DAX calculations and dataflows for AI models

Hi community, 

I would please need some advice if I do the scenario below in the most efficient way whilst staying the realm of the power platform.

Scenario:

- I have some base data and do some relatvely complicated calculations in order to calculate some features per record derived from these base data

- At the moment I use DAX to produce a dataset  and read the features with a scheduled cloud flow and write them into a Dataverse table

The reason why I do this is based on the assumption that need ot have the data in my dataflow to feed the AI model. 

This all a little bit cumbersome as as the calculated data exists, but I have to work around to get them into a flow. Furthermore I have to some update schedule instead of just pulling the data as needed.

My understanding is:

- dataflow is mostly ETL: if the calculations are to complicated M is not suitable as it is ETL and not data modeling DAX

- datamart would be nice, but the SQL capability is only for ad hoc queries and cannot be used to feed this queries into something else (at least the queries could not be used in a report last time I tried)

So, is there a smarter way to do this, and/or is my above understanding wrong? As I said no way that these calculations can be efficiently done with Power Query.

Thank you for your help. 

Best regards 

Christian