Your robot has never seen the real world. We'll show it, frame by frame.
We capture skilled human work in Asia's homes and factories and turn it into robot-ready data, graded frame by frame.
CHECKS–
Real footage and real numbers: hand keypoints from our pipeline, lighting, shake, hands and contact measured on every frame, step labels from the vision-language model.
Robots won't be limited by their brains. They'll be limited by what they've seen, and how much of it they can trust.
Your robot works in the lab and fails in the home.
A perfect world
Clean physics, tidy objects, nothing out of place. Homes don't look like that.
A controlled world
One room, one set of hands, one way of doing the task.
A raw world
Video you have to clean, check and mostly throw away before it trains anything.
Robot-ready, not raw.
No guessing. No cleanup. Every frame comes with its report card.
Give us one task your robot struggles with.
Name the task
The one your robot keeps failing. Tell us your cameras, labels and format.
10+ real homes
We bring back graded, robot-ready data captured in more than ten real homes.
Then we scale
If it moves your model, we scale the collection with you.
What does your robot need to see?
Tell us the task, the setting and your capture spec. We'll come back with coverage, timing and a pilot plan.
Want to teach robots instead? Become a partner →
Simulation shows your robot a perfect world. We show it the real one, and prove it's real.
The robots of the future will learn from someone. Let them learn from you.
Big tech took the internet's knowledge for free.
The people whose skill teaches the robots are paid, credited and protected.
Pick your side of the classroom.
The world's largest classroom for robots: every home, every factory and every skilled hand in the majority world.
Every robot will learn from someone.
We make sure it's everyone, and that everyone gets paid.