Gimmick: Using Neuro's real-time vision capabilities, feed her a long driving video and get her to pretend to a taxi driver.
She can chat with us, and comment on stuff on the road. I would argue it isn't a react stream because the video is secondary to the neuro zatsudan.
Ngl the driving element sounds quite dull, but maybe we could brainstorm an evolution of this idea leads to interesting Neurotalk.
Perhaps the footage would end up being too distracting, or maybe this gimmick has no substance?