Large language models are already transforming the way consumers find and buy products, letting them ask questions in natural language to get more tailored recommendations than what they typically receive from traditional keyword searches.
That’s just the start. In September 2025, OpenAI announced deals with Shopify and Etsy to let shoppers buy from their platforms directly through ChatGPT, while Google and Perplexity have unveiled agents that can complete purchases on a shopper’s behalf.
At the moment, shopping is still just a fraction of consumers’ total use of AI. Researchers found 2.1 percent of a sample of 1.1 million messages sent to ChatGPT between May 2024 and July 2025 sought information about purchasable products. Still, that’s a significant amount of traffic given the authors said ChatGPT received 18 billion messages per week, and it continues to grow.
The situation changes the online playing field for fashion brands. If consumers now get a short list of product recommendations, rather than a long list of links to sources to sift through themselves, brands not appearing in that consideration set may as well be invisible. It’s kicked off a race to improve what’s known as answer-engine optimisation (AEO) or generative-engine optimisation (GEO), as opposed to traditional search engine optimisation (SEO).
Profound is among a crop of startups on the frontlines of assisting retailers such as Mejuri to adapt to this new environment. In August 2025, the company announced a $35 million funding round led by Sequoia Capital, with participation from Kleiner Perkins, Khosla Ventures, Saga VC and South Park Commons.
BoF: We hear a lot about the rise of AI platforms, but overall, they’re still a small share of total traffic to retailers. How much do brands really need to think about how they’re showing up on these platforms?
James Cadwallader: You’re speaking to someone extremely biased, but I think this is the biggest platform shift maybe in the history of the internet. The front door of the internet is changing for the first time in 25 years. The brands that are discussed and covered via AI will naturally receive billions of eyeballs over the coming years. Today, I think we see typical [traffic] referral rates of around 10 percent from AI answer engines like ChatGPT. These numbers are only increasing.
BoF: Are there any factors that you see driving more consumers to use AI platforms for shopping?
JC: It’s fantastic for comparison. If you imagine that the old paradigm was that you’d have 10 tabs open in your Chrome window, and you’re comparing one against the other, you can literally feel your rote human cognition working. That all has just shifted now into a world where the entire funnel is being compressed into a ChatGPT window. You might ask for help in finding a new pair of sneakers and it would give you a shortlist. Then, you narrow down that shortlist to maybe two options that you’re thinking about.
BoF: Are there other ways the shopping journey could change as these platforms evolve?
JC: I think we are going to see a platform like ChatGPT emerge with the personalisation of a social-media feed, because they just capture so much intent through these interactions that just didn’t exist in a Google search. You tell these models so much more about yourself, and so much more about what you’re looking for and what you’re thinking. That allows them to tailor the results a lot more.
BoF: Are there things brands should be doing for what’s often called answer-engine optimisation (AEO) or generative-engine optimisation (GEO) that differ from what they’re already doing to ensure they’re optimised for traditional search (SEO)?
JC: The way to think about it is you are creating content for bots, not humans. In the old world, let’s just say a brand like Chanel would create web pages for humans to read, but the future of the internet looks like Chanel creating lots of content for AI crawlers to hoover up into their results. Content is essentially machine optimised, it’s far more structured, it’s information dense. You wouldn’t pay as much attention right now to imagery, for example, because it’s the words that matter. I’m sure that’s going to evolve and change over time. There’s [also] technical nuance: JavaScript is invisible to the crawlers that are retrieving information. The way that you structureyour website, the way that you actually build your website will change.
BoF: Is it realistic to expect AI answer engines to overtake traditional search at some point, or will consumers use both to gather as much information as they can?
JC: I think it would be over-dramatic to say that Google search and blue-link search is going to disappear off the face of the earth in the next 12 months. I think as these consumer platforms like ChatGPT get better and the shopping experiences improve, which we are expecting them to, we’re going to see a lot of consumer behaviour [shift]. As a techno optimist, I tend to bet on the best technology will win over time.
