Third Eye AI

What this is

Third Eye AI is a specification for a conversational layer that lets someone navigate without looking at a screen: voice in, voice out, no menus to see. I wrote it as an engineering document first, the way I did with the Kit, because I would rather work out how something has to behave before I build it.

The idea underneath it

I build on the same idea everywhere: someone who cannot see a menu should not need one, so the interface is a voice that listens and a voice that answers.

The document targets a working prototype two to three weeks after work begins, and work has not begun. What exists right now is the document.