Decision intelligence for the physical world.

Every factory is a stream of decisions: what to buy, what to make, what runs next, who runs it. Each is taken under constraints that are physical, human and unforgiving. Most are still made by hand, under pressure, with a spreadsheet.

We exist to make those decisions as good as the best planner on their best day, every hour, in every plant.

The intersection of machine learning and operational research. Optimisation gives you the right answer to a well-posed problem; learning tells you what the problem actually is. Neither is enough alone. We work on systems that read a plant, pose its problem precisely, solve it, and explain the result. They improve as the people who use them push back.

The mathematics of deciding well under constraints. It scheduled convoys in the 1940s, airlines in the 1980s and warehouses in the 2010s. It has never reached most factories, because every factory is its own problem and the modelling was too expensive to do by hand. That is the part we believe AI changes.

Because the stakes are physical. A plan is not a recommendation. It becomes steel, hours, shipments and shifts. Manufacturing is where a better decision does the most good for the most people, and where a bad one is felt immediately.