Haffner Energy turns to AI to steer its biomass thermolysis and hires Patrick Haffner
In a press release published on 29 October 2026, Haffner Energy presented a new stage in its technology strategy: gradually generalising artificial intelligence in the operation, control and optimisation of its biomass thermolysis technologies.
The approach adopted is based on the interaction between plant data and operator experience. Safety functions, for their part, remain entrusted to automated systems independent of AI.
AI working alongside operators to steer the plants
The company stated its intention to apply the so-called “human-in-the-loop” approach to its industrial processes. The stated principle is not to replace the operator, but to extend his or her analytical capabilities while allowing the AI to learn from human experience.
Data produced continuously (temperatures, pressures, flow rates, biomass characteristics, equipment performance) is expected to be used to detect weak signals indicating a drift. The operator retains the interpretation of this information and the decision.
Haffner Energy refers to a gradual evolution of the supervision interface towards an “intelligent driving copilot”, fed by the experience accumulated at Marolles and across the installed fleet. The objectives cited are earlier detection of anomalies, anticipation of maintenance and a reduction in unplanned shutdowns.
Philippe Haffner, co-founder and chief executive officer, has set the goal of translating these lessons into gains in performance and reliability, with, for customers, increased availability and savings on operating and maintenance costs.
A long-term consultant, brother of the co-founders and shareholder
Patrick Haffner will act as a consultant, within the framework of a long-term collaboration with the technical and operational teams. He is the brother of the company's co-founders and has been a shareholder since 2017 through Haffner Participation, according to his own statement.
A graduate of École polytechnique, Télécom Paris and holder of a PhD in signal processing, he has worked on neural networks since the late 1980s. He worked at Bell Labs and then at AT&T Labs in the United States, and most recently held the position of “Principal Applied Scientist” at Amazon Web Services.
In 1998, he co-signed with Yann LeCun, Léon Bottou and Yoshua Bengio an article on convolutional neural networks, published in Proceedings of the IEEE. He also led the deployment of a system for the automatic reading of amounts written on bank cheques in the United States.
The company specified that AI is not intended to replace automated systems or programmed safety loops. Changes to these devices will remain validated by the engineering teams.
