Skip to main content
The assistant on the right side of the model workspace receives the current workspace, model, active module, selected object, cube, and model quality summary. It can explain the model, suggest queries, and propose draft changes when editing is enabled.

​
Steps

  1. Open the model workspace and wait for the draft status to show Synchronized.
  2. Open the assistant and ask about the current cube, measures, or quality issues.
  3. Select an object in the data source browser, Dimensions, Cube, or Calculations before asking a question. This gives the assistant narrower context.
  4. When requesting a change, state the goal and constraints, such as “Add a product dimension to the current cube using product_id as the relationship.”
  5. Review the proposed change path, quality summary, and affected objects before applying it.
  6. Wait for the draft to synchronize, then run Model quality and a query probe again.

​
Boundaries and safety

The assistant changes the draft and does not bypass publishing requirements. Publishing still requires model quality checks and an explicit Publish action. Do not paste credentials, tokens, or sensitive data into the conversation. The assistant cannot replace data source permission checks; return to the data source browser or Run and cache when a query fails. The assistant is also available in XMLA workspaces. Missing remote bindings must be fixed by refreshing structure information and selecting the correct system name.