How A Headless Architecture Helped Unilever Reimagine The Customer Journey Around AI

Throughout the Internet and mobile revolutions, consumer brands designed their digital experiences to engage human users. Now, companies like Unilever are quickly reimagining their sales journeys with AI agents in mind. 

Customer interactions are no longer linear. Nearly half (45%) of consumers are already using AI to shop, according to Bain & Company research. The days of scanning pages of Google results, navigating brand websites, and manually checking out are quickly fading. Increasingly, shoppers are moving seamlessly between social platforms like TikTok, retailer apps and marketplaces, and AI engines — often within the same moment. 

“The traditional path to purchase is fundamentally changing,” Prashaant Huria, Global VP, Digital & Technology – Customer Development, told CIOnews. “Content now drives conversation, recommendations become purchase, and inspiration and transactions sit side-by-side.” 

At Unilever, the nearly 100-year-old maker of iconic brands like Dove, Domestos, Vaseline, Axe, and Hellman’s, the transformation extended beyond the technology stack. In the new AI-first world, serving the nearly four billion consumers the company reaches everyday means rethinking everything from product content, to fulfillment, data connectivity, and retailer collaborations. 

“Digital shelf visibility is becoming as important as traditional brand presence,” said Prashaant. 

Prashaant leads Unilever’s global digitization efforts for the customer development and sales functions and is spearheading work to turn the company’s e-commerce operations into an integrated ecosystem built around AI agents. The job brings together various aspects of Prashaant’s decades-long career, including enterprise architecture, software engineering, and data strategy — all at Unilever’s global commercial scale. 

“It’s about building a new growth model for how brands are discovered, chosen, and bought,” he said. “Creating experiences that work seamlessly for both consumers and AI agents represents a fundamental shift in how we design digital experiences, brand experiences and product data.”

CIOnews talked to Prashaant to learn more about the transition to an AI-centric shopping experience, the importance of APIs, and Unilever’s data transformation. 

‍The battle for AI relevancy  

The pillars of customer engagement used to be discovery through targeted, keyword searches, eye-catching layouts, and natural, seamless browsing experiences. But with AI agents now at the center of the consumer journey, these human-centric elements are less important. 

Instead of showing up on search pages, companies like Unilever have to make sure that when a consumer in London asks their AI agent for recommendations on treating dry skin or other conditions, for example, the system surfaces products like Dermalogica. But unlike humans, AI agents rely less on visual presentation alone and more on structured, trusted product information. Data and metadata are now key to discovery. And staying relevant to the AI engines takes coordination across Unilever’s operations. 

If inventory isn’t updated in real-time and items are out of stock or difficult to buy, the AI agent will ignore the seller, according to Prashaant. And to optimize for LLMs, product descriptions must have deep contextual details and evidence-based claims. 

 “If an LLM cannot seamlessly read, trust, or validate information, it simply will not recommend your brand to the human consumer,” he said. 

‍Staying ahead by going headless

Like other heritage brands, Unilever focused initially on unifying previously siloed data, and structuring the assets to draw the attention of AI agents tasked with recommending products. Real-time inventory feeds now eliminate gaps between the digital and physical shelves, for example, so AI agents aren’t suggesting out-of-stock items.

“Real time data unification across the entire value chain is not a back-end capability, but a modern demand-gen capability,” said Prashaant. “To feed an AI agent accurate information, your data cannot exist in silos. You cannot have one set of data for logistics and one for marketing. You need global data harmonized down to the SKU-level.”

To deliver on this mandate, Unilever relies on a so-called “headless” architecture, where data assets are easily accessible and reusable across various engines and consumer interfaces through APIs. The company is also starting to share its data feeds with external launch partners. 

“In traditional commerce, the product data is bound to a specific webpage layout. In a headless layout, the data exists independently as a clean, harmonized product," said Prashaant. “Using APIs, we treat each part of the value chain as an independent microservice.”

‍Building the ‘Perfect Store’ 

Prashaant and team’s efforts are paying off. Vaseline, a 150-year-old brand, set a new benchmark for social commerce performance during its first-ever UK TikTok Live last year: “We’ve reinvented that product for the digital age,” said Prashaant. 

But it’s still early days in the AI transition. Ultimately, the shift to the new AI-centric integrated growth model doesn't boil down to a hard science. For every TikTok Live that sells out, there are items that don’t rank well or never make it out of the checkout baskets. 

It’s why Prashaant and team constantly review and update product content to drive conversation and growth. Reports indicate the SKUs that aren’t performing — whether that be a ranking drop or inventory bottlenecks — so they can correct the listing, adjust the fulfillment path, and keep the digital shelf ready for shoppers. 

Overall, like other departments, Prashaant’s team is guided by Unilever’s “Perfect Store” initiative, a strategic global program underpinning the company's growth ambitions aimed at flawless execution across every customer channel — both digital and in the real-world. 

“It’s all about getting the basics right. Right product, right price, and always available — in the supermarket or the digital shelf,” said Prashaant.