AI-generated descriptions for recorded steps are functioning correctly and are contextually relevant to the recording.

However, the one of our customers reported that the descriptions are too generic, they describe the general area/outcome of an action rather than the specific UI interaction performed (e.g., "select from the dropdown," "enter value in domain field," "move cursor to tenant name field").

Customer used AI-generated recordings a few months ago and found the descriptions more granular/action-specific at that time.

It would be great if the descriptions should narrate the precise mouse/UI action being performed at each step, not just describe the resulting screen state.

Requesting that the AI content generation logic be enhanced to reflect the specific user action taken (click, select, type, drag, etc.) at each recorded step, rather than a general description of the resulting UI state.