Physicl raises $10 million, launches Studio for physical AI data

Physicl's seed round arrives as physical AI developers confront a persistent gap: models that perform well in controlled demonstrations can stumble on unusual lighting, unfamiliar objects or unexpected physical arrangements. These edge cases are hard to capture comprehensively through conventional data collection, which is the problem Physicl's simulation-based approach is designed to address.

Studio lets researchers select or describe problematic scenarios, adjust variables such as lighting, camera angles and object placement, and then generate training data aimed at those weaknesses. SpAItial, which develops world models for explorable 3D environments, used Physicl-generated data to improve its model's ability to reconstruct real rooms, reducing reconstruction error rate by more than half, according to the companies.

The new funding is intended to support expansion of Physicl's team and further development of its synthetic data generation platform, as the company plans to make targeted physical AI training data more accessible to developers. Physicl says it already works with AI laboratories including Meta, World Labs, Epic Games and SpAItial, suggesting early traction among researchers building robotics systems and world models.