When looking at all the places where Google continues to blitz AI, we can add shopping to the list. Users can now enter gen AI-style prompts to find and discover fashion items. This joins Google’s many ongoing moves to use multimodal AI – including text and visual search – to let users find apparel.
There’s a lot to unpack there. Starting with the latest, the new “Vision Match” feature lets users enter prompts to describe their desired fashion gear. In addition to returning an imaginative image, a la generative AI, Google will search its Shopping database to match the image with the most fitting gear.
One question that arises is how this differs from traditional search and Google Shopping. After all, it’s a search query and set of results to match that query. That’s true, but what’s different in Google’s latest is that natural language queries and full sentences can be entered – a longstanding Google evolution.
The other thing that makes it different is that an image is generated based on the user’s prompt. That image doesn’t exist or correlate to a real product but is presented to the user as the basis for the visual search… which then happens for real products. Think of it like GenAI meets reverse image search.
Create & Shop
To further illustrate the new feature, users can type product attributes into the search bar and then scroll to the “Can’t find it? Create it” prompt. They can also end up in the same place through a “Create & Shop” option on the left panel of Google’s Shopping tab. From there they can start entering prompts.
Using an example provided by Google (see image below), users can enter descriptive terms like “colorful midi dress with big daisies.” That prompt will then generate a made-up image, as noted above, which serves as a sort of quick feedback loop for users to validate that it’s what they’re looking for.
That image is then used to find the right real-life products using image matching and object recognition – similar to the tech that’s behind Google Lens (and reverse image searches, as noted). Altogether, it takes keyword search and fuses it with generative AI and natural language for a new spin on shopping.
This feature has been in development since 2023 and available as an experimental feature to Google Labs users. It’s not being rolled out broadly but only for mobile users in the U.S. Based on its performance, we expect to see it roll out to other platforms and geographies in the near to mid term.
True to Life
Sticking with some of the visual aspects of this latest development, Google has coupled the update with additions to its AR beauty feature. For those unfamiliar, this is an existing Google feature that lets users virtually try on cosmetics from a range of brands including E.L.F., Fenty, and Glossier.
With the latest development, they can try on multiple products simultaneously. This offers a more holistic and comprehensive simulation of beauty try-ons. It’s also truer to life, where users can try on lipstick and eyeshadow at the same time to see how they look together, rather than imagining them one at a time.
This update also infuses some of the advanced natural language search that’s noted above. Before getting to the virtual-try-on stage, users have to first find the items they want. And they can now do so by describing specific looks in natural language – rather than just color attributes – such as “soft glam.”
Lastly, Google is expanding its virtual try-on experience that lets searchers see how garments look on a variety of models and body types. Specifically, it has added several new brands and items to be compatible with this feature. Expect many more moves from Google in visual and AI-fueled shopping.
Header image credit: Firmbee.com on Unsplash
