Fashion and machine learning don’t often make their way into the same conversation. Enter the new era of online shopping: Meet Aprizi, an etymological twist on the French word meaning to learn and the Italian word for prize. “We want you to discover on our site, and we want you to feel like you’ve won something,” explains one of the site’s co-founders Giff Constable.
The home page carousel displays four fashion items at once, and gives you the option to like and dislike each one. After twelve items, the carousel pauses to think, then returns with a new set of items that are smarter and catered towards your taste. This algorithmic approach to shopping is why people in New York City’s start-up scene are calling Aprizi, “the Pandora of online fashion.”
F**k it, we'll do it live!
Our biggest ever edition of TNW Conference is fast approaching! Join 10,000 tech leaders this May in Amsterdam.
The carousel refreshes slowly but trust me, it’s much faster than a sales attendant learning your taste based on which items you leave on the dressing room floor. While at first, I only liked about 10% of what was offered to me, after two rounds of liking and disliking, the carousel noticabely started to catch on to the fact that I like the color purple, prefer natural looking, funky jewelry and outfits with bold patterns in bright colors.
The site pulls from hundreds of different merchants and includes hundreds of different items, adding about 50 new items a day. Items are selected by Aprizi’s 10 curators who were cherry picked from hundreds of applicants. All items on the site have to meet 3 categories: They have to be special, beautiful or meaningful; they can’t come from a big brand or celebrity and they have to be for sale.
The algorithm learns based on certain repeating semantic tags such as eco-friendly, artisan, handmade and based on which curators you tend to graviate towards. If you dislike heels or chunky bracelets enough, it’s smart enough to pick up on that and not show you heels or chunky bracelets again. However the constraints are flexible to keep a bit of diversity in there, particularly because their user base, mostly women, can be so fickle.
Aprizi also has a neat blog that includes articles titled “Dress Me” for categories like “Warm Weather Escape,” and entries by other cool NYC fashion start-ups, like this recent one from the founders of Of A Kind.
Celebrating independent design isn’t cheap. Almost every item I liked was between $250-$500. So then I started browsing by price and specific category and found neighborhood’s worth of beautiful, affordable clothing and accessories. The site saves all of your likes, as well as items you find on other websites through it’s bookmarklet.
It’s fun to walk down the street in the East Village, but it’s not fun to shop on the web, and we want to change that. -Giff Constable
Co-Founders Giff Constable the design guy and CEO and Liz Crawford, the CTO with a PhD in Computer Science from Carnegie Mellon were introduced by venture capitalists intrigued by Crawford’s work in machine learning. They beta launched in late September 2010 and have been doubling users every month since. In October, they had 2,500 users which grew to 10,000 visits in December. The start-up is currently working within New York’s not-so-secret yet still hush-hush co-working space, General Assembly.
The early-stage bootstrapped business is talking to investors and plans to start making money in the traditional way, you know, by selling stuff. From their users, Giff and Liz are learning a lot about which designers are trending and what’s popular. They plan to use all of that data to sell curated, limited runs of certain items in the near future.