Cobilion is the marketplace where humans record physical tasks and get paid — and robotics companies buy the demonstrations to train their AI.
Be first to earn from your movements.
What is Cobilion
Robots learn by watching humans. Today, labs pay full-time employees to perform tasks in controlled environments — expensive, slow, and limited. Cobilion opens that market to anyone with a phone.
Anyone records everyday physical tasks using a phone or wearable — at home, at work, anywhere.
Curation, quality scoring, labeling, and marketplace layer.
Browse, filter, and license demonstrations by task type, environment, and diversity — at scale.
How it works
For Contributors
Use your phone or a clip-on camera to film everyday things — folding laundry, opening a box, cooking.
Our platform reviews, labels, and prices your demonstrations automatically. No technical knowledge needed.
Earn every time a robotics lab licenses your data. You keep 75% of every sale.
For Robotics Labs
Search by task type, environment, body type — or send custom requests directly to the contributor pool.
Browse a curated, quality-scored catalogue. Subscribe for unlimited access or buy individual datasets.
Diverse, real-world demonstrations that cut training time and improve generalization across environments.
Why now
Physical world training data is the next crowdsourced data business — and no open marketplace exists yet.
Built a business on crowdsourced data labeling. Cobilion is the same model applied to physical world demonstrations — harder to collect, higher switching costs, larger long-term market.
Well-funded labs are already hiring humans to generate demonstrations — but only for their own robots. No open marketplace exists yet. Cobilion is building it.
Chinese manufacturers are paying workers full-time to perform tasks for robot training at scale. The demand signal is real. The platform does not exist in the West.
We're building the platform that makes that possible at scale. Get early access — as a contributor, a lab, or an investor.
For contributors, labs, and investors.