In the past year, generative AI and Web3 have been competing for attention among tech hype cycles and brand innovation strategies. A new fashion company called Mmerch aims to combine the two.

Launching this autumn with an inaugural collection planned for the spring, Mmerch will offer limited-edition drops with unique, one-of-one clothing items designed using generative algorithms and linked to NFTs. The hope is that these relatively nascent technologies can be combined to reach a lofty goal: making individual clothing items accessible to the average consumer that provide value beyond the point of sale. It’s a high bar, especially when Web3 enthusiasm is waning, and manufacturers are trapped in old-school workflows that depend on the economy of scale.

Founder Colby Mugrabi calls the concept “neo-couture”. It means one-of-one products that meet the speed and scale of fast fashion. “I always give the analogy of, I can’t believe my mum only had five television channels on her TV growing up. I see a future not that far off where my daughter can’t believe I wore the same clothes as someone else,” she says, referring to mass production of the same design.

Mmerch is preparing for its first collection drop in the new year, positioned as a proof of concept. The collection will offer various approaches to hoodie designs for a total of 960 pieces. The designs were created using generative algorithms, based on the six pieces of a hoodie (hood, body, pocket, each sleeve and bands), with 20 different colours, 10 prints (such as camouflage or stripes) and two materials (French terry or cashmere). Similar to popular NFT PFP collections like Bored Ape Yacht Club and Cryptopunks, Mmerch designated various rarities for the design attributes, meaning that while there are no exact duplicates, certain colours or prints might show up more or less frequently.

The hoodies will be sold via a blind NFT drop, with a focus on both NFT art collectors and fashion fans. Once the designs are revealed to holders, they will be invited to claim their physical hoodie for a limited period. The hoodies come equipped with a QR code, an NFC chip and a soul-bound token that is associated with the garment.

Mmerch is experimenting with a number of technologies and concepts that could contribute to a more sustainable fashion industry: one-of-one pieces and on-demand manufacturing could create less waste by limiting overproduction of items that aren’t sold. As these technologies improve, they could enable more brands to convert to a design-sell-make model (opposed to today’s design-make-sell standard). Additionally, one-of-one pieces are often seen as more valuable and rare, suggesting they are more likely to retain value with owners and to the secondary market. Finally, digital identities can be used to authenticate items and share product details that can streamline resale workflows and inform recyclers of best practices once the garment is no longer worn.

While Mmerch’s goal is ambitious, it is not alone in experimenting with how AI can increase value and decrease waste in manufacturing fashion. During New York Fashion Week, emerging brands Bruceglen and Laura Garcia showed pieces produced by Resonance, a manufacturing company that uses proprietary AI, digital printing and automated cutting machines to make individual items of clothing after they are purchased. Today, H&M Group is introducing a generative AI design tool to its Creator Studio, that lets customers use text-to-image tools to mock up, then buy, pieces with custom artwork.

Beyond one-of-one manufacturing, generative AI design for fashion is still not the norm, but there is considerable appetite, says Andrew Wyatt, CEO of Cala, which was the first fashion company to get early access to OpenAI’s text-to-image product Dall-e. Collina Strada used generative AI to guide designs for New York Fashion Week, and retailer Revolve was a partner in AI Fashion Week to source generative AI design talent. Cala is seeing hundreds of new users each week, including teams from several of the top 25 global fashion brands as well as independent designers and fashion students, Wyatt says. Due to demand, it has added the option for people to instantly see new photorealistic designs on models and is testing the option for users to fine-tune the AI based on a brand’s proprietary assets and collective creative intelligence.

It’s novel still to provide customers with items that no one else has at a time when customer interest in individuality and personalisation are at an all-time high. “To me, the ‘scale’ part is the most interesting because it’s not necessarily new to provide people with one-of-one products — that goes back to the beginning of time,” Mugrabi notes. “But, what’s new and exciting to me is the ability to provide a lot of people with a one-of-one design.”

The challenge and promise of AI in design

Mugrabi studied fashion and art history before founding the popular fashion and art blog Minnie Muse in 2009, building on fashion week blog posts she wrote for Teen Vogue. More recently, Mugrabi became intrigued with the idea of applying the mechanics of generative art to physical products after the landmark sale of Beeple’s $69 million NFT artwork in March of 2021 (artist Mike Winkelmann is now a Mmerch advisor) and witnessing the success of generative art platform Art Blocks, which sells limited edition collections as NFTs. “I love contextualising information with where it stands in innovation of the past, present and future,” she says. “And I started thinking about how this entirely redefines or expands what it means to be an artist.”

While creating and refining the digital designs for the inaugural drop wasn’t easy — there are more than 732 million unique potential combinations based on the selected inputs — it turns out that translating the final selected hoodie designs into physical merchandise was even more challenging. Mmerch interviewed as many as 20 different factories to find a partner that was willing to change their standard workflows, which historically depend on the economy of scale, meaning that brands need to place bulk orders of the same items to be produced at once. Ultimately, she found two manufacturers: one for the cut-and-sew for the terrycloth, based in California, and one for the cashmere knitwear, based in Italy.

