Narrate first-person task footage step by step so robots can learn to understand it.
Robots can already move. What they can't do is understand what they're looking at, or break a task into the small physical steps a person does without thinking. Nobody writes that down. It lives in your hands.
So we pay people to say it out loud.
You get a piece of first-person footage. Someone's hands doing something real: tying a bow, wiring a socket, washing a plate, sorting laundry. You narrate it step by step, in the order the hands actually move, and you describe what else is in frame as you go.
Think of how you'd talk someone through a task over the phone when they can't see what you're doing. Small steps, specifics, no shortcuts.
You control the video. Hands move faster than speech, so you pause whenever you need to, describe what's happening, then unpause and continue. There's no live take. You are not being timed.
Everything runs in your browser: open the clip, record, submit. We'll walk you through exactly what a good recording sounds like once you're in.
Past that, it's a noticing job more than a voice job. The people who do well here are the ones who catch that the screw is stripped, that the knot is going the wrong way, that the hand hesitated before it committed. If you've ever talked someone through something over the phone and found yourself saying "no, the other way, the small one, next to the blue thing" — that's the skill.