ChatGPT can try on your products now. Can it find them?

TechCrunch reported the launch on 1 October 2026. A Try On button now appears in ChatGPT's shopping results. The shopper uploads a selfie or a full-body photo. ChatGPT shows how the clothing or accessory might look on them.

A shopper can also upload a screenshot of an item and ask to try it on. Favorites saves products to a Library in the app, next to the try-on images. Both run on OpenAI's new ChatGPT Images 2.5 model.

OpenAI also suggests other shopping uses. A shopper can describe a style and ask for the pieces to complete it. Or they can upload a celebrity outfit and ask where to buy the items.

Why this is about discovery

Six months earlier, OpenAI changed course on checkout. CNBC reported it on 24 March 2026. OpenAI said the first version of Instant Checkout was not flexible enough. It is letting merchants use their own checkout and focusing on product discovery.

OpenAI also said merchants can share product feeds and promotions. Target, Sephora and Nordstrom already supported the new discovery experience. We covered the checkout side in agentic checkout explained.

Read the two launches together. ChatGPT finds the product and helps the shopper picture it. Your site still takes the order. Try-on is a reason to click through, for products that showed up in the first place.

What try-on needs from your product

Try-on has to start from an image of your product. OpenAI has not said which image it uses. The likely source is the image in the shopping result. That comes from your feed or your product page.

So a clean product image matters more than it did. A flat lay on white is easy to place on a person. A crowded lifestyle shot with three items is harder. Give every variant its own image, and keep the main image to one product.

A generated try-on is still a simulation. It shows the cut. It cannot show how the fabric drapes after a wash. Customer photos answer that. Shoppers who just tried something on will ask about fit next. Reviews that mention fit, and photos from real customers, are what answers that question.

Five things to fix this month

1. 1Share your product feed. OpenAI said merchants can send feeds and promotions. Keep titles, variants, prices and images current.
2. 2Complete your Product schema. Include image, brand, offers with price, and every variant. Test it with our review schema validator.
3. 3Mark up your reviews. AggregateRating and named, dated Review blocks. Serve them in the HTML, so crawlers see them without JavaScript.
4. 4Let AI crawlers in. Check robots.txt does not block OpenAI's crawlers. Our AI visibility check tests this.
5. 5Publish an llms.txt. A short file that points AI crawlers at your key pages. Our llms.txt guide explains it.

Customer photos are worth tagging to the right product too. A photo tied to a SKU is evidence an engine can attach to that product. An untagged gallery is just pictures.

If you use a reviews or UGC app, ask where its content renders. Our guide to AI visibility tools for reviews and UGC compares six vendors on this.

Where Idukki fits

We make Idukki. It tags customer photos and videos to products. On Shopify it adds Product, Review and AggregateRating JSON-LD to product pages. Each store also gets an agent feed with an llms.txt file. Idukki does not send feeds to OpenAI for you. That step stays with you or your platform.