To solve the problem of one-of-one designs, Mmerch’s team developed a way to break down the entire collection into its individual parts, meaning that it instructed the manufacturers to, for example, cut a certain number of orange sleeves, red pockets and navy hoods, then gave instructions on final assembly. While it’s still more expensive to produce this way than to order 1,000 units of the same hoodie, it did make the process more efficient because the factories could cut multiple items from the same cloth at once, which is more similar to standard workflows. She also learned to limit the colours in the cashmere hoodie prints to four or fewer; otherwise, the weights of the materials would have been mismatched.

Even though her background was in fashion, Mugrabi was surprised at how complicated it was to produce individual items. “You kind of have to be slightly ignorant to these workflows to innovate atop an industry. To me, it was just like, ‘But why? Why can’t we do that?’”

There’s also the question of sizing. For the first drop, Mmerch has ordered one of each item in two sizes to ensure that the quality and production works (new owners can pick a size, and the other pieces will be repurposed). For future drops, the company plans to produce items after they are redeemed so that they are not holding inventory. Eventually, standard sizing might go the way of the pattern, thanks to AI.

Web3 continues the conversation

NFT collections with 10,000 avatars popularised the idea of generative art sold and resold as NFTs. By designing their hoodies using the same process and philosophy — inspired by the NFT generative art world — Mmerch is tapping into that phenomenon. However, with Mugrabi’s fashion-world network, she also hopes to welcome the traditional fashion consumer into the fold. For the genesis drop, she tapped comic book illustrator Gilbert Hernandez to create 48 original characters. Mmerch NFT holders will mint a unique artwork revealing one of the characters wearing one of the algorithmically generated hoodies, with only the hoodie’s colours and patterns visible in the art. Holders are then invited to redeem a physical version of that specific hoodie; upon redemption, the artwork because fully coloured in.

An NFC chip on each garment provides access to the soul-bound token associated with the piece. This essentially treats the item as a wearable wallet and the piece as the wallet’s seed phrase, Mugrabi says. As holders wear the garment at events, they might collect POAPs and other tokens that add to the garment’s story. Through this digital ID, Mmerch can continue to communicate with the wearer after the sale, and even collect incremental revenue and continue the conversation if the item changes hands on the secondary market.

Mmerch has partnered with fashion-tech firm Eon to provide unique QR codes for each garment’s interior label. Scanning the QR code provides access to information about the garment, including care instructions and details on the traceability and sustainability of the manufacturing process. Eon’s digital ID, somewhat like a food nutrition label, enables brands to attach provenance and material information to a garment. Once that ID is established, brands can add more features through the new Eon Exchange, which essentially hosts various “plug-in” integrations, such as those for resale websites or recycling programmes; Chloé and Vestiaire Collective have already established “instant resale” through Eon, for example.

The hope is that early adopter customers will help drum up buzz for the brand, leading to higher resale value. “I consider all of our collectors to be our affiliates, so there are no better promoters of the Mmerch brand than those individuals that got in early on,” Mugrabi says.

Other Web3-first fashion brands and startups have offered similar examples of using NFTs to transfer ownership and provide incremental revenue to brands. Phygital, Web3 fashion brand 9DCC enables holders to “vault” their purchases to retain resale value, while Cult & Rain linked up with Web3 marketplace LTD to fast-track transfer. Fashion-tech startup Mntge has retroactively attached NFTs to vintage goods, while Prada’s ongoing Timecapsule drops now all are attached to NFTs.

Mmercah’s inaugural drop is branded only as a Mmerch collection, but future collaborations with artists and brands are planned. Now that Mmerch has solved the workflow challenge, Mugrabi hopes that Mmerch will provide the “neo-couture” experience to fashion and art partners and introduce fashion natives to the world of Web3.

The current crypto “bear” market, in which even Bored Apes are tanking in value, is not a deterrent, she says, arguing it’s given Mmerch a moment to build and refine the concept. “Had we been in the midst of a bull market a year and a half ago, I would have felt more pressure to go to market, and it would have been a half-baked concept,” she says. The entry price for the first drop, while not yet determined, will be designed to be accessible. Hopefully, the crypto market will have slightly rallied, she adds.

While there are multiple barometers of success, a high mint price isn’t one of them, Mugrabi says. “I want to build a really excited community. I want to be over-indexed on demand and supply. I want to move product. I would rather mint out at a lower price and get people to redeem their physicals than sacrificing the potential success of the project by going to market with too high of a price. I think it shows that we understand the market and the collector mentality of the space.”

To become a Vogue Business Member and receive the Technology Edit newsletter, click here.

Comments, questions or feedback? Email us at [email protected].

More on this topic:

Buzz builds around made-to-order fashion manufacturing

Vault it, wear it or resell it? Inside Gmoney’s luxury label launch

The fashion exec’s guide to generative artificial intelligence