by rebel labs
The most consequential decisions in the world are still made on intuition and partial evidence. We believe every important decision should be tested before it is made.
Every institution, company, and government makes choices whose consequences ripple across millions of lives, and they make them with little ability to see where those choices actually lead. There should be a way to run a decision forward, observe its outcomes, and search across the alternatives until the best one is found. When the cost of being wrong is high, guessing should not be the only option.
The physics was never the hard part
Physics has long been simulable. We can model airflow over a wing, stress in a bridge, heat moving through a chip. What has remained out of reach is the rest of the world: people, societies, and the way they interact with the physical systems around them. A decision rarely fails because the physics was wrong. It fails because people responded in ways no one modeled.
Our research focuses on closing this gap. We are building simulations that couple the physical world with the human one, so that the second-order effects of a choice, the way people, markets, and communities actually react, can be studied with the same rigor we already apply to materials and machines.
One simulation is not enough
A single simulation tells you what might happen. The real value comes from running many. When a decision can be simulated cheaply and accurately, it can also be searched: thousands of variations explored, compared, and ranked, so that the optimal path is not merely plausible but discovered. We treat decision making as a search problem over possible futures, and we build the tools to explore that space exhaustively.
Human behavior is within reach
Modeling people and societies has resisted the methods that made physics tractable. Human behavior is noisy, adaptive, and dependent on context. But recent advances in AI give us, for the first time, models expressive enough to represent that behavior and fast enough to run it at scale. The verifiable structure that made progress rapid in domains like physics and code can now be extended to the coupled systems where people and physics meet.
Everything should be simulable
Our ambition is a world where everything can be simulated to accuracy, from a chip to a rocket ship to a decision about the people a policy will affect. We believe this is one of the most important capabilities a society can have: the ability to look before it leaps, and to keep looking until it finds the best way forward.