As artificial intelligence increasingly becomes a vital component in the food and beverage industry, Unilever has emerged as a leader in harnessing this once-futuristic technology. The consumer products giant is utilizing AI to transform and accelerate its food product development, resulting in the introduction of various innovative products such as Knorr Zero Salt Cube, Hellmann’s Vegan Mayonnaise, and Hellmann’s Real Mayo Squeeze Bottle. Unilever is integrating AI throughout its operations, leveraging the technology to assess shelf life, texture, and flavor, while also predicting how products will perform during manufacturing. Additionally, the company employs AI to forecast flavor profiles, gauge consumer preferences, and enhance analytics within its food portfolio.
Recently, Manfred Aben, the head of science and technology for Unilever Nutrition and Ice Cream R&D, discussed the role of AI at Unilever, its evolution, and future directions. This interview has been edited for brevity and clarity.
FOOD DIVE: How has Unilever implemented AI?
MANFRED ABEN: Over the last couple of decades, the availability of data and computing power has significantly increased. Initially, AI was primarily utilized in marketing and consumer insights. Now, it plays a crucial role in supply chain optimization and R&D, which is my area of focus. We aim to create products that consumers prefer while ensuring safety and longevity. We employ digital modeling and AI to predict the shelf life of products, which not only benefits consumers but also helps in reducing food waste. AI enables us to anticipate flavor profiles and consumer preferences, allowing us to innovate faster and meet consumer needs more effectively. This approach minimizes the need for extensive experimentation in both the market and the lab, ultimately accelerating the delivery of new products to our consumers.
What products have you developed using AI?
ABEN: One notable example is Knorr Zero Salt Cube, a product that required us to identify alternative ingredients to replace sodium without compromising taste or functionality. Using digital modeling, we analyzed various ingredient combinations to arrive at a select few options that met our requirements. Another example is our Hellmann’s Vegan Mayonnaise, where we substituted egg proteins with plant-based alternatives. Consumers seek nutritious options that are better for the planet, but they still want a creamy mayonnaise experience. AI models helped us predict flavor profiles and consumer acceptance in different markets, which significantly reduced the need for laboratory trials and allowed us to respond more rapidly to consumer demands.
Does Unilever utilize AI in other areas, such as supply chain management and minimizing trial and error?
ABEN: Absolutely. We aim to optimize every aspect of our products, ensuring they taste great, are stable, user-friendly, profitable, and environmentally sustainable. Gathering extensive data on consumer preferences is crucial. We analyze taste panel results and consumer research while also ensuring that our products are manufactured efficiently and consistently maintain high quality. AI excels in processing a vast array of data, enabling us to identify potential solutions to meet consumer needs effectively.
How crucial is AI for Unilever?
ABEN: The company has evolved alongside technology, making it nearly impossible to envision our current operations without it. AI has enabled us to develop solutions that would otherwise be too time-consuming or impractical. Its integration spans all areas of the business, leading to financial benefits by enhancing our ability to address consumer needs. Faster product launches and the ability to adapt to market challenges, such as ingredient availability or climate issues, are significant advantages. The flexibility to quickly switch input materials while maintaining product quality is invaluable and would be costly without AI.
Can you elaborate on how AI has assisted with ingredient sourcing amid supply chain disruptions?
ABEN: We’ve encountered fluctuations in commodity availability, particularly with vegetable oils. By switching sources or combining different oils, we can keep products on the shelves. Given the global nature of food production, variations in ingredient quality can occur. AI allows us to ensure that replacements maintain the taste and texture consumers expect. This capability is essential for successfully navigating supply chain disruptions while retaining customer satisfaction.
Have there been instances where you found that human judgment surpassed AI capabilities?
ABEN: AI is not a magic solution; it relies on data and statistical analysis to draw conclusions. The quality of the data is fundamental, and in some cases, we may lack sufficient data to make informed decisions. Given the vast array of ingredients and formulations in food, the creativity of chefs and the insights of taste panels remain irreplaceable. AI can assist in predicting outcomes, but the human touch is vital for innovation, especially when creating something entirely new. The most effective outcomes arise from a collaboration between human expertise and AI tools.
Is there a specific instance where AI allowed you to analyze millions of recipes or data points quickly?
ABEN: The exploration of the gut microbiome is a fascinating area where we’ve delved into how various microbes impact health. In collaboration with Holobiome, we examined millions of edible ingredients to identify those that positively influence mental and gut health. Such comprehensive analysis wouldn’t be feasible manually. While we may not have millions of mayonnaise recipes, evaluating hundreds of them to optimize formulations—like reducing sugar levels—would still be an overwhelming task without AI’s capabilities.
Are you surprised by the rapid evolution of AI?
ABEN: The pace of AI development is indeed astonishing, exhibiting an exponential growth pattern as more data is generated. However, one challenge lies in understanding how AI derives its conclusions. Since AI often relies on statistical correlations rather than causations, human expertise is essential to validate findings. It’s an exciting era for technology, but it’s crucial to combine AI capabilities with human insight to ensure sound decision-making.