A sage field jacket with pleated flap pockets, a brown felt cowboy hat, a shell of silver paillettes and a pair of marbled flares slide across the screen, each cut out and laid flat like a packshot on a product page. In the app, cut-outs like these sit on tiles under filter chips for tops, bottoms and skirts. This is Google’s demo for Wardrobe, a Google Photos feature that catalogues clothes from the camera roll and lets people try on what they have already paid for. Since 24 September it has been open to every eligible user in the US, Brazil and India on iOS as well as Android. Retailers bought try-on to sell the next garment; Google has now aimed a version at the garments already sold.

The coverage keeps reaching for Clueless, and TechCrunch reads Cher Horowitz’s computerised closet as a Beverly Hills privilege now open to anyone. Amy Heckerling, who wrote and directed the 1995 film, described something humbler to NYLON in 2020. The closet began with her childhood paper dolls and the thought of taking “pictures of yourself in all of your outfits”, which she called “a low-tech idea stuck into a computer”. What that closet needed most was photographs, and somebody had to take them.

Closet apps inherited that chore. Indyx puts self-cataloguing at about two minutes a garment, and for $295 it will send an archivist to shoot flat lays of up to 100 items. Wardrobe skips the shoot by scanning four years of photos of the user, Google’s help page says. It needs Face Groups switched on and more than 1,000 photos of the account holder, a threshold waived for AI Pro and Ultra subscribers. The pictures Heckerling imagined taking on purpose were already in the camera roll.

The catalogue records what the camera saw, and that is a different list from what was bought. Kabir Jain of Android Police found a red H&M sweatshirt and white Nike trainers in his first generated outfit; a second, less restrained pass turned up sunglasses that were in fact a Snapchat filter he had tried on years earlier. It also keeps track of when things were worn, and the help page offers a sort by “recently worn”. That’s a gap no retailer’s purchase history closes: no single brand sees everything a customer owns, and none tracks which of it gets worn. Google has not said how long Wardrobe keeps that record, Jain noted.

Google now has two “Try it on” buttons, and they end in different places. In Search, the button sits on product listings, and a finished look comes with a prompt to “shop similar styles”. Android Police’s test of the Photos version found a finished result offers only options to send feedback, regenerate, share or save. Nothing on that screen leads to a shop.

The trade still treats try-on as a sales tool, and vendors such as Veesual pitch it to retailers on higher conversion and fewer returns. Jain found Wardrobe most useful before family functions, for deciding which of his kurtas went with which of his trousers. Google pitches the feature as a way to “rediscover long-forgotten items that might still be buried in your closet”.

A rediscovered skirt is a skirt nobody needs to buy.

The checkout is not far away. Eight days before Wardrobe reached the iPhone, Tapestry put Coach and Kate Spade behind an inline Buy button in the Gemini app and AI Mode, with payment through Google Pay. Tapestry’s chief information and data officer, Yang Lu, told Retail Dive that the company receives the same order data it would get from its own sites. Google says, as TechCrunch reported, that data on how people dress is not shared with third parties such as retailers, though that assurance comes from Gemini’s general privacy notice in Photos, Android Police points out, not a policy written for Wardrobe specifically. The brand sees the sale, and the closet the bag joins stays with Google.