The demos look cool until the lighting changes or the table is two inches further left. Then they fall apart. That’s not a hardware problem. That’s a data problem. Robots have almost no real experience to learn from. Language models got the whole internet. Robots got a few hundred thousand hours total. @axisrobotics just scaled that to hundreds of thousands of regular people recording tasks from a browser. @verona_dev makes every one of those recordings verifiable the second it’s created, so companies can actually trust and reuse the data instead of re-auditing it every time. If this works, the robots that end up in kitchens and factories won’t mostly be trained by research labs. They’ll be trained by normal people, and the data will have receipts. That’s a different timeline than the one we’ve been watching.... Are you watching too?....