Manifesto

History

In 1997, IBM's Deep Blue beat Kasparov in Chess. Excellence in chess had long been the mark of intelligence and a computer beat our best.

In 2026, these machines have developed a superhuman understanding of code and mathematics, leaving us to ask what domino falls next. What else will computers understand better than humans? It is increasingly likely that the answer to that question is us.

Computer understanding by domain. Chess fell in 1997. Code and mathematics are at or past human level now. Understanding of humans is next.

What we are building

Jean's mission is to build computers that understand humans.

Jean started by developing downstream products that enable computers to better understand humans across domains (AI Memory, Context Engineering, Matching, Embedding, and Recommendation Systems).

Each has pushed us towards solving the more fundamental problem. In order to build these systems, we must first develop foundation models of human behavior.

For instance, instead of an LLM that predicts the next word, this means training a model that predicts the next action a user will take, the perfect product in the world for them, software that adapts to their presence, and matching systems that match one to their true soulmate based on their values and communication styles, rather than swiping or click data.

If we are successful, every human interaction with software will run on our models.