Intelligence that can act in the world.
We are building world thinking machines that learn how environments change, predict what could happen next, and choose actions with context.
Our work starts with a practical question: how can a machine move from understanding the world to acting in it?
01
world thinking machines #
Learning what can happen next.
Our models learn a representation of the world before they act. They use it to understand a scene, compare possible actions, and choose the next step with more context.
02
Understanding the world #
Predicting what an action will change.
A robot needs more than a picture of the world. It needs a model of how objects, spaces, and other parts of an environment change when an action takes place.
03
World Action Model #
A predictive model for action and control.
World Action Models are our research direction for learning how actions change the world. The goal is to help robots compare outcomes, plan across longer horizons, and act more reliably in changing environments.
This work is still under development. Read more in our research notes .
04
JEPA #
Predicting meaning instead of every pixel.
Joint Embedding Predictive Architectures let a model focus on the parts of an observation that matter for understanding and action. We use this family of ideas as a foundation for models that learn useful representations before they predict outcomes.
read the JEPA paper05
H-JEPA #
Planning across longer horizons.
H-JEPA extends predictive learning across time. It gives a model a way to think about several possible futures, so a robot can organize a sequence of actions instead of reacting to one frame at a time.
read the H-JEPA paper06
Our goals #
Prediction, control, and recovery.
We want our systems to predict the result of an action before it happens, notice when a plan is failing, and recover without starting over. The larger goal is reliable behavior across tasks that take time and require several decisions.
07
Project Monarch #
Our earlier model.
Project Monarch was the internal name for an earlier model and research direction. Its work on representation, prediction, and longer tasks helped lead us toward World Action Models for robotics and automation.
read about Project Monarch08
The timeline #
From simulation to capable machines.
We are starting with simulated environments where we can test prediction and control carefully. From there, we will study longer tasks, broader environments, and the steps needed to move these systems toward real robots.
FAQ