Behind the scenes of Revolve’s AI-generated drop

Today, collections designed by the three winners of the inaugural AI Fashion Week will go live on Revolve. The teams behind the collections break down the creation process for Vogue Business.
Behind the scenes of Revolves AIgenerated drop
Photo: Courtesy of Revolve

To receive the Vogue Business newsletter, sign up here.

Today, three new collections will drop on Revolve. What do they have in common? They were all designed using artificial intelligence.

In April, AI creative studio Maison Meta, founded by Cyril Foiret, hosted the first AI Fashion Week (AIFW) in New York’s Soho neighbourhood. More than 400 participants from 50-plus countries submitted AI-generated collections for consideration by a panel of judges including Tiffany Godoy, Vogue Japan’s head of editorial content; Natalie Hazzout, Celine’s head of men’s casting; Erika Wykes-Sneyd, VP of Adidas’s Three Stripes Studio; Matthew Drinkwater, head of London College of Fashion’s Fashion Innovation Agency; and Michael Mente, Revolve CEO and co-founder.

Now, the three winning collections are for sale — in physical form — on fashion e-commerce site Revolve. Designed by AIFW winners José Sobral of Paatiff, Matilde Mariano of Molnm and Opé Majek of Opé, each collection consists of 10 to 12 pieces. Prices across the collections range from $228 to $1,598 and run from XS to L in sizing. The collections were made in Revolve’s atelier in Los Angeles. The process highlighted how AI might influence the go-to-market process for fashion collections, both in terms of design possibilities and efficiencies. The initial process for the Paatiff, Molnm and Opé collections began in July — just four months ago.

“Each designers’ collection was produced in small quantities and are priced on the higher end of the spectrum for us, as each item is very unique,” says Mitch Moseley, CEO of owned brands at Revolve, who oversaw the design and production process. He declined to comment on how the profits will be split between the designers and Revolve.

Photos: Courtesy of Revolve

The three winners — none of whom are trained designers (Sobral and Mariano were architects, Majek a stylist) — were the first to go through Revolve and Maison Meta’s AI Fashion Incubator, which refers to the process through which the winners were guided through creating a collection from their AI-generated designs. Maison Meta partner Nima Abbasi describes the incubator as a “holding hands” to give the winners a better chance at success, by alleviating cost (Revolve fronts the bill) and providing the necessary training. Following the competition, Revolve brought the three winners in to work with the brand’s internal teams to bring the pieces to life. “We wanted to expose the winners to the process of taking designs from the concept phase, to the actual creation and sale of a garment,” Revolve’s Moseley says.

The winners were assigned a project manager and given access to all aspects of Revolve’s supply chain and teams: the atelier team, product developers, pattern-makers and technicians. In partnership with these teams, the three winners sourced fabrics and perfected iterations of their designs. These were then produced in-house. (It was the first of two collections the winners will produce with Revolve under the year-long incubator programme.)

The process also illustrated AI’s current limitations — and offered room for creative workarounds. Some designs required “toning down” in the development stage to make them physically feasible to produce, Moseley says. Majek had to rethink some of her original AI designs. “In the initial concept, my collection leaned more towards fantastical, metaversal fashion,” she says. “I faced questions about how certain dresses could be translated into real-life garments and what their exact purpose was. When we created the physical samples, we made adjustments to ensure real-life functionality, comfort, ease and aesthetic on the fit model. The Revolve team was also very committed that the sample remained true to the AI garment sketch.”

Sobral, on the other hand, generated the AI designs with physicality in mind. “Even before winning, while still in the design phase, I constantly engaged with the technical aspects, contemplating garment construction, materials and exploring innovative alternatives,” he says, crediting his background in architecture to his technical approach.

Photos: Courtesy of Revolve

“Currently, when we generate AI sketches, they are flat, 2D sketches. I know that 3D is coming eventually, but today there is a front of the AI sketch and no back,” Majek says. She worked with Revolve’s teams to create the back of the dress on a tight timeline.

Sobral, Mariano and Majek also worked with Revolve teams on the marketing and editorial side to produce photoshoots and gain insight to the strategy of a collection launch, Moseley says. While the collections will be marked as AI-generated, the marketing won’t centre around this fact, he says, flagging that Revolve’s main focus is to bring consumers the most relevant product — irrespective of the design method.

For Revolve, the project is about testing and learning. “We have a massive team of incredibly talented designers, and we produce around 6,000 to 8,000 designs annually. If we can provide our designers another tool to help improve their process and expand their mind to better service our customer, it makes our business stronger,” Mosley says, noting that Revolve views AI as a supplemental tool — not a replacement for traditional design. He also notes that Revolve is exploring AI utilisation “across all aspects of the business”.

It was a lesson in efficiency. Though the design phase still takes time — it’s not just a matter of plugging a few words into an interface — once these are complete, the production process is far quicker. “What is unique is that the design process allows you to come up with an incredibly realistic view of what the design will look like at the initial concept phase,” Moseley says. “This is incredibly powerful as it can help eliminate design fallout and provide better direction for the development teams, making the organisation more efficient.”

The launch coincides with ‘Season 2’ of AIFW, which is set to take place in Milan 16-19 November, in partnership with photography platform PhotoVogue. They’ll also host a panel with Vogue China editor-in-chief Margaret Zhang on the final day. AIFW will then return to New York’s Spring Studios later in November. This time around, the team says, there’ll be more winners (they’re looking at five); clearer guidelines (to reduce misunderstandings and refine the production process); and more locations (Milan and New York). “We want to expand,” Maison Meta’s Foiret says, noting that they’re eyeing global growth.

Through AIFW, individuals with an interest in fashion design, who might lack the formal training, can participate in new ways. “Now, architects can become fashion designers; more people can step into that role,” Maison Meta’s Abbasi says on reducing barriers to entry. He’s also keen to emphasise the utility for trained designers and brands by reallocating resources. “Everyone in fashion is underpaid and overworked,” he says. “If this frees up some resources, that’s a great contribution of AI.”

Comments, questions or feedback? Email us at feedback@voguebusiness.com.

More from this author:

What would an IPO mean for Tory Burch?

Moncler Group revenues rise 16% on DTC sales

The Farfetch-YNAP deal is on. But how will the terms change?