Jess Weatherbed is a news writer focused on creative industries, computing, and internet culture. Jess started her career at TechRadar, covering news and hardware reviews.

Google is injecting more generative AI into its online shopping experience in Search. An upcoming feature for AI Mode will generate images of outfits and decor ideas based on user descriptions, to help people find visually similar products. Also launching is a new tool that allows people to virtually try on clothes.

The new AI Mode shopping feature will launch this fall in the US. It’s designed to provide a visual aid that can better guide what kinds of products should be recommended, compared to just using search descriptions. If users search for “a green flowy dress for a garden party,” for example, AI Mode will generate images of fake dresses in a variety of different styles that allow users to find the closest match to the dress they were envisioning. Google will then show the users real listings for visually similar dresses in online stores.

This seems like a better way to use AI as a shopping inspiration tool than browsing through AI images hosted on moodboard platforms like Pinterest, which don’t match users to a real-world equivalent that they can actually buy. The initial images that Google AI Mode will generate are still fake, however, which may cause some disappointment for users who can’t find an exact match for the AI clothing and decor suggestions they’ve fallen in love with.

For something that’s actually available today, Google is rolling out a virtual try-on tool in the US that allows shoppers to see what clothes look like on themselves by uploading a photo. After previously being introduced as a limited experiment in Search Labs, the feature is now launching in Search, Google Shopping, and product results on Google Images.

Users can access the feature by tapping on any clothing product results in Search, Shopping, or Images, selecting the “try it on” icon, and uploading a full-length body photo. Google’s AI will then show the user what they might look like wearing the apparel they’ve selected, giving users a more personalized experience than seeing those clothes modeled on someone else.