Every clip we train on is cleared.
Our models learn from human demonstration, which means real people doing real work in real places. That imposes obligations. We license and capture data on terms the people in it agreed to, we can show provenance for what we use, and we treat a defensible corpus as a product requirement rather than a legal afterthought.
Consent, recorded
Every operator who appears in a capture program signs a release covering AI training use. We do not train on footage scraped from the open web, and we do not accept libraries that cannot show consent.
Provenance, traceable
Each clip carries its source, rights terms and capture conditions through the pipeline. We can tell you what a skill was trained on and under what licence.
Creators, paid
We license libraries and commission capture at commercial rates. People who record for us are compensated for the work, and owners retain their underlying rights.
Bystanders, protected
Capture in shared spaces follows a protocol for incidental subjects: signage, avoidance of identifying detail where it is not needed, and removal on request.
Sites, confidential
Data captured in your buildings is used for your policies. It is not folded into a general corpus or resold without an explicit agreement.
Scope, deliberate
We do not build skills for surveillance of individuals, enforcement, or weapons applications, and we decline capture programs aimed at those uses.
Hold a library and want to know if it qualifies? Our data intake asks about rights and consent status up front, because that is the first thing that determines whether a library is usable.
See the data page →Build the business case before you buy a robot.
Our Deployment Impact Assessment baselines the labor in scope, quantifies what automation is worth on your floor, and returns a costed plan your finance team can underwrite.