Origami Robotics

General manipulation in robotics has been the holy grail for decades—every robot that can stack blocks or pick up an Amazon package is doing something a human toddler does without thinking. Quanting Xie, Tongzhou Liao, and Yonatan Bisk looked at this problem and said "we're going to solve it." All three have PhDs from CMU, which is basically the Vatican for robotics research, and they just got backed by Y Combinator to do exactly that. They're based in Millbrae, CA, and they're building the infrastructure that lets robots manipulate anything—literally anything.

What does that mean in practice? Their research tackles the "Dexterity Deadlock"—the reason most robots are still clunky, fragile, and can't handle real-world objects that twist, slip, or require subtle force. They're going after the fundamental hardware and software layer that makes general manipulation possible. If they crack it, every warehouse robot, surgical bot, and home assistant gets exponentially better. This is the kind of deep tech that doesn't get headlines until it suddenly changes everything.

They're a team of PhDs and engineers from CMU who are genuinely passionate about physical AI—these aren't people who stumbled into robotics, they've been living in this world for years. They have a product waitlist going and they're looking for early partners who want to push the boundaries of what robots can actually do. If you're building something that needs hands that work like human hands, this is the team to watch.