Can AI help brands figure out what will sell before they make it?

Fashion trends move fast. But when it comes to getting the clothes onto shelves, Raspberry AI moves faster.

The platform wants to close the gap between prediction and production by expanding beyond AI-powered design into the work that follows, from technical specifications to e-commerce imagery and marketing campaigns.

Currently, the company works with 460 brands, per CEO and Founder Cheryl Liu, who said the expansion grew out of seeing how much work remained after a design was approved. “Design is not where the work ends,” she told Retail Brew. “That’s just the beginning.”

The idea is to connect tasks across design, merchandising, and marketing, allowing brands to create digital product concepts and the assets needed to sell them before physical goods are ready.Liu explained that brands could move from an initial sketch to marketing and e-commerce assets in the same day. That doesn’t mean the clothes are manufactured that quickly; the speed comes from moving more of the development and visualization process to digital.

From sketch to screen: Take a denim collection, for example. A brand could give Raspberry’s AI agent a broad brief, such as developing a menswear range for an upcoming fall season. The system would then review signals from runways, social platforms, and Google searches, then generate mood boards and design concepts, Liu said.

Designers and merchandisers can also select and refine those options before the platform generates technical specifications for their teams to review and visualize products on models and create imagery customized to fit different retailers’ requirements. Pricing starts at $49 a month, with customized enterprise offerings coming at a steeper price point.

Incoming: But that’s just half of the story. Another part of Raspberry’s ambition is helping brands identify what shoppers will want before committing to production. Liu argued that traditional forecasting struggles to keep pace with trends emerging from social media and celebrity appearances.

But whether Raspberry’s predictions translate into sales remains an open question.

“On the prediction side, we haven’t tested it against actual sales yet, mostly because, as we talked about, the product development cycles are still so long,” Liu said, adding that the company has tested its AI-generated imagery, reporting 15% higher conversion compared with historical photo shoots. In fact, she attributed the improvement to brands being able to show more angles, styling options, and body types. The interview did not establish the test’s sample size or methodology.

Liu sees opportunities to reduce sampling rounds and let wholesale buyers review digital assortments before brands produce selected items, potentially cutting costs and waste.

However, she cautioned against leaning too much on AI. “I probably wouldn’t forgo physical sampling altogether or photo shoots altogether,” she said.