The general idea is to have an AI system be aware of where or what you are pointing at, and then use the microphone on your computer to give simple commands without careful prompting.
This should allow for easier AI interactions like «show me directions» when looking at a building, or «book a table» while hovering over a restaurant.
This sort of pointer needs the OS or app to be context-aware, and there is no system like this yet. It is, however, rolling out in Gemini in Chrome «starting today,» and will roll out «soon» as Magic Pointer, a major feature of the freshly announced Googlebook.
It is also available to test in AI Studio, where you can use it to edit an image or find places on a map. Do note that the system uses your microphone.
Read more: Google’s blog, 9to5Google.