Raspberry AI CEO Cheryl Liu: AI should make the fashion calendar more responsive, 'not more frantic'

As generative AI compresses the time it takes to move a garment from idea to market, Raspberry AI CEO Cheryl Liu believes the fashion calendar should become "more responsive, but not necessarily more frantic." The goal, she told FashionUnited, is not to push brands into producing more collections, but to let them make decisions later, with better information.

"Today, brands often have to make decisions very far in advance because product development takes so long," Liu said. "If you can compress that process, you can make some of those decisions later, with better and more current information."

Founded in 2022, Raspberry has shifted from AI image generation towards what it calls an agentic platform, designed to carry a collection from trend research and mood boards through tech packs, assortment planning and e-commerce assets. In September, it launched a partnership with the Pantone Color Institute through Pantone's spring/summer 2027 Fashion Color Trend Report.

From image to workflow

Liu explained that image generation gives you an output from a prompt, while an agentic platform helps you complete a workflow and conduct much of the administrative, non-creative work on your behalf.

In practice, a design team could start with a brief such as "create a men's denim collection for Fall 2027" and use the platform to gather trends, develop mood boards, explore materials, refine designs, assemble an assortment and produce presentations and commercial assets. Designers can also generate tech packs directly in the platform and send them to manufacturers. "The important shift is from isolated outputs to connected work: the context, decisions and product direction can carry forward from concept through commerce," Liu said.

The Pantone partnership follows the same logic. Raspberry attaches the official Pantone identifier to a selected colour, so the reference stays with the concept as it moves on to technical teams and suppliers. "We're not trying to replace physical validation; our role is to reduce ambiguity and unnecessary iterations earlier in the process," Liu said.

Where humans stay in charge

Liu, who came to fashion through investing rather than design, said much of the delay she saw in product development had little to do with creativity. "It was spent on handoffs, revisions, waiting, file management and coordinating across different teams and systems."

According to Liu, people set the brief, define the consumer, pick which trends to pursue and decide what is right for the brand. Designers step in most often on silhouette, proportion, fabrication, wash and detailing; merchandisers shape the assortment and pricing; technical teams decide what can actually be produced. "The highest-value decisions including taste, differentiation and commercial judgment, remain human," she said.

It is also how she pitches the platform to design directors wary of what AI means for their teams. "So we don't position Raspberry as a substitute for the designer. We built it because we believe the designer's judgment is actually more valuable in an AI-enabled world," she said. "The technology can generate options, but it can't decide what your brand should mean or which idea is worth believing in."

Speaking fashion's language

Part of Raspberry's case against general-purpose image tools is vocabulary. Liu said a term like "fuzzy sweater" carries a precise meaning for a designer, describing a type of tactile, hairy yarn rather than an image that simply looks fuzzy. To build that understanding, Raspberry hired designers, merchandisers and product specialists in-house to work alongside its engineers. "You cannot build the right technology for an industry if you do not understand the way the industry actually works," she said.

Liu is equally direct about what AI cannot yet do. Without custom training, she said, it does not know which idea a brand should back, and it cannot judge fit from a screen. "We're not trying to pretend AI can replace every part of the process," she said. "The strongest workflow is one where AI is doing what it's good at: exploration, visualization, iteration and execution, while people are focused on judgment, taste, authorship and accountability."

Liu wants Raspberry to become the platform through which retail teams bring physical products to life, from the first consumer insight through design, development, merchandising, wholesale, marketing and e-commerce. For that to happen, she said, the platform has to earn the trust of creative teams, integrate with the systems brands already use and deliver measurable gains in speed, cost and decision-making.

"Most importantly, it has to help brands make products people genuinely want, with less waste and far more room for human creativity," she said.