The next frontier of AI is the physical world.
We extend foundation models downward into reality;
an artificial general scientist, engineer, and
inventor that learns through interaction.

From Generative AI to
Creative AGI

Our mission

Every solution paves the way for solving what comes next.

We build self-improving world models. The LLM already holds a vast ontology of the world and externalizes it by writing programs, because a program is a model. It writes an executable model of a domain, proposes minimal experiments to resolve what's uncertain, and tests its predictions against reality.