AI's quiet retail win is behind the camera

I spent the first part of my career making the pictures. Print design and technical artworking for food and drink brands, the sort whose packaging you have wheeled past ten thousand times in a supermarket aisle without once looking at it properly. So when a company tells me it has taken product photography from weeks down to minutes, I read it with two reactions firing at once. One half of me knows exactly which weeks it means: the studio booking, the sample that turns up creased, the reshoot, the retouch queue that never gets shorter. The other half reaches for the eyebrow.

ABOUT YOU Group launched SCAYLE STUDIOS on 20 July, and it is now selling the thing as software to other brands. You upload one product image, choose a model, style the outfit, pick a set and a pose, and it generates the full set: product-page stills, social, video, campaign. What took weeks is done in minutes at roughly 90% lower cost. Inside ABOUT YOU, around 90% of e-commerce production now runs this way, saving the group more than €8 million a year in studio and editorial costs. Co-founder and co-CEO Tarek Müller frames it as taking the fix they built for themselves and renting it out: "we're bringing the technology that solved this bottleneck at ABOUT YOU to other brands and retailers."

The numbers that actually pencil out

Here is why I take it seriously, and why you should too. Most of the AI aimed at shoppers this year has underwhelmed. The very fortnight SCAYLE STUDIOS launched, CI&T's UK research found that 68% of British shoppers have used an AI agent, but overwhelmingly to compare and research rather than to buy, and the US craft retailer Michaels reported the same lopsided split from its own assistant. The demand side keeps saying browse, not delegate.

Product photography is the opposite kind of problem. It is bounded, it is expensive, it is measurable, and nobody has to trust a chatbot for it to pay off. ABOUT YOU says its AI-generated images beat traditional studio shots by 9.2% on GMV in A/B testing, with add-to-basket up 5.1%. Those are the company's own numbers, from its own tests, and I would want to run the experiment on my own catalogue before I believed the exact figure. But the shape of it is right. This is the AI spend that pencils out: the unglamorous, back-office, provable kind that the customer never actually sees. If you run a UK fashion business and your competitor is taking a 90% cut on content production while you are still booking studio days, that gap compounds quietly every week.

So far, so good. And then it gets complicated in three directions at once.

The shopper nobody asked

The first catch is a good one, because SCAYLE walked straight into it with its own research. The same group that sells this tool publishes an annual survey of how shoppers feel about AI in fashion, and the UK number is the bluntest thing on the page. Nearly half of British shoppers, 48%, say they will not use AI to shop for clothing at all, the highest rejection rate of any market it surveyed.

UK

48%

Germany

38%

US

37%

China

10%

South Korea

8%

Britain is the most AI-resistant fashion market in the world, and SCAYLE is admirably clear about the reason. The resistance, its report says, "isn't about lack of awareness" but "about agency and transparency." Shoppers want AI to be optional, to leave them in control, to be honest about when it is steering them. Read that back against what SCAYLE STUDIOS actually does and the tension is hard to miss. The shopper who tells a survey she does not want AI making her fashion decisions is, right now, browsing product pages whose images an AI made, and no one has mentioned it to her.

I want to be fair here. An AI-generated packshot is not a con by itself; a good one is more accurate than a rushed studio shot on a bad lighting day. But "the customer would object if they knew, so we won't tell them" is not a position that survives being said out loud, and it is roughly the position the industry has backed into.

The law moved on Sunday

The second catch has a date on it. On 2 August, the EU AI Act's transparency rules took effect, the part that requires providers of generative AI to mark synthetic images so they can be detected as machine-made. There is a grace period to December for tools already on the market, and a carve-out for AI that only assists with standard editing rather than substantially generating the picture.

Whether a fully AI-generated shot of a real dress, on a model who never existed, counts as "substantial" is exactly the sort of question that sounds academic right up until a regulator asks it. I don't know yet how it lands for product photography specifically, and if I'm honest, nor does anyone: the Code of Practice that will define the actual labels is still being drafted. The obligation as written sits on the provider and is about machine-readable provenance, not necessarily a visible badge on your product page. But the UK has no equivalent rule in force, our own advertising code already frowns on materially misleading images, and any UK brand selling into the EU is inside the Act's reach regardless of Brexit. "We'll wait and see" is a strange posture to adopt about every picture on every product page you own.

The model and the photographer

The third catch is the oldest one, and the least comfortable. AI models are not new to fashion. Levi's began testing them back in 2023, Mango ran an entire teen campaign generated by AI, and Louis Vuitton and Nike have used them too. Each time, the same two objections came back. That a brand can now buy the look of a diverse cast without going to the trouble of hiring one. And that somewhere behind every generated shot there used to be a photographer, a stylist, a model and a studio who got paid for it.

I have built a career partly on the belief that technology done with care makes more than it breaks, and I still mostly hold it. But I have also spent enough evenings in the pub circling the question I can't quite put down, which is who exactly buys all this merch once the AI has taken the jobs that paid for it, to know that "faster content creation" is not a complete answer to anything.

There is a colder version of the same worry, and it should trouble a Head of Technology more than the ethics do, because it lands on the P&L. A product image exists to set an accurate expectation. If the generated version flatters the drape, the fit or the exact colour, the gap comes back to you as a return, and 65% of UK shoppers already say they feel let down when they can't return an online purchase easily. Returns are where online fashion quietly bleeds out. An image that sells 9.2% harder but comes back heavier in returns has not made you money; it has lent it to you at a rate you never agreed.

Worth being honest about where we are

So here is where I actually land, wearing the Head-of-Technology hat rather than the old artworker's one. I would use this. The production ROI is the most credible AI number I have seen all year, and turning it down on principle while a rival banks the 90% saving would be its own kind of malpractice.

But I would not switch the studio off and call it finished. I would mark the AI images honestly, in the product data and, where it matters, to the customer, before a regulator or a competitor makes the decision for me. I would watch the return rate on AI-shot lines the way you watch a database under load. And I would keep shooting real people often enough that the day the rules or the shoppers demand it, I still remember how. The picture got cheap this month. Trust didn't.

---

Data sources and further reading

• ABOUT YOU Group launches AI-powered photo studio with SCAYLE STUDIOS. ABOUT YOU corporate newsroom, 20 July 2026 (launch, figures, quotes; efficiency and GMV figures are the company's own A/B tests).
• About You launches Scayle digital AI studio to speed up photo/video production for UK brands. FashionNetwork UK, 20 July 2026 (UK-market framing).
• UK Fashion Retail Trends to Watch in 2026. SCAYLE (Fashion Retail Trends 2026 survey; the 48% / 51% / 2% figures and the "agency and transparency" framing).
• Many sceptical Britons still don't trust AI shopping. FashionNetwork UK reporting Visualsoft's "One Shopper, One Journey" research, 17 June 2026 (33% open to AI assistants; 65% returns-friction).
• The EU AI Act's Transparency Rules: A Practical Guide to Article 50. Future of Life Institute, on the 2 August 2026 obligations and the marking carve-outs.
• Why Mango is replacing fashion models with AI. RetailDetail EU (the AI-models debate; Mango, Levi's, Louis Vuitton, Nike).