That would be so cool if there was an AI inside your laptop that you could grant keyboard/mouse control to and that you could communicate with verbally, and it would understand what you were asking it to do in context of what was currently on the screen. Basically like you're always screen sharing with your AI helper. Since it would have access to everything this would probably only be workable if you were running locally or self hosting. In any case I wouldn't be surprised if UX converges on something like this in the near(ish) future