Fashion has always been a driver for innovation — from the invention of the sewing machine to the rise of e-commerce. Like tech, fashion is forward-looking and cyclical.
At $2.2 trillion, the fashion sector is also one of the largest industries in the global economy. And today, fashion technology is growing at a faster pace than ever. From robots that sew and cut fabric, to AI algorithms that predict style trends, to VR mirrors in dressing rooms, technology is automating, personalizing, and speeding up every aspect of fashion.
We take a close look at the evolution of the fashion industry and where AI is taking it next, from automating design to style predictions and beyond.
Tech is automating fashion design processes
Fashion brands of all sizes and specialties are using technology to understand customers better than ever before. As those data collection efforts grow more sophisticated, artificial intelligence will reshape brands’ approach to product design and development, with a focus on predicting what customers will want to wear next.
Outside of fashion, manufacturers are already using AI to generate out-of-the-box prototypes for products ranging from aircraft parts to golf equipment. Generative design software is expected to be a $44.5B market by 2030, per CB Insights’ Industry Analyst Consensus.
Google has tested the waters of user-driven AI fashion design with Project Muze, an experiment it deployed in partnership with Germany-based fashion platform Zalando in 2016. The project trained a neural network to understand colors, textures, style preferences, and other “aesthetic parameters,” derived from Google’s Fashion Trends Report as well as design and trend data sourced by Zalando. From there, Project Muze used an algorithm to create designs based on users’ interests and aligned with the style preferences recognized by the network.
AI to become co-designer
Amazon is innovating in this area as well. One Amazon project, led by Israel-based researchers, would use machine learning to assess whether an item is “stylish” or not. Another, out of Amazon’s Lab126 R&D arm in California, would use images to learn about a particular fashion style and create similar images from scratch.
If that sounds like “fast fashion by Amazon,” that’s because it probably is. In 2017, the e-commerce giant patented a manufacturing system to enable on-demand apparel-making. The tech could be used to support its Amazon Essentials line or the suppliers in Amazon’s logistics network. Of course, the outcomes of human-free AI design aren’t always runway-ready. Many designs created for users of Google’s Project Muze were unwearable scrawls and scribbles, while some reports on the Amazon Lab126 initiative called the design results “crude.”
Furthermore, using algorithms to generate clothing has backfired at times. In 2019, for instance, it was unveiled that a number of online T-shirt vendors were deploying bots to scrape images (under which people had commented “I want this on a T-shirt” or the like) and uploading them to marketplaces to be produced and sold on-demand. This quickly drew criticism and allegations of copyright violation and IP theft.
Nevertheless, the gap between AI-developed designs and human-made ones is closing. In April 2019, an AI “designer” called DeepVogue placed second overall and won the People’s Choice Award at China’s International Fashion Design Innovation competition. The system, designed by China-based technology firm Shenlan Technology, uses deep learning to produce original designs drawn from images, themes, and keywords imported by human designers.
Clearly, more R&D is needed before brands rely on AI-only designers. But artificial intelligence is already helping brands create and iterate their designs more quickly.
AI is influencing brands
In 2018, Tommy Hilfiger announced a partnership with IBM and the Fashion Institute of Technology. The project, known as “Reimagine Retail,” used IBM AI tools to decipher:
- Real-time fashion industry trends
- Customer sentiment around Tommy Hilfiger products and runway images
- Resurfacing themes in trending patterns, silhouettes, colors, and styles
Knowledge from the AI system was then served back to human designers, who could use it to make informed design decisions for their next collection.
Stitch Fix is already at the forefront of AI-driven fashion with its “Hybrid Design” garments. These are created by algorithms that identify trends and styles missing from the Stitch Fix inventory and suggest new designs — based on combinations of consumers’ favorite colors, patterns, and textiles — for human designers’ approval. The company details how it works (shown below) in the “Algorithms Tour” on its website.
Stitch Fix has developed over 30 pieces of apparel using the Hybrid Design methodology. The company has said that the AI-designed pieces perform comparably in “keeper” sales to the garments from its fashion- brand suppliers. That’s likely because Stitch Fix has such vast troves of customer data informing its AI, thanks to its subscription-based, feedback-focused business model.
“We’re uniquely suited to do this,” said Eric Colson, chief algorithms officer at Stitch Fix. “This didn’t exist before because the necessary data didn’t exist. A Nordstrom doesn’t have this type of data because people try things on in the fitting room, and you don’t know what they didn’t buy or why. We have this access to great data and we can do a lot with it.”
Design isn’t the only area where Stitch Fix is putting AI and machine learning initiatives to work. The company employs a team of more than 85 data scientists to oversee machine learning algorithms that are used to inform everything from client styling to logistics to inventory management. According to Colson, the company is already seeing ROI from its AI investments, including increased revenue, decreased costs, and improved customer satisfaction.
AI assistance tools advancing
As more and more AI “assistance” programs advance, they will help brands make smarter strategic decisions around product development and new business lines. 3D design platforms like CLO also make it easy to tweak designs on the fly. These allow brands to use real-time AI insights to modify fashions right up to the minute they hit production. Below, we illustrate how tech is automating away the fashion designer, as styles become more personalized and influenced by digital signals.
Similar to Amazon’s Lab126 initiative and Google’s Project Muze, scientists from UC San Diego and Adobe have outlined a way for AI to learn an individual’s style and create customized computer- generated images of new items that fit that style. The system could enable brands to create personalized clothing for a person based solely on their engagement with visual content.
At a more macro level, it could also allow a brand to predict broader fashion trends based on historical data from its entire user base. The predictions could ultimately be used to guide the design of a product or an entire label.
What is next?
The next era of fashion is all about personalization and prediction. With more and more data, algorithms will become trend hunters — predicting (and designing) what’s next in ways that have never been possible.
True Fit, for example, closed a $55M Series C round in 2018 to bring its total equity funding to $102M. The company’s big data platform facilitates capabilities like AI-powered fashion discovery and exact-fit clothing and shoe recommendations.
With over 100M registered users, the platform uses transaction data to determine customer preferences that “better personalize all touchpoints of the consumer journey” for brands, according to CEO William R. Adler. Virtusize, another company capitalizing on the smart fitting trend, enables online shoppers to buy the right size, either by measuring the clothes in their closet or by comparing specific brands and styles to their own.
Virtusize claims that, by removing uncertainty around size and fit, it can increase average order values by 20% and decrease return rates by 30%. The Japan-based company counts Balenciaga and Land’s End among its clients, as well as Zalora — a leading online fashion store in Asia.
Ultimately, consumer preferences will guide every aspect of the design and production process. Platforms like True Fit may help identify the types of materials shoppers prefer, or even pinpoint how important sourcing and manufacturing conditions are to a unique shopper.