While using my phone I often get screens which clearly imply I need to create a new contact. Today, that's a lot of work clicking around, yet LLM+vision is already good enough to do it all for me, if the parts were connected.

The claim:

When will that ship on a Google Pixel 7 or later so I can use it? It can't just be for Gmail or a limited set of apps. It has to be raw, for any screenshot.

It only counts if it's in broad release, but if I have to pay I will. As long as most anyone with a new enough phone can do so.

No rules about exact mechanism of confirmation/disambiguation/etc. the point is, when will the manual steps for this be skippable.

Demo

Here is a demo of gpt4V doing a reasonable job already, first try, April 2024