Andrew Barry on Why Dexterity Is the Next Breakthrough in Physical AI

Physical AI development centers on achieving robotic dexterity, the most critical hurdle for creating truly useful, general-purpose machines. Generalist, led by CTO Andy Barry, addresses this challenge by training models from scratch on massive, diverse datasets captured via inexpensive, mass-manufactured handheld devices. This approach bypasses the limitations of traditional hard-coded controllers, enabling "improvisational intelligence" where robots successfully navigate novel, untrained scenarios—such as reorienting objects in mid-air. By prioritizing real-world tasks like cable threading and household manipulation, the company creates a data flywheel that improves performance through continuous, scalable interaction. Rather than relying on static internet-based data, this methodology leverages high-fidelity, physical-world experience to build foundation models capable of adapting to unstructured environments, ultimately bridging the gap between theoretical research and commercial viability in the robotics industry.