AI models are rapidly learning to reason, converse, and act in the world. But the moment AI takes physical form, something is missing. It cannot walk into a room and make you feel comfortable. It cannot look you in the eye when you are scared. It cannot step aside politely when you are in a hurry. The missing layer is behavioral: the acting itself — movements with meaning and purpose, made of the small physical choices that make someone present. The hesitations, the glances, the weight shifts that make a body feel real and alive.

We call it the Behavior Layer.

Imagine a hospital kiosk that makes eye contact at the right moment. A robot whose posture shifts when you step too close. Every embodied AI today has the same problem: people don't trust them because they move like machines. AI systems can generate language, images, and policies. They have never been able to generate behavior.

Building the Behavior Layer isn't just a model. It's a new substrate for embodied AI — real-time behavioral coordination across the body, body-agnostic retargeting across humanoids, robots, and avatars, real-time motion synthesis, character-grounded reinforcement learning, and the encoding of decades of craft from animation, theater, and film. Safety and ethics are central to how we build.

Our founding team has shipped AI systems at Google, Meta, and Nvidia — and spent careers making characters move across Pixar, Disney, Lucasfilm, DreamWorks, Sony, Telltale, EA, PullString, and Baobab. Between us, we have contributed to frontier AI research, Oscar-winning performances, and creative tools used by billions.