The defining question of this decade is how AI moves into the physical world. Trace exists to answer it – connecting robots to reality through operational rigor, engineering excellence, and frontier research.

People say robotics has no "internet" of data to learn from. They're wrong. It's the real world around us. Unlocking models' ability to learn from the physical world is the missing ingredient for general-purpose robotics. Transforming everyday activity into useful robot learning involves hard, new challenges across disciplines. Trace exists to solve them.