Physical Intelligence is an artificial-intelligence company whose stated mission is “bringing general-purpose AI into the physical world” by developing learning algorithms that can “control any robot to do any task.” Rather than building robots, it builds the software brain - robot foundation models in the vision-language-action (VLA) mold - meant to run across many different robot bodies. Its team is drawn heavily from the academic robot-learning community, including researchers such as Sergey Levine and Chelsea Finn who were behind much of the foundational VLA and imitation-learning work.
The company’s flagship is the pi (pi-zero) model family. pi0, announced in October 2024, pairs a pre-trained vision-language model with a flow-matching action head to drive eight different robot platforms on tasks like folding laundry, bussing tables, and bagging groceries. Physical Intelligence released pi0’s weights and code openly in February 2025, and has since iterated with follow-on models the company describes as steerable robotic foundation models showing step-changes in generalization.
Physical Intelligence has been backed by investors including Bond, Khosla Ventures, Sequoia Capital, and OpenAI, making it one of the most prominent venture-funded bets that the foundation-model approach which reshaped language and vision will do the same for physical robots.