You can now consume process insights from Power Automate Process Mining in Power BI reports you create in a Fabric workspace by using the DirectLake semantic model on top of Fabric Lakehouse.

Compared to the semantic model created in Power BI workspaces, the new DirectLake model:

Is simpler and more streamlined, so it removes unnecessary complexity.
Uses DirectLake mode, which works with a smaller SKU even for large process models.
Uses fewer resources because DirectLake mode loads to memory only the data a query needs.
Offers performance that matches the import mode semantic model.
Supports custom enhancements with composite models while keeping the DirectLake performance benefits (with some limitations for DirectLake composite models).
Includes custom metrics and view configuration that you create in the Power Automate Process Mining desktop app.

In the Create process wizard, select the location for the semantic model:

Power BI Embedded: Creates a lighter version of the default report in a shared workspace you provide. The semantic model uses aggregated values. Choose this option for small proof of concepts and trials.
Power BI workspace: Creates a full semantic model (older version with import mode) in your own Power BI workspace. Choose this option for production, small, and midsize use cases.
Fabric workspace (preview): Creates a full semantic model with DirectLake in your Fabric workspace. Choose this option for any production use case.