COCO Matrix

COEXIST · COWORK · COLLABORATE · COEVOLVE

Building Physical AGI in real-world.

About COCO Matrix

COCO Matrix is founded by a team passionate about ROBOTS and AI.

We want to bring something exciting to physical AI: instead of enumerating the vast physical world in advance, robots should adapt and learn as they do.

In-Context Learning (ICL) — that's the direction COCO Matrix bets on. We turn the real world into rich "CONTEXT," where robots can learn more immersively, more efficiently, and more intuitively. This paradigm shift profoundly excites and motivates us.

We want to redefine the learning approach of our robot models. With strong world representation learning, we empower our models with better perception, richer diversity, and longer memory. They are not just identifying objects, but gaining a deeper understanding of humans, environments, actions, and consequences. They are not just seeing this exact moment, but turning past demonstrations, corrections, and experience into the "CONTEXT" that guides the solution for future problems.

On top of that, our robot models can leverage the entire ecosystem. The model paradigm decides what data to collect, what forms of interaction to build upon, and what kind of robot morphology we invest in. Data, hardware, and models are no longer individual concerns — they co-evolve as our robots' learning capacity expands.

Eventually, learning becomes a natural daily routine. Anyone can be a robot's teacher, and the robot becomes your best partner. Let's leap into the AGI future and CO-EVOLVE with robots.

Robots for all, shaped by all.

Four Technical Dimensions of Real-World Robot Deployment Four capabilities flow into the central robot — world model representation, in-context learning, learning from human, and human-robot collaboration — culminating in real-world deployment. World Model Representation In-Context Learning Learn from Human Human-Robot Collaboration REAL-WORLD DEPLOYMENT

Join Us

We're hiring across data, model training and product. See open roles and apply directly on our Feishu recruiting page.

See Open Roles