Skip to main content

Practical Magic

How Unilever uses AI to cut petrochemicals out of laundry soap

Unilever used AI to identify replacement ingredients for its laundry and cleaning products five times faster than previously possible.

Unilever product laboratory

Source: Unilever

Unilever is speeding development of petrochemical replacements for ingredients in its cleaning and laundry products with the help of artificial intelligence. 

It’s a strategy that will continue under new CEO Hein Shumacher, who has pointed to research and development in innovative disciplines — notably synthetic biology — as central to Unilever’s sustainability strategy.

One example is Uniliever’s multiyear partnership with Arzeda, a Seattle-based company using physics-based computational protein design and machine learning to identify renewable and biodegradable proteins. Enzymes are responsible for chemical reactions. They’re also naturally derived, and Unilever sees them as one potential ingredient to help decarbonize products such as OMO, its largest detergent brand.  

"It is believed that only 5 percent of the enzymes that occur naturally have been identified and fully understood," said Neil Parry, head of biotech for Unilever, in July 2021 when the company’s Arzeda relationship was announced. "So when looking for new discoveries, it’s a bit like trying to find a needle in a haystack — then you need to tackle that one key find and optimize it to be as effective as possible in your application."

So far, the two companies have identified enzymes that can eliminate stains while reducing the amount of energy and water used by cleaning products and potentially cutting the number of ingredients needed for certain products in half. They did this in just 18 months, five times faster than traditionally possible, Parry said this summer.

"Not only can you optimize from the templates you have today, but you can create entirely novel models that you don’t have today," he said. 

This isn’t just about directing computers to solve a problem. Physicists, biologists and data analysts were all involved. As part of the process, Unilever’s design teams worked with Arzeda to create lists of requirements for its products, Parry said. The algorithms draw on past research to make suggestions, helping prioritize the most viable options. "It’s an amalgamation of the science, which is leading to a step change in how things are developed," he said.

Arzeda, which raised a $33 million Series B round of venture funding in March 2022, doesn’t just help companies with discovery. It’s also a co-development partner. It has already worked with Unilever to design, engineer and test the enzymes in real-world testing quantities. The next step will be to scale production of the enzymes and help commercialize them. "We are getting better and better at the design with every undertaking," said Arzeda CEO Alexandre Zanghellini.

The two companies declined to discuss their financial relationship or when the enzymes will actually show up in products, but Unilever is working against real deadlines. The company has committed to net-zero carbon emissions for its products "cradle to shelf" by 2039 and aims to make its product formulations biodegradable by 2030.

"We’re not working on something that doesn’t have a tangible outcome," Parry said.

[Want more great insight on technologies and trends accelerating the clean economy? Subscribe to our free Climate Tech Rundown newsletter. ]

More on this topic