Foodpairing uses Google Cloud to predict consumer tastes with 75% accuracy, cutting development time to days.
Foodpairing uses Google Cloud to predict consumer tastes with 75% accuracy, cutting development time to days.
Analyzing the data behind the perfect bite
In the food and beverage industry, launching new products successfully is a challenge. More than 80% of product launches fail, due in large part to the sector’s traditional reliance on slow, expensive consumer testing. Historically, brands have faced a difficult choice: spend months conducting surveys to understand what consumers really want—by which time tastes may have already changed—or move fast and risk getting it wrong.
We were looking for a provider that could give us building blocks so we could focus on where we actually make a difference as a company. With Google Cloud, everything is tied together, making it easy to build and scale without losing time managing infrastructure.
Bernard Lahousse
CSO and Co-founder, Foodpairing
Foodpairing was created to solve this bottleneck. The Belgium-based company analyzes the compounds of specific aromas, tastes, textures, and appearance that drive human preference, enabling food companies to quickly understand what consumers want before they launch a product. To map this level of scientific detail onto the global food market, the team needed a technology ecosystem capable of managing robotic laboratory measurements, vast datasets, and complex consumer simulations in a single environment.
Foodpairing chose to build its platform on Google Cloud because its unified ecosystem provided the data storage and analytics, advanced AI capabilities, and agile infrastructure the team needed to turn a complex scientific mission into a global platform.
By using BigQuery as a high-speed backbone for their molecular data and Firebase to launch new apps quickly, they could focus entirely on their scientific research.
As Bernard Lahousse, CSO and co-founder of Foodpairing explains, “We were looking for a provider that could give us building blocks so we could focus on where we actually make a difference as a company. With Google Cloud, everything is tied together, making it easy to build and scale without losing time managing infrastructure.”
From supermarket shelf to BigQuery—and back again
The innovation journey begins in the laboratory, where Foodpairing digitizes the retail experience by breaking down 100 products every day into their core molecular components. These data points—covering aroma, taste, trigeminal, appearance and texture—are streamed into Cloud Storage and funneled through automated pipelines into BigQuery. What was once weeks of manual chemical analysis now happens autonomously in hours, creating a vast database that serves as the foundation for the platform’s intelligence.
This data is combined with extensive consumer research to fuel a platform of 200,000 digital twins—virtual consumers built from real-world preferences and behaviors. Using Gemini Enterprise Agent Platform, Foodpairing models how these virtual individuals will react to new flavors or packaging. Because these simulations run on Cloud Run to scale processing power automatically, brands can use this platform to test dozens of concepts in a single morning to see which formulations will generate the most interest. This allows companies to validate ideas before a single physical prototype is manufactured.
To deepen these insights, the company uses virtual agents, running on Google Cloud and powered by Gemini, to simulate market launches. Instead of recruiting physical focus groups, Foodpairing uses these agents to communicate with one another to replicate months of consumer behavior in minutes, accounting for variables like social media trends and competitive pricing.
This high level of automation allows Foodpairing to maintain a highly efficient operational model while managing global data sets that would otherwise require significant manual oversight.
“With Google Cloud we have built a closed loop,” explains Jan Stout, senior data scientist at Foodpairing. “We analyze products in the lab, build a molecular map in BigQuery, and predict consumer responses with Gemini Enterprise Agent Platform and Gemini with over 75% accuracy, enabling new products to be launched with confidence. It all runs on Google Cloud.”
As well as serving global industry leaders, Foodpairing also empowers the culinary community through the Inspire platform for chefs. Hosted on Cloud SQL and Firebase, Inspire provides professional chefs with instant access to molecular pairing data, helping them move beyond intuition to discover new data-backed combinations and optimize recipes in real time.
With Google Cloud we have built a closed loop. We analyze products in the lab, build a molecular map in BigQuery, and predict consumer responses with Gemini Enterprise Agent Platform and Gemini with over 75% accuracy, enabling new products to be launched with confidence. It all runs on Google Cloud.
Jan Stout
Senior Data Scientist, Foodpairing
Building a self-driving lab for the future of food
The shift from traditional research to predictive AI has fundamentally changed how Foodpairing’s clients operate. Innovation cycles that once lasted 18 months now last just days. This agility allows brands to respond immediately to changing tastes, such as viral social media trends, or quickly reformulate a product due to ingredient shortages. It bridges the gap between marketing and research and development, providing a shared language based on data rather than subjective taste.
The next phase for Foodpairing is the self-driving lab—a 24/7 autonomous facility where Gemini-powered agents decide which experiments will be most useful to run next.
This will further expand the map of human taste and refine the accuracy of the digital twins, helping the company go further in its mission to help the food industry navigate rapidly changing consumer tastes.
For Lahousse, being able to use Google Cloud to understand consumer tastes and forecast their behavior has been fundamental to this transformation. "From using BigQuery to store 70TB of molecular data for our digital twins, to powerful AI capabilities to predict intent, to agile infrastructure to power it all, Google Cloud gives us a complete set of tools to help clients understand the science of what people love to eat," Lahousse concludes.
From using BigQuery to store 70TB of molecular data for our digital twins, to powerful AI capabilities to predict intent, to agile infrastructure to power it all, Google Cloud gives us a complete set of tools to help clients understand the science of what people love to eat.
Bernard Lahousse
CSO and Co-founder, Foodpairing
Foodpairing is a Belgian food-intelligence leader that blends molecular science and AI to empower global brands and chefs to create the next generation of food and drink people love.
Industry: Food and Beverage
Location: Belgium
Products: BigQuery, Gemini Enterprise Agent Platform, Gemini, Firebase, Cloud Storage, Cloud Run, Cloud SQL, Google Kubernetes Engine
