How Unilever is using AI to improve its products

It’s been nearly 100 years since British soapmaker Lever Brothers and Dutch margarine company Margarine Unie merged to create CPG giant Unilever, and the company has evolved a lot since then, most recently using AI to advance its product portfolio.

Alberto Prado, Unilever’s head of R&D digital and partnerships, is leading this transformation. He told Retail Brew part of his strategy is introducing digital tools to the R&D process to make innovation “better, faster, and more effective.” That means a shift from physical to digital labs, using “in silico science” (aka experiments done on the computer).

“In digital, we can plan, discover, design, and optimize in ways that we cannot in the physical space,” Prado told us. “We can do that without friction, without constraints, and without the limitations of physics and the economics around the physical world.”

Beauty secrets: Molecular models help Unilever improve its product formulas much faster than in a physical lab, cutting down the time to a few hours, Prado said. These processes are “very intimately” tied into Unilever’s sustainability goals, he noted.

• Last year, the company announced its commitment to make its products’ ingredients and formulations biodegradable by 2030. With modeling, it can simulate how a formulation will break down over 10 years rather than waiting a whole decade to find out.

Unilever has also used AI to improve its lipsticks—and save beetles. Its Hourglass Red 0 lipstick, released earlier this year, uses a vegan replacement for carmine, a commonly used lipstick pigment that requires crushing 1,000 female beetles per tube to produce, according to Prado. Unilever used modeling to determine which non-animal-derived ingredients could combine to pack the same color punch as carmine.

“You can apply the same logic and the same methodology to any kind of formulation in order to make it more sustainable, more animal-friendly without necessarily impacting the performance or even the cost of the formulation,” Prado said.

• He said Unilever has also collected and delved into 12 terabytes of data around microbiomes (the microorganisms on the skin), allowing it to add product claims on its skincare brands like Living Proof, Vaseline, Dove, and Tatcha to state how the products support the microbiome.

Work in progress: Of course, applying this technology to an over 100-year-old legacy company takes time and adjustment, Prado shared. The process involves finding not only scientists skilled in biology and chemistry, but those with expertise in data science and statistical analysis, and requires people to trust the outcome of these models over seeing the results IRL.

“The ability for you to rewire or reframe the way that you can create value is something that is not easy,” Prado said. “[It] still is a multi-year journey.”