Foodpairing is Building a Model of the World — The Sensory World
Belgian AI company maps the journey from molecule to sensation to purchase. 100 products digitized daily. 200,000 consumers modeled. A self-driving lab to accelerate the mission.
Ghent, Belgium, December 10, 2025 --(PR.com)-- A new debate is emerging in artificial intelligence: the future may not belong to systems that predict text, but to those that model reality itself. One company has been quietly building exactly that for the consumer goods industry.
Foodpairing, the Belgian AI company specializing in sensory analytics and consumer prediction, has spent over a decade constructing what it calls "a model of the sensory world": a computational system that maps the journey from molecule to human experience to purchase behavior.
"The future belongs to those bold enough to code a model of the world itself," said Bernard Lahousse, founder of Foodpairing. "Ours is the world of sensory experience. We've digitized 200,000 real consumers and +23,000 products, creating a model that doesn't predict from history — it computes from the fundamental relationship between molecular reality and human response."
Industrial-Scale Digitization
Foodpairing's laboratory analyzes between 50 and 100 products every day, digitizing their aroma, taste, texture, and trigeminal properties. This continuous stream of molecular data feeds into the company's predictive models, enabling consumer packaged goods companies to simulate market response before a product exists.
The company is now evolving toward a self-driving laboratory — an automated system designed to dramatically accelerate this digitization process. The goal: to build the most comprehensive computational model of how humans experience products.
From Word Models to World Models
The critique of large language models is growing louder: they predict words without understanding physical reality. Foodpairing's approach is fundamentally different. While LLMs analyze text, Foodpairing's systems analyze the sensory properties that determine whether a consumer will reach for a product on a shelf.
"Word models tell you what people say about flavor," Lahousse noted. "World models tell you what people will actually experience, and whether they'll buy it again. That requires understanding the physical world — the chemistry, the biology, the psychology. Not just the language."
About Foodpairing
Foodpairing is an AI company that helps consumer packaged goods companies develop products faster through advanced sensory analytics and predictive modeling. The company's digital twin technology creates computational models of real consumers, enabling prediction of buying intent, liking, and portfolio optimization. Foodpairing serves leading global brands across food, beverage, and adjacent categories.
Foodpairing, the Belgian AI company specializing in sensory analytics and consumer prediction, has spent over a decade constructing what it calls "a model of the sensory world": a computational system that maps the journey from molecule to human experience to purchase behavior.
"The future belongs to those bold enough to code a model of the world itself," said Bernard Lahousse, founder of Foodpairing. "Ours is the world of sensory experience. We've digitized 200,000 real consumers and +23,000 products, creating a model that doesn't predict from history — it computes from the fundamental relationship between molecular reality and human response."
Industrial-Scale Digitization
Foodpairing's laboratory analyzes between 50 and 100 products every day, digitizing their aroma, taste, texture, and trigeminal properties. This continuous stream of molecular data feeds into the company's predictive models, enabling consumer packaged goods companies to simulate market response before a product exists.
The company is now evolving toward a self-driving laboratory — an automated system designed to dramatically accelerate this digitization process. The goal: to build the most comprehensive computational model of how humans experience products.
From Word Models to World Models
The critique of large language models is growing louder: they predict words without understanding physical reality. Foodpairing's approach is fundamentally different. While LLMs analyze text, Foodpairing's systems analyze the sensory properties that determine whether a consumer will reach for a product on a shelf.
"Word models tell you what people say about flavor," Lahousse noted. "World models tell you what people will actually experience, and whether they'll buy it again. That requires understanding the physical world — the chemistry, the biology, the psychology. Not just the language."
About Foodpairing
Foodpairing is an AI company that helps consumer packaged goods companies develop products faster through advanced sensory analytics and predictive modeling. The company's digital twin technology creates computational models of real consumers, enabling prediction of buying intent, liking, and portfolio optimization. Foodpairing serves leading global brands across food, beverage, and adjacent categories.
Contact
Foodpairing
Bernard Lahousse
+32497418588
www.foodpairing.com
Bernard Lahousse
+32497418588
www.foodpairing.com
Contact
Categories