You’re speaking to someone extremely biased, but I think this is the biggest platform shift maybe in the history of the internet.
BoF: The large language models (LLMs) that power AI answer engines can be a great way to learn what people online say is the best electric toothbrush or the best mountain bike — functional products that you can rank based on performance. But fashion is very individual, it has to spark an emotional connection and there’s often an aspirational aspect to it. Do you think consumers will use LLMs to search for fashion?
JC: The word I would rotate around here is ‘search.’ I don’t know if they would use LLMs to discover fashion. These conversational interfaces are not the best place to stumble across something beautiful. However, let’s say you had decided that you were in the market for a Chanel 2.55. Absolutely you’re going to ChatGPT to ask about the provenance of the bag, why is the Chanel 2.55 bag so iconic, what’s the history of it, where did it come from and how much is it, what should I look for when I’m buying one, how do I make sure I don’t buy a replica? All of that part of the consumer journey, these AI answer engines are perfectly suited for that.
BoF: If consumers use LLMs in greater numbers, does it create a winner-takes-all environment where the brands and products the AI recommends become the entirety of what consumers consider purchasing, and anything that doesn’t show up is invisible?
JC: We’re actually going to see the opposite be true. I think we’re going to see a huge, long tail of coverage because every answer is different. [Cadwallader types ‘what’s the best luxury sneaker for men right now?’ into ChatGPT in two separate browser windows.] In response one, the answer was Maison Margiela, Tom Ford, Burberry, D&G, Amiri, whereas in the second response, we had Brunello Cucinelli, we had Gucci. You see this probabilistic nature, the answers change.
BoF: AI is notoriously a black box, especially as models get more complex. How do you go about determining what works and what doesn’t for getting a brand to show up in AI search results?
JC: It’s by looking at where they’re going to build these answers. When we asked this question earlier about ‘what’s the best luxury sneaker for men right now’ what happened in the background is ChatGPT went to a bunch of different sources to build this answer. What we’re able to do is we can show the sources: where the models are going, what are the names of the websites, which websites are being used most frequently.
Lean in and take it really seriously, understand that there are billions of people using these platforms on a weekly basis. This is the new marketing channel.
BoF: A lot of the content these LLMs are drawing on is not under a brand’s control. How much are brands able to influence how they show up inAI answers?
JC: [Cadwallader shows a breakdown on Profound’s platform of where an LLM is finding information for a query about trail-running shoes.] You can see that Nike.com is the ninth most-visited site [by the LLM], so Nike.com is not that far below Reddit or YouTube. What this shows is that these models are going to Nike.com. Nike is actually controlling a lot of the narrative here.
BoF: When you look at the rise of LLMs and potentially agents that can actually take action on a consumer’s behalf, does that change the work of fashion marketers?
JC: At this point in time, at least — and I expect this will evolve and change as well — I think marketers in fashion right now should focus on description and lots of coverage around the quality of the product, having nice, structured text on their website that describes what makes their product special, that describes the look and feel in a very detailed way, that is designed to get pulled into responses. That’s how we’re working with fashion brands right now.
BoF: When we think about how AI is changing shopping, what do brands need to do to win in this new environment?
JC: Honestly, it sounds so trite, and so unimpressive, and it’s not like the magical response that I would imagine you’re hoping for, but it’s just lean in. Get set up, pay attention, start. Once you understand how you’re showing up in answer engines, which is a very self-serving statement, but that’s what we help you do, I think lean in and take it really seriously, understand that there are billions of people using these platforms on a weekly basis. This is the new marketing channel. It’s happened. This isn’t a speculation into the future. You shouldn’t put your entire 2026 marketing budget into your AI visibility strategy, but in the same vein, you should have a strategy.
This interview has been edited and condensed.
This article first appeared in The State of Fashion 2026, an in-depth report on the global fashion industry, co-published by BoF and McKinsey & Company.
