Imubit

Founders and development in Israel

Imubit was founded in 2016 by the brothers Gilad (Gil) and Nadav Cohen and by Meir Tseitlin, according to CTech. A 2021 article described a research and development center in Modi'in and sales and support teams in Houston. The official website today lists an address in Houston, and the leadership page shows Gilad (Gil) Cohen as CEO and Nadav Cohen as chief technology officer and president. The company develops artificial intelligence systems for manufacturing processes, including refineries and chemical plants. So its Israeli attribution rests on the documentation of the development center and on the company's founding, alongside commercial operations in the United States.

What the model learns from plant data

The website describes models of nonlinear, dynamic processes, built on the plant's historical data and guided by physical principles. The system links operating decisions, process disturbances and the measured results. Reinforcement learning is used to learn control decisions aimed at targets within defined constraints. In this way the AI does not just summarize an engineer's reports, but takes part in setting operating actions. In the company's description, engineers can see relationships the model has learned and test the controller before using it. The product is built around a specific physical process and its data.

Mapping, modeling and repeated control

Imubit presents a workflow of mapping the gaps between the plan and the state of the plant, building a model, and running control decisions. In the mapping stage, targets and constraints are defined and linked to results. In the control stage, the system carries out decisions, measures the gap against the plan again, and allows retraining when the plant changes. This is what closed-loop control means in the product description: the new data feeds back into the decision process. The company targets the chemicals, oil, cement and other materials industries. This is not a single model run without adaptation on every plant, but a system presented as learning the local process.

The funding round and the scale of investment

On 16 August 2021, CTech reported a $30 million Series B round led by Zeev Ventures and Insight Partners. Spider Capital and Upwest Labs also took part. At the same time, cumulative funding of $50 million was reported. The article described the solution being integrated into the facilities of large companies, but did not give a full list of contracts or a price per facility. No report of a later funding round was found.

Customers, operators and engineers

The company's website features testimonials from people at Citgo, Ash Grove Cement, Oxbow, Monroe Energy, Big West Oil and Preem. They describe use in control, collaboration between planning and operations, and support for training control room staff. One testimonial deals with learning through open-loop operation, in which decisions can be tried and their effects understood. These are testimonials the company chose to publish. The website emphasizes targets and constraints and engineers' sign-off on the controller. The product page describes installation at the plant site and integration with the existing control systems. Operators and engineers keep operating authority, and define the constraints and the units within the scope of control.