Doji puts you
in the clothes
before you buy.

Here is a problem that sounds small and turns out to be enormous: you cannot try on clothes over the internet. You can look at a photo of a jacket on a model who is not you, order it, wait, and then discover it fits like a tent. Then you send it back. The clothes industry has quietly built an entire logistics apparatus around the fact that people guess wrong, and the returns pile up, and everyone pretends this is normal.

Doji, a New York startup, would like to fix the guessing. The pitch is that you give the app roughly six selfies and two full-body photos, wait about half an hour, and it produces a photo-realistic avatar - a version of you, rendered by the company's own diffusion models, that can then wear things. Not a cartoon. Not a Lensa-style stylization. You, in the clothes, before the card comes out of the wallet.

The name is the tell. Doji is Dorian plus Jim: Dorian Dargan, who worked on Apple's VisionOS and on Meta's Oculus Quest experiences, and Jim Winkens, a former researcher at Google DeepMind. They met on Twitter around 2022, discovered a shared interest in fashion, and started building side projects together. Naming your company after yourselves is either supreme confidence or a joke you can't take back later; in this case it doubles as the product spec.

The interesting claim Doji makes is not really about clothes. It's about taste. "People who have explored this idea before either haven't had the technology chops to make it good or the taste to understand what actually makes a person feel good about an image of themselves," Dargan has said. Virtual try-on has existed, in gimmick form, for years. The graveyard is full of features that let you slap a shirt onto a stiff mannequin version of yourself, look once, and never return. Doji's bet is that the failure was aesthetic, not technical - that the reason nobody came back is that the avatar looked wrong, and looking wrong makes you feel bad, and feeling bad is not a foundation for a shopping habit.

So the whole company hinges on a distinction that is hard to put in a spreadsheet: does the avatar look like you, in the way you'd want to be seen? If yes, the try-on becomes something you do for fun, not just utility. If no, it's another abandoned app icon. That's a lot of weight to hang on model quality and design judgment, which is precisely why Doji built its own diffusion models rather than renting someone else's.

The product is more social than a fitting room usually is. There's a feed of looks styled by tastemakers and brands. You can remix someone else's outfit, style one from scratch, or pull a product from anywhere on the internet via a browser extension and try it on instantly, saving to a wishlist with a tap. The clothes are the medium; the play is the point. Whether that's a durable behavior or a novelty is the open question, and it's the one the $14 million is meant to answer.

About that money. In May 2025, days after publicly launching on the App Store in more than 80 countries, Doji announced a $14 million seed round led by Thrive Capital, with Seven Seven Six participating. The order of operations is worth noting - ship first, prove people want it, then raise on the traction. It's a tidy little playbook, and it works better when the launch actually lands. Doji's did, at least well enough to get Thrive to write the check.

None of this means the fitting-room problem is solved. Doji still can't tell you whether a specific garment will fit your specific body - it visualizes, it doesn't measure - and the 30-minute, eight-photo onboarding is a meaningful amount of friction to ask of a consumer who is used to installing an app and poking at it for ninety seconds. The bet is that the payoff, a wardrobe that fits your actual face, is worth the setup. We'll find out. For now, Doji is the rare fashion-tech company where the most convincing thing about it is that its founders clearly care whether you like how you look.