OpenAI puts a fitting room and a wishlist inside ChatGPT

A shopper on a retailer's product page takes a screenshot of a coat. She uploads it to ChatGPT with a full-body photo and asks to see it on ^[3][2]. If she likes the result, she saves the coat to her Library, where it sits next to the try-on image ^[6]. The launch coverage does not say whether retailers see any of what happens after the screenshot.

Digital Trends describes the intended loop. Shoppers discover items through try-on, decide by experimenting, then buy or save for later ^[12]. OpenAI also says ChatGPT will shop the pieces of a look the user describes, or find items for sale from a photo of a celebrity's outfit ^[8]. TechCrunch puts that in territory Pinterest and Google have held as fashion discovery that converts to sales for online retailers ^[5].

What users actually do is go to ChatGPT for ideas. According to Digital Trends, more than 1.2 billion people use it every week, and many already turn to it for shopping inspiration ^[14]. What commerce teams tend to tell themselves users do is finish the purchase in the same window. The one test of that so far went badly. OpenAI's Instant Checkout feature did not perform well, and the company had to pivot away from it, TechCrunch reported ^[10]. A single feature is a small sample, and it is the only evidence in these sources on whether people will pay inside ChatGPT.

Everything here depends on the render. OpenAI says Images 2.5 produces more natural lighting and richer textures, follows editing instructions more reliably and cuts generation latency ^[7]. Those are the company's claims. Digital Trends framed the open question as whether shoppers trust an AI's rendering of themselves ^[9]. Google has offered a comparable try-on since July 2025 ^[4].

Both features wait for the shopper to act, by pressing the try-on button or uploading a picture ^[2][3]. That design stays clear of the complaint against Instinct, the agentic AI startup whose proactive product recommendations struck some users as closer to ads than help ^[13].

In my view, commerce teams should treat ChatGPT as the place where shoppers consider and shortlist, and keep their checkout work on their own sites. The tradeoff is timing. OpenAI has tried in-chat checkout once ^[10], and if a better version comes back, teams that waited will be building late.

To size your own exposure, sort the catalog on two questions. The first is whether your products sell on how they look on a person; try-on covers clothing and accessories ^[2]. The second is whether your shoppers save items before they buy. Favorites keeps those saves in ChatGPT's Library ^[6], duplicating a function Digital Trends notes e-commerce sites already run ^[11]. Both yes puts you in the quadrant this launch was built for, and the figure to track is how quickly ChatGPT referrals to product pages convert. Looks without saving means try-on is your exposure. Saving without looks means Favorites competes with your wishlist. If neither applies, this launch changes little for your Monday.