Investing in AMI

For the past decade, AI progress has largely been measured by how well systems can predict or generate language. Models that once struggled with basic pattern recognition now write code, synthesize research, and are embedded in workflows across industries.

But much of the real world does not behave like language. Real-world data is continuous, high-dimensional, and noisy. In dynamic environments shaped by sensors, motion, uncertainty, and constraint, reacting to historical data alone is not sufficient. Operating in these settings requires systems with memory, an understanding of cause and effect, and the ability to reason and plan as situations evolve. 

We believe this shift can define the next frontier of AI. That conviction led us to Advanced Machine Intelligence, or AMI. 

AMI is building intelligence centered on world models. These systems learn structured representations from continuous streams of real-world data, including video, sensors, and actions, to understand how environments change over time. Rather than focusing solely on output generation, this approach is designed to model how situations unfold and how actions shape outcomes. 

By grounding intelligence in world models, AMI aims to enable systems that can operate within real-world constraints while remaining controllable and safe. It builds on the past decade’s advances in representation learning and generative modeling and extends that progress beyond digital tasks into physical and dynamic environments. We expect applications in areas such as industrial systems, robotics, manufacturing, and other industries where intelligence must operate under physical and operational constraints.

While we believe in the thesis behind AMI, as with all startups, our conviction stems from the dynamism of this ambitious team. 

Yann LeCun has spent decades shaping the trajectory of modern AI. His contributions to deep learning helped lay the foundation for today’s systems, and he has long argued that prediction alone is not sufficient. Systems capable of true reasoning require internal world models that allow machines to understand how environments evolve and how actions shape outcomes.

That scientific perspective is paired with the leadership of CEO and co-founder Alexandre LeBrun. In one of our earliest meetings, what stood out was not just his experience founding three prior businesses, but the clarity of his sequencing and commercial discipline in translating foundational research into real-world systems.

Around them is a global team of researchers and operators spanning representation learning, human-centered AI, systems architecture, and global operations including, Saining Xie (Chief Science Officer), Pascale Fung (Chief Research & Innovation Officer), Mike Rabbat (VP of World Models), and Laurent Solly (COO). From day one, AMI has approached this as both foundational research and company building, pairing scientific depth with commercial execution.

We are proud to partner with AMI as it works toward this next chapter in artificial intelligence.