Unilever leverages AI to transform plant proteins and vegan innovation

17 Jul 2023 — Unilever executives dived into how artificial intelligence (AI) and big data biology are transforming the face of consumer goods and helping to develop nutritious food at unprecedented speed and with a low carbon footprint during a recent London event centered on a new era of biotechnology and microbiomics.

Food Ingredients First attended the event, with experts sharing insights on the science behind the company’s next-generation products, which have been designed using machine learning. 

Speaking to us in London, Carla Hillhorst, Unilever’s chief R&D nutrition officer, explains the complex processes behind F&B Innovations. 

“If you take food products that have been around for hundreds of years, like yogurt, sauerkraut, beer, wine and bread, these fermented products are made using biotechnology. And biotechnology technology is, if you simplify it, a process where you take ingredients, such as sugar or proteins, you select a microorganism and then that transforms that into another product.”

Hellman’s mayonnaise, vegan mayonnaise and Knorr chicken stock.“At the moment, we at Unilever are using our digital capability combined with our continuously evolving scientific understanding to optimize this space. And we do not only do that in F&B, but we also really do it in the area of personal care, beauty, well-being and homecare,” she comments. 

Alternative protein development 
Notably, the plant-based meat alternative space is at the height of development and although this is not a new space of innovation for Unilever, it is crucial. 

Unilever is exploring precision fermentation to create plant-based products based on mycoprotein with its partnership with Enough. 

“We are continuously expanding the number of products that we have in our plant-based portfolio, and the reason why there is such a need for alternative proteins on the market starts with the bigger picture,” says Hillhorst. 

“If you look at the world of food and agriculture, livestock is a huge contributor to global greenhouse gas emissions. So, especially in the developed world, moving to more of a plant-based diet is important. Suppose we want to move to a world that is sustainable from a production point of view but also sustainable from a consumption point. In that case, we need to remove animal-based proteins from our diets. We cannot ignore the huge need for suitable alternatives that taste good and are nutritionally similar.”

“Exploring this whole space of alternative proteins gives us a basket of opportunity,” continues Hillhorst. And by utilizing AI and big data, these combinations and biotechnologies can help speed up the process and progress in the plant-based space.

“By putting these systems in place, you can end up with a product that uses 90% less water and 90% less speed. And in the end, you get a product with a carbon footprint at least 95% lower than it,” she notes.

“So whether that is meat alternatives for The Vegetarian Butcher, a Hellman’s Vegan Mayo, a plant-based boullion or a vegan ice cream, digital and big data can be essential to the development and the design of really great products that are appealing to the consumer.”

Working smarter, not harder
Unilever has been scaling digital technology for years and Hillhorst says, “It’s really helping us to work smarter in this area.” 

“Today, we have game-changing technology tools that help us to strengthen our insights, develop the next level of innovation and aid us with scientific discovery.”

In the instance of Hellman’s vegan mayonnaise, Hillhorst explains that “we do the majority of that in silico using our modeling capabilities. So it’s all happening behind the computer.”Carla Hillhorst, Unilever’s chief R&D nutrition officer.

“We are predicting what the shelf life and microbiological stability of the product will be. We are predicting whether the product can run on our processing lines in the factory and using our models to inform us what the formulation of a product needs to look like at the moment.”

To achieve a specific texture using AI, you can create a sensory profile that you are looking for and then you use your modeling techniques to assess in a relatively short amount of time: “30 seconds in this case,” Hillhorst tells us. 

Adapting to vulnerable supply chains
Unilever is also leveraging AI and digital data in the area of portfolio analytics, which means having data at your fingertips that allows you to optimize the cost structure of your part continuously or if you have to reformulate because an ingredient is no longer available.

“You can do that really fast. But unfortunately, we live in a world where supply chains are quite vulnerable at the moment, so we do have to do these reformulations quite a bit, which also means it can help us improve our products’ nutritional profile,” Hillhorst explains. 

“This level of digital capability is becoming more paramount to our business. It’s also been very empowering and time-saving. Our experts in a specific field can focus their attention on creativity, innovations and values. With these models in place, we can create concepts that benefit both people and the planet.”

Risk to human jobs?
Replacing repetitive tasks frees up a lot of time where value can be added and used more efficiently, but what implications might this have for human workers?

Digital and big data can be essential to the development of products that are appealing to the consumer.Hillhorts says she “isn’t concerned” there is a threat to human productivity when using computer models, such as AI and big data, but she does point to safety concerns about how data can be shared and accessed. 

“We’re using a cloud-based system, and while we can start to access each other’s data, the number of additional insights that can be created from that can be enormous and effective in future learning.”

“Right now, we’re in the middle of development, so more data will be made available, which means we can start tapping into smarter ways of product design and using less energy, for example,” she explains, adding that the end products can be much more customizable with processes like this put in place. 

“It’s also not about replacing human workers, we still need experts in their field, but it’s about making processes simpler, more robust and whereby we can free up some of that time to be more creative in the entire development process,” Hillhorst concludes. 

By Elizabeth Green

To contact our editorial team please email us at
[email protected]