Ciresk is an applied research lab building enterprise world models.

Our mission is to make every company autonomous. Companies produce most of what people need, from software and banking to medicine, food and energy. How much they produce, and who can afford it, are still limited by the hours people can work. A company that runs itself can produce far more, for far more people.

Every company runs on feedback loops. A decision is acted on, and its result shapes the next one. Most loops that need judgement still run by hand, on workflows designed for a time when only a person could make that call.

An enterprise world model learns how a business actually works, from its systems and its own history. It predicts what a decision sets in motion, including its second- and third-order effects. Delay one shipment of parts, and it sees whether the line stops, which orders slip and what it costs the quarter.

On top of these models, we build agents that take on those loops. Each agent is given a clear target and is tested against decisions the company has already made. It then works beside the people responsible for that loop today, and every correction they make becomes another test it has to pass. It is trusted with more as it matches their best, until it runs the loop on its own.

That frees people for the work with the most leverage, the decisions only they can make.

Our research puts evaluation and data before model size, because that is where most of a system's reliability comes from.

Our systems are built to operate entirely inside a company's own infrastructure. The models, their memory and what they learn belong to the company.