📰 Key Takeaways

OpenAI announced two new global shopping features in ChatGPT, letting its conversational AI assistant dive deeper into helping users shop online. The first is “Virtual Try-On,” where users upload a selfie or full-body photo to visualize how clothing or accessories would look on them, with a new “Try On” button appearing on shopping results pages; users can also upload a photo of an item (like a screenshot from a website) and ask ChatGPT to simulate the try-on. The second is “Favorites,” which lets users save items they discover to an in-app Library for later reference — these saved items get stored alongside the try-on images. OpenAI says both features run on the newly released ChatGPT Images 2.5 model, which the company claims produces more natural lighting and richer material detail, follows editing instructions more accurately, and reduces image generation latency. Beyond that, ChatGPT can help with shopping in other ways too — users can describe a style they want to try and ask the AI to find the pieces needed to put it together, or upload a photo of a celebrity’s outfit and ask the AI to find the same or similar items available for purchase. This pushes ChatGPT further into territory long dominated by Pinterest and Google for fashion inspiration and product discovery. Worth noting: OpenAI previously tried rolling out an “instant checkout” feature but pulled back after it underperformed, and Google already launched a similar virtual try-on feature last year — signs that competition in this space is heating up, with it still unclear whether ChatGPT can become users’ go-to entry point for shopping inspiration.

💬 JudyAI Lab’s Take

OpenAI’s rollout of virtual try-on and favorites in ChatGPT formally plants a conversational AI assistant right at the decision point of online shopping — upload a photo, and you can see how the outfit actually looks on you.

What’s worth paying attention to here isn’t the feature list itself, but the fact that OpenAI chose to solve e-commerce’s old “can’t see how it looks on me” problem with generative imaging (ChatGPT Images 2.5), rather than beefing up search or recommendation algorithms. That suggests reducing decision friction might matter more than piling on more information. The original piece also mentions that OpenAI’s earlier instant-checkout attempt didn’t pan out, and Google already shipped a similar try-on feature last year — which tells you the jump from “conversation” to “transaction” has a much higher trust bar than it looks, and the competition is only getting hotter.

If you’re building AI products, here’s something worth sitting with: is your feature actually helping users make a decision, or just handing them more information?

📅 Source Info

🔗 Further Reading