We talk a lot about speed in Ai, but the more interesting question is accuracy.

This article has been written by Natalie Mosley

We talk a lot about speed in Ai, but the more interesting question is accuracy. Not whether a model can produce something that looks plausible, but whether it can produce something that is right in the way retail needs right to be. Author and industry expert Jamie Bartlett’s point about hallucination is useful here because it cuts through the hype and lands on the real risk: Ai does not “know.” It predicts.


In a consumer context, prediction is often good enough. In a retail brand context, it can be expensive. A model can confidently invent a detail, misread a material, bend a logo, shift a tone, or quietly change the meaning of what an image is communicating. The result is not always obviously wrong at first glance. Sometimes it is worse than that. It is almost right. Close enough to pass a quick check, but wrong enough to chip away at trust when it appears across a product page, a campaign grid, or a marketplace feed.


Bartlett frames hallucination as a reminder that language models operate on patterns rather than understanding. In visual generation, we see the same behaviour, just expressed in pixels instead of words. The model reaches for the most likely answer, not the most faithful one. It fills gaps with what it has seen before. It tries to be helpful. It tries to complete the picture. And in doing so, it can create an output that is coherent, attractive and still misaligned with the brand reality you are trying to build.


This is why “just prompt it” is not a strategy. Prompting is not the work. Prompting is the interface. The work is knowing what the model will misunderstand, where it will drift and how to guide it back to the truth of the product and the truth of the brand. That guidance is not a single sentence. It is a system. It is reference, calibration, iteration and judgement applied with consistency.


At ACi Studios, our Advanced Creative Intelligence platform exists to make that system repeatable. The technology matters, but the bigger differentiator is that it runs inside a studio discipline. We are not asking Ai to guess what “premium” looks like for a brand, or what “summer light” means for a category, or how a material should behave when it moves. We define it. We set the boundaries. We create the conditions where the model is less likely to improvise.


In practice, that means we treat hallucination as a production problem, not a novelty. We expect it, we design around it and we build checks that catch it before it ships. We use precise direction, but we also use experienced eyes. Retail imagery is full of small signals that customers read instantly, even if they cannot explain what they are seeing. A shadow that feels wrong. A proportion that breaks believability. A surface that reads as plastic instead of fabric. A styling choice that is just off. These are the moments where Ai can lose the plot and they are also the moments where expert review makes the difference.


There is also a quieter form of hallucination that brands overlook. Style drift. The slow accumulation of “almost on brand” variations that, over time, creates an inconsistent visual world. A single image might be fine. A hundred images might start to tell a different story. In ecommerce, where customers compare products in grids and scroll at speed, consistency is not a nice-to-have. It is what makes a brand feel reliable.


So the ambition is not to eliminate Ai’s unpredictability entirely. The ambition is to harness its power without inheriting its randomness. When you combine Ai generation with studio led intent, you get the upside without the meme risk. You get faster production without letting the model define your aesthetic. You get scale, but you keep authorship.


Bartlett is right to remind us that these systems do not see the world as we do. The best use of Ai starts from that humility. It is not magic. It is a tool that needs direction. The brands that win will not be the ones that generate the most. They will be the ones that generate with the clearest intent, the tightest controls and the strongest craft behind every image that goes live.

