Reflo is disrupting sportswear with sustainable design and redefining the creative and operational backbone of brand-building using AI. But where is the line between automation and authenticity, and who gets to draw it?
For a brand founded on sustainability, creativity, and a healthy disregard for industry convention, it is fitting that Reflo, the London-based sportswear disruptor, is building its future with artificial intelligence. At first glance, AI and apparel may not seem the most obvious pairing, but behind the scenes, the technology is rapidly transforming the company’s workflows, decisions, and ambitions.
Reflo Co-Founder Rory McFadyen describes his startup as “a speedboat among oil tankers,” which might be the perfect metaphor for how AI allows smaller players to outmanoeuvre far larger incumbents. “We realised early on that we could be a lot more efficient by using AI,” McFadyen explains. “We did not start by using it creatively. It began as a way to sense-check documents, analyse information faster than we could manually, and act as a first pass on things like contracts. But we quickly saw that it was about far more than operational savings; it was about punching above our weight.”
From AI-powered image generation and pitch decks to the creation of ghost mannequin videos and scalable ad testing, Reflo is embedding AI across every business unit. Unlike enterprise-scale incumbents’ heavy, centralised systems, the company’s approach is pragmatic, lean, and modular, driven as much by experimentation as by infrastructure.
A new kind of creative process
While generative AI is frequently touted as a design disruptor, Reflo’s creative adoption of the technology has been anything but automatic. The team first deployed generative AI to address a challenge common to smaller, global brands: how to launch a location-specific capsule collection creatively and cost-effectively without flying a whole creative team across the Atlantic.
Their campaign for the WM Phoenix Open, one of the world’s most sustainable golf tournaments, became a testbed for AI as a standalone creative engine. “The collection was based on an old underground bowling alley in Phoenix from the 1950s, and we had loads of reference photos,” McFadyen says. “But how do you bring that visual world to life without spending a fortune on post-production or flying a team out there? It was a question of sustainability and budget, but also creativity.”
Image generation tools such as ChatGPT’s 4.0 multimodal model and platforms like Kling were put to work transforming stills into videos and static packshots into dynamic on-model visuals. What emerged was not a shortcut but an entirely new creative discipline. “It was not quick, and it was not easy,” McFadyen admits. “It took hundreds of iterations to get just a few usable bits of video. The tech has already improved massively since then, but at the time, getting AI to replicate graphic prints on a polo shirt or generate accurate sports visuals was incredibly hard.”
It taught them how to develop a visual story with AI from concept through to execution without sacrificing brand identity. The team now uses mood boards and prompt engineering to ensure AI-generated visuals align with their evolving design language. “Brand consistency still comes down to creative direction,” McFadyen continues. AI is not a substitute for that. It is the same as working with a designer. You do not get a good output if you do not provide a good brief.”
Between the virtual and the real
One of the more provocative frontiers Reflo is exploring is using AI for product imagery, turning flat-lay garments into full-body model shots using synthetic models and inferred fabric dynamics. However, the team is cautious about how far it can go without compromising trust.
“There is a difference between showing an authentic image of how clothes fit a person and a stylised rendering,” McFadyen explains. “AI can get the pose, the setting, the lighting. But it does not know what the fabric weighs, how it drapes, and how it catches the light when someone moves. That nuance is critical in fashion. People want to see what they are buying, not a fantasy version of it.”
The challenge, then, is not just technological; it is ethical. The tension between accuracy and aspiration, between operational efficiency and customer expectation, is a tightrope all AI-augmented retailers must walk. Reflo is content to test these systems behind the scenes while keeping human oversight in the loop. “We have not used any of those AI-modelled images on the website yet,” he says. “But we are constantly experimenting. You must be agile and see where the tools can take you.”
A numbers game, but with nuance
While brand storytelling may still require a human hand, advertising is where AI is having its most measurable impact. “Paid marketing is becoming a numbers game,” McFadyen explains. “The more creative you test, the more winners you find, and the quicker you scale. This week alone, we have created 30 different ads using AI.”
The company uses this creative volume to test visual hooks, calls to action, and campaign structures that can be iterated and refined over time. If a particular concept performs well, it may justify a future physical shoot. If not, the brand has saved thousands in production costs while maintaining campaign velocity.
Beyond the numbers, this also allows for a more dynamic feedback loop between the audience and the brand. “If we find winners quicker, we can refine and scale faster. It changes the entire cadence of how we work,” he says. “It is not about AI replacing human creativity. It is about using AI to get us closer to the audience faster.”
The energy equation
For a brand that positions sustainability at the heart of its mission, the environmental implications of large-scale AI adoption are not lost on the team. As image generation models become more powerful and compute-intensive, questions about energy consumption and carbon impact are becoming more pressing.
“A lot is being written about how much energy is used to generate a single AI image,” says McFadyen. “But when you compare that to the alternative, eight people travelling to a heated, lit studio with high-powered lights, wardrobe, crew, and transport, it is not so straightforward. We are currently doing a deep dive into what is more sustainable. The problem is we do not know which servers these tools use. Some are renewably powered, and some are not. There is a lot of work to be done to make those comparisons meaningful.”
This is perhaps one of the most under-discussed aspects of AI adoption at scale: the lack of transparency in cloud infrastructure and its impact on a company’s Scope 3 emissions. For sustainability-led brands, it is not enough for the tools to work; they must also align with the organisation’s values. “The headline numbers do not tell the whole story,” McFadyen adds. But we are determined to understand it better.”
From startup experimentation to systems thinking
With AI now embedded across most parts of the business, Reflo recently hired a dedicated AI and Systems Manager to formalise processes, ensure compliance, and support scale. This reflects a shift from opportunistic experimentation to long-term strategic planning. “Every business unit now has a remit to explore how they can use AI to improve efficiency,” McFadyen continues. “But we are also putting guardrails in place. Data protection, tone of voice, and brand integrity still matter. You cannot just paste prompts into a chatbot and call it transformation.”
The company’s AI stack remains largely off-the-shelf, built on platforms like ChatGPT, Microsoft Copilot, and specialist image tools. However, the longer-term question of whether to build a more integrated AI-native backend is actively under discussion. “We are exploring how our tech stack needs to evolve,” McFadyen admits. “We are currently finding SaaS tools that plug into our existing systems. But we may get to a point where we must rethink the whole thing.”
The future of AI-native brand building
Perhaps the most profound insight McFadyen offers is that AI, for all its potential, is still not the idea engine. It is a catalyst, not a creator. “To build a brand, there still needs to be a human element,” he says. “What ChatGPT cannot do is originate a great idea. It can help refine one, test it, build around it, and even interrogate it. But the spark still needs to come from people.”
This distinction is critical for executives looking to apply AI across their organisations. While automation may improve outputs and reduce costs, the value of originality, judgement, and vision remains unchanged. In fact, as more companies adopt AI, human differentiators such as values, voice, and culture may become even more valuable.
“You can build faster and smarter with AI, no doubt,” McFadyen concludes. “But brands that succeed will be the ones that integrate AI into what they do, not the ones that build their identity around it. There is still no substitute for authenticity. And AI, for all its power, still needs a brief.”
