For the 100th anniversary of its Lons-le-Saunier plant in the Jura region, Bel has crossed a technological milestone with the announcement of a “digital twin” for La Vache qui rit.
The group sees it as a world-first for a cheese.
The digital twin brings together in a single virtual environment the ingredients, the formulation, the physico-chemical properties, the sensory perception, and consumer preferences. Teams can simulate thousands of recipe combinations and only launch physical pilot tests on the most promising ones.
Anne Pitkowski, Bel’s Director of Research and Applications, summarizes the tool’s value: to predict the taste and sensory perception of a recipe. For example, whether it will be more or less melting, salty or creamy, and whether it will suit American or Vietnamese palates. The brand already has more than a hundred variations worldwide.
Traditionally, the development of a food product relies on a sequence of trials, followed by analyses and tastings. Bel presents the simulation as a complement to this expertise, not a replacement. The group also aims for a continuous improvement of the nutritional profile and the environmental footprint of its products.
Two years of development
The technology was designed by Bel’s research, innovation and development teams. It is planned to be rolled out at industrial scale by 2028, and the group hopes to release the first recipes generated by the tool by that date.
According to Les Echos, two years of development were required with substantial work in extracting and curating internal data. The amount invested is not disclosed, but Bel describes it as a “heavy” investment.
Other building blocks are already in service. Predictive microbiology solutions and formulation-assistance tools, stemming from work with external partners, are used in the new developments.
Dassault Systèmes, partner since 2024
The project stems from the agreement announced in July 2024 (signed in Q4 2023) between Bel and Dassault Systèmes.
This agreement envisions deploying the “Perfect Production” solution, built on the 3DEXPERIENCE platform, across 11 of the group’s factories. It relies on Manufacturing Operations Management (MOM) and Product Lifecycle Management (PLM) components, on food and materials science, and on artificial intelligence.
Dassault’s digital twin technologies are already used in aerospace, automotive, and life sciences.
The plant’s digitization (twin, sensors, analyzers) remains in the experimental phase. It aims in particular to reduce losses and to manage the variability of milk, a living raw material. The group reiterates that this is not about replacing humans.
Antoine Fiévet, Bel’s president, acknowledges the risk with Les Echos. The group invests tens of millions of euros every year in this data-and-tech approach, “with returns that remain to be demonstrated.” With €3.8 billion in revenue, Bel weighs less than Lactalis or Danone. “We are too small not to innovate,” he justifies.