By Nick Aldrich August 18, 2026
For decades, if a major retailer wanted great imagery, the route to getting it was pretty well established. Creative concept, samples, studio or location, photographer, assistants, models, stylists, sets, props, travel, production, retouching and then, hopefully, a library of beautiful assets at the end of it. Done properly, it can be brilliant. We know because we have spent more than 30 years doing it. But it can also be expensive. A significant campaign shoot can easily become a £50,000, £100,000 or considerably bigger production once all the moving parts are added together. Historically, brands accepted that because there simply wasn't another credible way of achieving the quality they needed. Ai changes that equation. Not because photography is dead. It isn't. Not because brands should suddenly stop shooting real products, real people or real locations. They shouldn't. Ai changes the equation because for the first time the £100,000 photoshoot becomes a business choice rather than a production necessity. And that is a much bigger change than simply making imagery cheaper. Imagine a retailer planning its next major seasonal campaign. There may be every reason to shoot the hero campaign traditionally. A particular photographer, model, location or creative idea might be fundamental to the brand. The experience around the shoot might generate PR, social content or cultural relevance that goes far beyond the images themselves. Great. Shoot it. But then ask a different question. Does every other image surrounding that campaign need to be produced in the same way? Does every PDP image, lifestyle variation, social asset, marketplace image, CRM execution, regional adaptation, seasonal version and alternative crop really need another physical production? Until recently, the answer was often yes, because there wasn't a serious alternative. Now there is. And this is where we think the conversation around Ai imagery has been slightly misunderstood. Everyone talks about cost reduction. We do too. At ACi Studios we can produce commercial grade imagery at a fraction of the cost and time of traditional production. But saving money is only the first part of the story. The more interesting question is what brands choose to do with the money, time and creative capacity that Ai releases.  If producing ten additional lifestyle images used to require another shoot, another location and another chunk of budget, the obvious response was to produce fewer images. Marketing teams became very good at deciding what they could live without. Ai turns that around. Instead of asking, "How much content can we afford?" brands can start asking, "How much content would actually help us sell?" That is a fundamentally different question. Perhaps a sofa doesn't need one beautifully art-directed lifestyle image. Perhaps it needs ten. Different rooms. Different seasons. Different customer profiles. Different social formats. Different environments for different international markets. Perhaps a fashion product should appear in multiple contexts rather than one campaign execution. Perhaps a homeware collection should evolve visually throughout the year rather than waiting for the next scheduled shoot. Perhaps creative teams should be able to test ideas rather than rejecting them because producing the assets would cost too much. When the marginal cost of creating another high-quality image falls dramatically, the creative possibilities expand dramatically with it. That doesn't mean producing endless content for the sake of it. More isn't automatically better. The opportunity is to produce more of the right content. That distinction matters because modern retail has developed an enormous appetite for imagery. Ecommerce, marketplaces, social media, paid advertising, CRM and international trading have created hundreds of places where a product needs to show up, often in different formats and with different creative requirements. Traditional production was never really designed for that world. It was designed around shoots. Ai production can be designed around products. Once you make that shift, something else happens. The relationship between the marketing calendar and the production calendar begins to change. A retailer no longer necessarily needs to wait until every sample has arrived, every location has been booked and every member of the production team is available before content creation can begin. With accurate product reference and a controlled Ai workflow, imagery can start being created much earlier. Products can potentially be marketed while stock is still travelling through the supply chain. Campaign directions can be explored before physical production would traditionally have started. Additional assets can be created in response to what is actually happening in the market rather than what everyone predicted three months earlier. And if something changes, you don't necessarily reshoot it. You change it. That might ultimately prove to be one of the biggest differences of all. Traditional photography creates an image. Ai can create a visual system around a product, one that can evolve as the requirements of the business evolve. None of this removes the need for creative judgement. In fact, we think it makes judgement more important. When producing imagery becomes easier, deciding what imagery should exist becomes harder. Brands still need art direction. They still need taste. They still need people who understand customers, products, composition, lighting, materials and what makes an image feel unmistakably theirs. Ai doesn't replace any of that. It simply gives those people a much bigger canvas to work with. At ACi Studios, that is how we see the future developing. Not photography versus Ai. Not old versus new. And certainly not technology replacing creativity. It is about choosing the right production method for the right job. Spend £100,000 on a photoshoot when the idea deserves a £100,000 photoshoot. But don't spend £100,000 simply because that's what it has historically cost to create the imagery a modern retail business needs. For the first time, brands have a genuine choice. And we suspect the businesses that understand the difference will be the ones that do the most interesting things with it.
By Nick Aldrich August 18, 2026
