An arrow on the screen solves “where do I click”; it doesn’t solve “should this happen”

To me this feels like it takes away from what the human is supposed to do (read, understand the consequences of the action, then.. consent or abort)

There is a reason your AI Agent won't automate these clicks for you