What OpenAI shipped
OpenAI rolled out virtual try-on inside ChatGPT on October 9. Shoppers browsing clothing and accessories now see a try-on button on eligible product listings. Tap it, take or upload a selfie, and ChatGPT generates images of that person wearing the selected item. Results can be saved to Favorites or organized into folders inside the ChatGPT Library for later comparison. The feature works on both the ChatGPT website and its mobile app, and it sits directly inside the same conversational flow shoppers already use to research purchases.
The launch extends a push OpenAI started last year with its shopping research feature, which helped users build personalized buying guides inside chat. Retail Dive's reporting frames this as another step to court both shoppers and retailers onto OpenAI's platform rather than a standalone feature. The company is building out a full discovery-to-purchase loop inside ChatGPT, and try-on fills one of the more obvious gaps between browsing a product and feeling confident enough to buy it.
Catching up to Google, not inventing something new
This is not a new category. Google introduced virtual try-on for women's tops back in 2023, extended it to dresses in 2024, and added shoes last year. OpenAI's version closes a gap that Google has held open for three years, and the company arrives with a feature set that is functionally comparable rather than ahead of it. Shoppers gain a second credible place to do this kind of research, which is good for adoption, but the underlying capability itself is now table stakes between the two platforms rather than a differentiator for either.
The real signal for a CTO sits in what catching up reveals about intent, more than in the technology itself. OpenAI is now willing to match Google feature for feature on retail-specific functionality, inside a product with enormous existing consumer reach through ChatGPT's user base. That is a direct contest for the same real estate: the moment a shopper decides what to buy, before they ever land on a retailer's own site. Retailers who assumed only Google mattered here now have two gatekeepers to satisfy instead of one.
The retailers already building inside ChatGPT
Some retailers are already treating ChatGPT as a real storefront rather than an experiment. David's Bridal set up shop on Shopify's Agentic Storefronts for both ChatGPT and Microsoft Copilot in April. Etsy followed the next month with its own dedicated ChatGPT app. Both moves predate this try-on launch, which means those retailers will likely have try-on availability sooner and with fewer integration surprises than retailers starting from zero once the feature is already live and visible to every ChatGPT user browsing their category.
Most retailers still treat conversational AI platforms as a fifth or sixth sales channel, worth a pilot but not a budget line. That framing was already questionable before try-on shipped. It gets harder to defend now that the feature directly affects conversion on a channel increasingly used for pre-purchase research, and where a competitor's listing already renders a customer's photo convincingly while yours does not. The engineering lift to integrate is modest compared with most roadmap items competing for the same quarter, which makes the continued hesitation less about capability and more about which team owns the decision to prioritize it at all.
Why try-on converts better than browsing
The financial case for prioritizing this is not speculative. A 2024 Perfect Corp survey found that brands using virtual try-on saw their sales conversion rate more than double, and 38.4% of respondents said virtual fitting rooms were the closest thing available to recreating physical shopping online. For apparel and accessories specifically, uncertainty about fit is consistently the largest driver of both cart abandonment and post-purchase returns, so a feature that resolves that uncertainty earlier in the funnel moves the needle on both ends of the P&L at once, not just on top-line sales.
That effect should be larger inside ChatGPT than on a typical product page, because the shopper arriving there is usually already mid-research rather than mid-browse. They have asked a question, compared options, and are closer to a decision than a cold visitor landing on a category page. A try-on result that lands well at that point in the journey has an outsized chance of converting, which is exactly why retailers slow to integrate are not just missing a feature, they are missing their highest-intent traffic.
The catalog debt problem this exposes
Virtual try-on is only as good as the product data behind it. Rendering someone convincingly wearing an item requires clean, consistent, high-resolution imagery across colorways and sizes, along with sizing and fit metadata that most mid-market retail catalogs simply do not have in usable form. Years of treating product photography as a merchandising afterthought, shot inconsistently across seasons and vendors, now shows up as a direct, visible quality gap the moment that catalog feeds an AI model instead of a static product grid.
That gap will be visible to customers in a way old catalog debt never was. A shopper comparing two near-identical jackets inside ChatGPT will see one render cleanly and one render poorly, and will draw a conclusion about brand quality that has nothing to do with the garment itself. Retailers who deprioritized catalog standardization as a low-value back-office task are about to discover it is now a front-line conversion variable, visible in real time, next to a competitor's listing.
What this means for your product data roadmap
The practical move is to audit your product imagery and metadata against what both Google's and OpenAI's try-on systems actually require, not against your existing merchandising standards, which were built for a different purpose. Treat catalog photography and sizing data as infrastructure that your AI discovery strategy depends on, with a budget and an owner, rather than a marketing asset that gets refreshed opportunistically. Retailers waiting for a platform mandate before investing here are choosing to find out the cost of deferral after a competitor's listing already looks better.
Whichever of OpenAI or Google ultimately wins more of the shopping discovery layer, the retailers who benefit will be the ones whose catalog was ready before the feature arrived, not the ones scrambling to catch up once their competitors already render well inside the same chat window their own customers are using to decide what to buy next. Put catalog readiness on the same roadmap as any other agentic commerce initiative, with an owner and a deadline, and treat the next platform feature announcement as a test you have already passed rather than one you are scrambling to study for.