For decades, if a major retailer wanted great imagery, the route to getting it was pretty well established. Creative concept, samples, studio or location, photographer, assistants, models, stylists, sets, props, travel, production, retouching and then, hopefully, a library of beautiful assets at the end of it. Done properly, it can be brilliant. We know because we have spent more than 30 years doing it. But it can also be expensive. A significant campaign shoot can easily become a £50,000, £100,000 or considerably bigger production once all the moving parts are added together. Historically, brands accepted that because there simply wasn't another credible way of achieving the quality they needed. Ai changes that equation. Not because photography is dead. It isn't. Not because brands should suddenly stop shooting real products, real people or real locations. They shouldn't. Ai changes the equation because for the first time the £100,000 photoshoot becomes a business choice rather than a production necessity. And that is a much bigger change than simply making imagery cheaper. Imagine a retailer planning its next major seasonal campaign. There may be every reason to shoot the hero campaign traditionally. A particular photographer, model, location or creative idea might be fundamental to the brand. The experience around the shoot might generate PR, social content or cultural relevance that goes far beyond the images themselves. Great. Shoot it. But then ask a different question. Does every other image surrounding that campaign need to be produced in the same way? Does every PDP image, lifestyle variation, social asset, marketplace image, CRM execution, regional adaptation, seasonal version and alternative crop really need another physical production? Until recently, the answer was often yes, because there wasn't a serious alternative. Now there is. And this is where we think the conversation around Ai imagery has been slightly misunderstood. Everyone talks about cost reduction. We do too. At ACi Studios we can produce commercial grade imagery at a fraction of the cost and time of traditional production. But saving money is only the first part of the story. The more interesting question is what brands choose to do with the money, time and creative capacity that Ai releases.  If producing ten additional lifestyle images used to require another shoot, another location and another chunk of budget, the obvious response was to produce fewer images. Marketing teams became very good at deciding what they could live without. Ai turns that around. Instead of asking, "How much content can we afford?" brands can start asking, "How much content would actually help us sell?" That is a fundamentally different question. Perhaps a sofa doesn't need one beautifully art-directed lifestyle image. Perhaps it needs ten. Different rooms. Different seasons. Different customer profiles. Different social formats. Different environments for different international markets. Perhaps a fashion product should appear in multiple contexts rather than one campaign execution. Perhaps a homeware collection should evolve visually throughout the year rather than waiting for the next scheduled shoot. Perhaps creative teams should be able to test ideas rather than rejecting them because producing the assets would cost too much. When the marginal cost of creating another high-quality image falls dramatically, the creative possibilities expand dramatically with it. That doesn't mean producing endless content for the sake of it. More isn't automatically better. The opportunity is to produce more of the right content. That distinction matters because modern retail has developed an enormous appetite for imagery. Ecommerce, marketplaces, social media, paid advertising, CRM and international trading have created hundreds of places where a product needs to show up, often in different formats and with different creative requirements. Traditional production was never really designed for that world. It was designed around shoots. Ai production can be designed around products. Once you make that shift, something else happens. The relationship between the marketing calendar and the production calendar begins to change. A retailer no longer necessarily needs to wait until every sample has arrived, every location has been booked and every member of the production team is available before content creation can begin. With accurate product reference and a controlled Ai workflow, imagery can start being created much earlier. Products can potentially be marketed while stock is still travelling through the supply chain. Campaign directions can be explored before physical production would traditionally have started. Additional assets can be created in response to what is actually happening in the market rather than what everyone predicted three months earlier. And if something changes, you don't necessarily reshoot it. You change it. That might ultimately prove to be one of the biggest differences of all. Traditional photography creates an image. Ai can create a visual system around a product, one that can evolve as the requirements of the business evolve. None of this removes the need for creative judgement. In fact, we think it makes judgement more important. When producing imagery becomes easier, deciding what imagery should exist becomes harder. Brands still need art direction. They still need taste. They still need people who understand customers, products, composition, lighting, materials and what makes an image feel unmistakably theirs. Ai doesn't replace any of that. It simply gives those people a much bigger canvas to work with. At ACi Studios, that is how we see the future developing. Not photography versus Ai. Not old versus new. And certainly not technology replacing creativity. It is about choosing the right production method for the right job. Spend £100,000 on a photoshoot when the idea deserves a £100,000 photoshoot. But don't spend £100,000 simply because that's what it has historically cost to create the imagery a modern retail business needs. For the first time, brands have a genuine choice. And we suspect the businesses that understand the difference will be the ones that do the most interesting things with it.
By Nick Aldrich August 4, 2026
