How digital twins are shaking up food and beverage development

Digital twins are virtual models that replicate real-world entities, enabling companies to simulate and predict various outcomes under different circumstances. In food product development, these models can, for instance, provide a comprehensive simulation of how consumers might react to new product ideas.

Simulating reality

“We were the first company that used digital twins for product innovation in the food space,” says Langenbick. He explains that Foodpairing has earned this pioneering position thanks to the database it has developed over many years. “We have digitised the flavour profile of thousands of ingredients, creating detailed chemical breakdowns that help us understand how different combinations work,” he says. This means that Foodpairing can know exactly which molecules are in a product, measuring up to 1,200 physiochemical parameters.

The scale-up has found several uses for the virtual doppelgangers, including finding unexplored product categories (white space mapping), analysing the competition, creating and testing new product ideas, and developing recipes and (re)formulations.

Formulating with your virtual friends

When applied to product formulation, the process of applying a digital twin begins with creating a “gold standard” in the form of the ideal flavour profile for a product, where Foodpairing can draw on its extensive database. Once established, AI generates multiple recipe variations that align with the gold standard. A chef prepares these recipes, allowing a comparison with the established standard. According to Langenbick, this process takes days, compared to months with traditional methods.

Digital twins can also simulate consumer reactions to newly developed products. By leveraging AI and machine learning, Foodpairing can predict how consumers will perceive and respond to different flavours, textures, and combinations.

This AI-assisted process allows for rapid iteration and refinement of product formulations, reducing the time and cost associated with traditional trial-and-error methods. Langenbick claims that – thanks to the ability to simulate very large volumes (thousands to millions) of consumer responses – this approach also enhances the accuracy of the product development process, ensuring that the final products are more likely to succeed in the market.

“A lot of companies started to validate the technology against a real consumer panel. With some companies, it is a KPI [key performance indicator] within the collaboration with them,” he says. “Today we have models that can predict exactly what a consumer panel would say; it is highly accurate.”

The rise of digital twins in food innovation

While digital twins are already a common practice in manufacturing, they are only just becoming instrumental in the process of new product development. The drastic reduction in time and cost compared with traditional development methods is driving a lot of the interest of industry players.

Smaller brands were the first to recognise and apply the opportunity to test new products on limited budgets. “We often see smaller companies using our digital twin technology to bridge the gap quickly and cost-effectively," Langenbick says. Large companies, initially somewhat resistant to these new technologies, are now also beginning to embrace them. The agility demonstrated by big firms during the Covid-19 pandemic highlighted the need for quick adaptation and innovation. As digital twins become more integrated into standard practices, their adoption is expected to grow across the industry.

Foodpairing’s collaboration with the brand Sigma Alimentos showcases the practical applications of digital twins and AI in food innovation. The companies jointly developed hybrid meat products that blend meat with vegetables, maintaining the flavour and texture of traditional meat while incorporating up to 40% vegetable content. This innovation addresses consumer demand for healthier, more sustainable options and reduces the environmental impact of meat production.

By creating virtual models of the hybrid meat products, Foodpairing refined the recipes and ensure they met consumer expectations. This approach significantly reduced the time and cost associated with traditional development methods.

Artificial intelligence and the future of product development

As more companies increasingly recognise the benefits of digital twin and the larger universe of AI technologies, their adoption is likely to increase. Langenbick envisions a near future where digital twins and AI are integral to the development process.

He expects to find the biggest low-hanging fruit in generation and validation: “Within the next three to five years, digital twins technology will disrupt traditional consumer panels.” When applied to product formulation, the gains are even more significant, as companies can struggle for years at this stage of development. While AI will eventually accelerate and de-risk the process significantly, this is also a more complex area that will take longer to mature.  

Awareness of AI and its potential has grown rapidly in recent years, and this trend is likely going to continue.

“Manufacturers and brands are generally more aware of AI's potential due to industry discussions and pilot projects. Chefs are becoming increasingly interested, especially as AI-powered tools emerge in the culinary space. Everyday consumers are less aware of the specifics but are experiencing the impact indirectly through new products and experiences.”

Integrating the technologies successfully and smoothly involves not just a focus on success stories, but also a greater understanding of limitations.

“Transparency is key to building trust and ensuring ethical use of AI. Collaboration between technology providers, industry experts, and consumers will be crucial to harnessing AI's full potential in the food and beverage sector. This includes sharing knowledge, developing best practices, and addressing concerns openly.”