Unilever Foods is treating AI-driven discovery as a parallel universe to the supermarket aisle: one where visibility happens inside voice assistants, AI search tools, and large language models, not just search blue links or endcaps.
Olivia Kirby, Unilever Foods’ director of integrated demand generation, says the company is using AI across marketing and product development, especially by reworking recipes and site content to be more machine-readable. The goal is for brands like Knorr and Hellmann’s to have a better shot at being surfaced — and recommended — when consumers ask AI systems what to cook.
One concrete case: ahead of Super Bowl 2026 in the US, Hellmann’s found it had low visibility for “Game Day sandwich recipes.” The fix wasn’t a new campaign so much as a formatting intervention: adding a dedicated “Game Day sandwich” listicle, tightening keywords, reformatting existing content into AI-friendly lists, and expanding the recipe section.
Kirby says the changes produced a ten-position improvement in visibility rankings and nearly doubled the overall visibility score from the 10% benchmark. An unusually specific datapoint in a space that’s often vibes-based and unverifiable.
Unilever is also using AI to speed up product development. For Knorr’s Fast & Flavourful Paste, the company digitally tested recipe formulations before moving to physical trials; Kirby says this cut development time in half by reducing trial-and-error in kitchens, labs, and pilot plants.
Over in foodservice, Unilever Food Solutions is feeding AI with proprietary inputs — market research plus around 35,000 chef-authored recipes — alongside customer “online signals” like menus, reviews, and public social content. The result is a conversational interface that gives chefs and foodservice operators personalized recommendations.
Still, Unilever’s line is that the technology gets you “80%” there, but humans make the final calls. Because, as Kirby puts it, you taste things with your mouth, not a model.
Read more at Food Ingredients First.