Brands don’t reject Ai imagery because they dislike technology. They reject it because they dislike risk. If you’re responsible for a brand, “pretty good” isn’t a win. Consistency is the win. Control is the win. Trust is the win. And most teams have learned the hard way that new production methods only look easy right up until the moment something goes live that’s slightly off, the material reads wrong, the proportions drift, the styling signals the wrong price point, or the image simply doesn’t feel like it belongs in the world you’ve spent years building. So the decision to adopt Ai usually comes down to one thing, confidence. Not confidence that Ai can make an impressive image in general. Everyone has seen that now. The real question is whether it can make your image, for your product, under your standards, without quietly weakening the brand. That’s why “proof” matters so much in this category. And it’s also why most proof is the wrong kind. A gallery can be curated. A deck can be polished. A case study can be the best day in the best lighting with the easiest product. None of that tells you what happens when the brief is yours, the product is yours and the bar is the one you actually work to. The proof of the pudding is always in the eating. In retail imagery, that means seeing the result on your own product, with your own brief. That’s exactly what our free ACi Studios test image offer is designed to do. We invest in that process because we know this is how brands make decisions. Not by being told. By seeing. By comparing the output to their current imagery and asking a simple, practical question, would we publish this? Would we put this on a PDP? Would this sit in our grid without calling attention to itself for the wrong reasons? If the answer is yes, the Ai conversation changes instantly. It stops being “is this real?” and becomes “how fast can we scale this?” Because then you’re not buying into a promise. You’re responding to evidence from your own context. And the context matters, because Ai has levelled execution, not judgement. The advantage now isn’t access to tools. It’s direction. Art direction. Brand understanding. Customer psychology. Knowing what must stay consistent, what can flex and what signals quality in your category. Ai can generate options, but it can’t own that responsibility. At ACi Studios, that’s our lane. We’re not asking brands to adopt Ai as a leap of faith. We’re giving them the most honest trial possible, your product, your brief, our output, free, so you can experience what “great” looks like when Ai is guided by real studio judgement. Because when a brand chooses to adopt Ai, what they’re really choosing is a new way to produce at speed without losing control. And the fastest way to know if that’s possible is to see it working on your world, not someone else’s.
By Giles Mosley July 29, 2026
For years, world-class visual content was a privilege of scale. Big teams, big budgets, big production. If you were an established retailer, you could afford an imagery machine, studio infrastructure, specialist crews, locations, sets, logistics, retouching pipelines and the time it takes to make everything consistent. Ai changed that almost overnight. Now smaller brands can show up looking surprisingly polished. A two-person ecommerce team can generate lifestyle scenes, campaign-style visuals and social-ready assets in hours, not weeks. That’s exciting and it makes competition healthier. But it also creates a new pressure point for established brands, when everyone can execute, your advantage can’t be execution. The uncomfortable truth is that “looking good” is no longer a moat. Ai has democratised production and the floor has risen fast. When a challenger brand can produce “pretty good” imagery at volume, cheaply and test dozens of creative directions without booking a single shoot, it eats into the advantage established brands used to rely on ie the ability to outspend the market on content. This is why Ai can feel like it’s hurting bigger brands, not because their imagery has got worse, but because everyone else’s has got better. That leaves established brands with a fork in the road. One instinct is to compete the traditional way by doubling down, more shoots, more locations, more content, more speed, more teams, more oversight. But trying to win that way at the pace the market now demands can quickly become a multi-million-pound production problem, not just in shoot costs, but in the operational drag of approvals, reshoots, asset management, internal politics and the constant stress of keeping quality consistent across hundreds of deliverables. Traditional production can still be brilliant; it’s just no longer the only way to look good, and it’s rarely the most scalable way to look consistent. The stronger path is to use Ai, but with real direction. Ai isn’t the strategy. Ai is the camera. The strategy is the part that can’t be automated, art direction that decides what belongs in frame (and what never does), brand understanding that clarifies what you’re trying to make the customer feel, customer psychology that knows what builds trust and drives conversion and consistency systems that protect a brand across ranges, seasons and channels. Ai can generate, but it cannot judge. And judgement is now the advantage. Ai imagery often fails in a quiet way, it looks great in isolation, but it doesn’t hold a visual line over time. It produces “nice” images that are interchangeable and interchangeable is deadly for established brands because brand equity is built on recognition. Customers don’t remember “nice.” Customers remember coherent. The next era isn’t about producing more images, it’s about producing more images that do the right job, in the right way, every time. This is where ACi Studios sits. We work in the gap between what Ai can produce and what brands actually need. We use Ai intentionally inside a studio-led process built on over 30 years of delivering retail imagery at scale, so brands get speed without chaos, cost efficiency without sacrificing premium standards, work that still feels authored rather than “tool-made,” and brand control through consistent art direction, reference frameworks and expert review. Small brands now have access to the camera. Established brands stay ahead by having the better director. In a world where everyone has access to the same technology, technology stops being the advantage. Judgement does.
Show More

Share this article