BI service or workspace

  1. Choose a published semantic model in a workspace.
  2. From the context menu choose Write DAX queries. If Write DAX queries is greyed out, you may need to enable Users can edit data models in the Power BI service (preview) in the Power BI General section of Workspace settings.
  3. Create a new query tab.
  4. Open Copilot by clicking on the Copilot button or using CTRL + I.

Screenshot showing Power BI service with Fabric Copilot in DAX query view.

Note

DirectQuery and Direct Lake model authors can also use DAX query view. View the data in your tables whenever you want.

Use Copilot

Let’s look at what Copilot can help you with in DAX query view:

  • Natural language to DAX query. Just ask for what you want from your semantic model and have the DAX query generated by Copilot. You can also select an existing DAX query and have Copilot modify it.
  • DAX query to natural language. Ask Copilot to explain what any DAX query is doing.
  • Understand DAX concepts. Maybe you heard about a certain DAX function or need to find the right DAX function for the job. Copilot can help you here too.

To use Copilot to create a DAX query, type in a prompt to describe what DAX query you would like generated and select Send or press Enter. Alternatively, you can click the Write DAX query inspire button. To run what is returned, select Run or press F5 to see the results of the DAX query. To keep the query and exit Copilot, select Keep query to add it to query tab.

To use Copilot to explain a DAX query, select the DAX query and click the Explain this query inspire button. A short explanation is shown with the option to show more to see a longer explanation.

To use Copilot to explain a DAX topic, click the Explain a DAX topic inspire button, or type in a prompt to ask for an explanation of a DAX function or topic and select Send or press Enter. A short explanation is shown with the option to show more to see a longer explanation.

Features of the inline Copilot

1. Run the DAX query before you keep it

You can Run the DAX query then decide to Keep or Discard the DAX query.

Screenshot of running the DAX query before you save it.

2. Build the DAX query conversationally

After the initial DAX query is generated, you can simply adjust by typing in additional user prompts.

Screenshot of asking DAX to explain the query.

3. Syntax checks on the generated DAX query

The generated DAX query syntax is checked, and the prompt automatically retried once. If the retry is also invalid, the generated DAX query is returned with a note that there is an issue, giving you the option to rephrase your request or fix the generated DAX query:

"This query contains errors. Rephrase your request and try again or try fixing it yourself."

4. Inspire buttons to get you started with Copilot

Type in a prompt or select any of these buttons to see quickly what you can do with Copilot.

Screenshot of showing Copilot offers suggestions.

Sample prompts

These sample prompts can illustrate how to use Copilot to write DAX queries.

Write a DAX query

In a blank query tab, invoke Copilot with CTRL+I and enter this request:

"Write a DAX query to show profit by product"

Screenshot showing Copilot with the prompt Write a DAX query to show profit by product.

After a few moments I get a DAX query.

Screenshot showing DAX query generated by Fabric Copilot.

I can select run to see the results before I keep the query.

Screenshot showing Run DAX query generated by Fabric Copilot.

I can conversationally make updates to my initial request. Type in "Add in Orders" and hit enter and Copilot adds in the additional measure. I can again run the query to see the results.

Screenshot showing how to build the DAX query conversationally.

When I select keep query, the query is added to the query tab.

Add a column with the diff view

Let's now add a column to the existing DAX query. First I select the DAX query, then invoke Copilot. This time the selection is showing fitted to the text and the background shading highlights what is being sent with the request to Copilot. Type in this request, and hit enter:

"Add a column to show profit per order"

Screenshot showing a selection included with the Copilot prompt.

Now the diff styling in the DAX query editor shows exactly what Copilot has changed.

Screenshot showing diff styling to see what Copilot has changed.

I select run to see the results, and keep query to add it to the query editor.

Screenshot showing the query in the DAX query editor.

Explain a DAX query

Copilot can provide a detailed explanation of a DAX query. Again, select the entire DAX query and invoke Copilot. This time select the Explain this query button.

Screenshot showing Explain to me what this query does.

Now, instead of writing DAX queries, it’s responding in the inline Copilot.

Screenshot showing brief explanation of the query.

Copilot generates a few lines. I select Show more to see the full explanation.

Screenshot showing a long explanation of what the query does.

Now I understand each part and the DAX functions used.

Explain a DAX topic

Copilot can also explain DAX functions with more detail. Let's ask about SUMMARIZECOLUMNS. In a new query tab, invoke Copilot, and sent the following request:

"Explain what SUMMARIZECOLUMNS is for"

Screenshot showing an explanation of what the SUMMARIZECOLUMNS does.

Again, Copilot generates brief description:

Screenshot showing a brief description of the SUMMARIZECOLUMNS DAX function.

I can select Show more to see a more detailed explanation with an example.

Screenshot showing Copilot generates a much more in-depth description.

The example also uses columns in my model.

Fabric Copilot to help write DAX queries: Responsible AI FAQ

What is Fabric Copilot to help write DAX queries?

An inline Copilot found in Power BI DAX query view, available in the DAX query view of Power BI Desktop or write DAX queries in Power BI workspace, for model authors and consumers to request help creating DAX queries. The help can include explaining a DAX query and DAX topics, in addition to creating DAX queries themselves.

What can Fabric Copilot to help write DAX queries do?

The user request can be returned by Copilot as a DAX query the user can see, then run to see the data returned in the results grid. Additional requests can be made, and a new DAX query can be returned, creating a DAX query conversationally. If the request is about explaining a DAX query or topic, a natural language response is generated in the inline Copilot. DAX topics can include what a DAX function does or which DAX function is good for returning data in a specific way.

What is Fabric Copilot to help write DAX queries’ intended use?

Copilot’s intended use is to help create DAX queries on a specific semantic model the user is currently working with in Power BI.

How was Fabric Copilot to help write DAX queries evaluated? What metrics are used to measure performance?

DAX queries from multiple Power BI semantic models using a variety of user requests, categorized by data availability in the semantic model and/or if additional calculations were needed to create the correct response, and then graded for accuracy and performance by members of the product team. This included a semantic model adhering to best practices. Explanations from multiple DAX queries and DAX function topics, and then graded for accuracy and readability by members of the product team.

What are the limitations of Fabric Copilot to help write DAX queries? How can users minimize the impact of Fabric Copilot to help write DAX queries’ limitations when using the system?

  • To use Copilot to help write DAX queries, you need to select a workspace with a Fabric capacity.
  • Copilot uses the model metadata (tables, columns, measures), so DAX query responses will answer only in context of the current semantic model.
  • Copilot uses the model metadata (tables, columns, measures), so providing rich metadata that is descriptive and understandable will yield better results.
  • Copilot may use min/max to get sample values in import storage mode, but cannot use other data in the tables of the semantic model.
  • Copilot may not be available or use reduced version of the model metadata (tables, columns, measures) when the number of model metadata items exceeds 500 tables, 1,000 columns or 3,000 measures in a table, 2,000 relationships, 10,000 total columns, 5,000 total measures, 5,000 characters in a single DAX expression, or 256 characters in a name.
  • Copilot may use the description property on model metadata, but is limited to only the first 200 characters.
  • DAX functions added recently may not be known to the Copilot, and explanations may be incorrect.
  • Copilot requests include all text in the current query window, and if text is selected the selection information it is noted as selected, but all text in the query window is included.

What operational factors and settings allow for effective and responsible use of Fabric Copilot to help write DAX queries?

How do I provide feedback on Fabric Copilot to help write DAX queries?

  • Use the thumbs up and thumbs down icons in the inline Copilot to provide feedback.
  • Submit feedback using the Power BI support.

Learn more with these resources