How AI is changing the fashion industry

Published by

Of all the terms that spring to mind when we talk about the fashion industry’s upward trajectory, ‘deep learning’ and ‘big data’ probably aren’t the first to come to mind. Behind glitzy exteriors though, these errant terms are directing fashion brands in technologically-driven new directions.

This week, fashion AI startup Omnious revealed an investment of US$885K (over $250,000 US) from Korea Investment Partners, a deal the head of KIP said was hinged on the fact that “Omnious has the highest level of technical expertise in the deep-learning based image recognition field”. This expertise is deployed through the company’s visual search and image-recognition technology which will return a precise match or similar options from a fashion brand’s catalogue, as its ‘fashion-trained’ AI systems recognise item categories, colours, patterns, shapes and other details.

Ok, so what exactly does all this mean? Well without going into specifics about what ‘big data’ and ‘deep learning’ are (I’d love to but really, I’d only disappoint), in fashion terms the two phrases mean brands can now adjust supply and customer demand intelligently based on buyer’s behaviours, tailor customer service and use AI to simplify design processes. Here’s how...

The first taste of AI-assisted design was brought to us last year when Berlin-based fashion platform Zalando teamed up with Google and Stink Digital to develop Project Muze, a fashion ‘experiment’ as it were, which uses machine learning to create virtual fashion designs. The technology allows users to create 3D designs “inspired by you, designed by code”, using a neural network pre-trained by over 600 fashion experts and fueled by data from the Google Fashion Trends Report.

This technology is very much in its infancy (many of Project Muze’s ‘designs’ are unwearable scrawls) but teething problems aside, the experiment does set the bar from which other brands and platforms will now need to work off.


Elsewhere in the industry, fashion customer service is evolving beyond stuffy storefronts. Recently we shared an example of how Tommy Hilfiger and Burberry utilised chatbots during London Fashion Week allowing viewers to shop collections quicker than ever before. The medium was also used by Hilfiger in Fall 2016 when his brand launched its TommyXGigi [Hadid] collection and developed the TMY.GRL chatbot to boost engagement. Created in partnership with artificial intelligence platform, msg.ai, the chatbot replicated a concierge-style experience by allowing users to type questions or select pre-made queries, go behind-the-scenes of the collection and, of course, shop for items.

Other movers in the area of fashion customer service include AI-powered visual chatbot, mode.ai, which won a spot in the CB Insights AI 100 2017 (a ranking of the 100 most promising private artificial intelligence companies globally).

As for fashion’s supply and demand chains, these are undergoing huge adjustments after decades of stagnancy as companies such as Edited reinvent fashion retail. Edited has revolutionised traditional buying and merchandising teams with its real-time data flows which allow retailers such as Topshop, Net-A-Porter and Tommy Hilfiger to immediately respond to buying trends, better understand markets, optimise prices and stock, as well as ultimately selling more and discounting less. For us mere consumers it means the end of the sale as we know it, for retailers it means slick, smooth operations and increasing profits.


Fashion may be an inherently traditional industry at heart, but paying attention to AI is the only way forwards for brand domination. Never have ‘big data’ and ‘deep learning’ been so in vogue.


More Leaders



WHAT’S UP 2022? Three media experts predict the year’s big trends

Long form, short form, educator content, social media and inclusivity - a trio of industry leaders predict what they expect to be big this year, whilst examining where they come from in the first place. Simon Binns, editor of LAD Bible, Matt...

Posted by: Creativepool Editorial