About You, the German online fashion retailer Zalando acquired in July 2025, has cut product photography spending by roughly 90 percent using generative AI — a saving of around 6 million euros on a budget line that previously ran at about 7 million euros. Tarek Mueller, managing director at About You, disclosed the figure at Zalando's annual "Partner Day" in Berlin last week, where about 750 attendees from brands including Adidas, Nike, Levi's, Calvin Klein, Lacoste, Guess, New Balance, Next and Triumph gathered to hear the platform's technology roadmap.

The event, held since 2014, put artificial intelligence at the center of the agenda. "AI is changing everything we do," Zalando co-founder David Schneider told the main stage. "The rise of mobile [technology] changed customer behavior… the rise of AI is much deeper, a fundamental change."

But Mueller framed the commercial questions bluntly in his opening remarks: "Does AI help create awareness for my online shop? Does AI help me convert traffic into orders? Does it help me to get more efficient and improve margins?"

The answers are mixed. Large language models such as OpenAI's ChatGPT and Anthropic's Claude currently send less than 0.5 percent of traffic to online fashion stores, Mueller said. Google and Meta remain the dominant traffic drivers for online fashion.

Where AI has already delivered measurable results is content production, for both product detail pages and marketing campaigns. Zalando and About You have both almost completely eliminated the dedicated studios they previously operated for catalogue-style product photography. "Around 70 percent of what you see behind me was created with AI," said Matthias Haase, Zalando's vice president for content solutions, gesturing at a video wall of campaign imagery.

One exception: content featuring real talent, such as brand ambassador and actress Lily Collins, who attended the Berlin event, is still produced conventionally.

From cost cutting to conversion

About You had been working with AI-generated model and product shots for about 18 months, but Mueller said the "real breakthrough" came roughly six months ago — the point at which AI stopped merely reducing costs and started increasing the platform's conversion rate.

The reason, Mueller argued, is replicability. A brand may find a model who drives sales, "but then they move to Milan and you can't book them again. Or the stylist has a bad day, or certain products are not available. It's very hard to replicate what works. But AI is learning what works — and it can replicate it."

About You houses the Scayle e-commerce software business and its AI content tool, Scayle Studio. The service offers brands AI-generated models across age ranges, with selectable poses such as "open laugh" and "confident cool," styling options including shoes and sunglasses, and lifelike animated backgrounds. The system can generate dozens of images per look, then decide which performs best — informed by A/B testing and increasingly by fine-grained shopper data accumulated over Zalando's 18 years of operations.

"It's a very individual thing. It's the same for styling, it really depends on who and what you are," Mueller said. Personalization will extend to how products are presented: "It might recommend me a product I like, but also styled in a way I like it, or even styled on me, or maybe a model I relate to."

Speed to market

AI content is also compressing campaign lead times. Haase receives a report every morning around 7 a.m. from an AI agent that crawls social media for trending content, pulling the five most relevant "micro-trends" and predicting each trend's life cycle. His team validates the data against Zalando search behavior.

After detecting the "arty fashion" trend at the most recent Met Gala in New York, Zalando launched a campaign the day after the event, producing an uptick in wish lists and followers for relevant brands. Zalando's fastest campaigns previously ran six to eight weeks — "But even six to eight weeks is too long to grab a micro trend," Haase said.

James Rothwell, senior vice president for marketing and content, said AI has shifted Zalando from traditional advertising toward publishing its own content, including short films. "Eighteen months ago, Zalando was not a content publisher," he told the audience.

The studio closures have had human costs. Of the 90 positions lost when Zalando's product studio closed, around 60 people were relocated elsewhere in the company, Haase said. Human judgment still gates the output: "Maybe [what the AI selects] fits the trend, it fits the style but it just doesn't look good. That's where the human comes in and says, 'out, out, out.'"

Where AI stalls

AI-powered shopping assistants remain unproven. Mueller noted that OpenAI has stepped back from shopping-assistant plans and turned to selling advertising instead, while Google's assistant can search for products but cannot complete transactions. Even when the technology matures — Mueller predicts within about five years — he doubts it will serve fashion shoppers, because most fashion spending is driven by inspiration rather than transactional efficiency.

Mueller's three near-term predictions for AI in fashion: AI-driven content creation, deeper personalization including virtual try-on, and faster, more reliable, less frictional delivery and returns. "More inspirational, more personalized and less friction," he concluded. Zalando insiders say future gains will come from expanding data inputs, including the text conversations shoppers have with its online assistant, which the company is not yet analyzing but is likely to use going forward.