How are fashion brands actually using AI

But where agreement does seem to have been reached is that we’ve moved past just being wowed by the idea of this shiny new technology. Now businesses and consumers are both looking for concrete examples of how AI can be used well and the benefits it brings.

 

So how are four major fashion brands actually using the tech?

 

Employee AI assistant (Levi’s)

Levi’s is using AI to make sure in-store employees are as informed – or more so – than customers when they come in the doors.

 

The idea for its employee AI assistant, STITCH, actually came from an employee hackathon last year. Levi’s then ran a pilot in 10 stores before rolling STITCH out to over 70 Levi’s stores in the US due to the success of the tool – improving customer satisfaction by as much as eight points in stores with the tech compared to those without.

 

Store staff can ask the AI assistant operational and product questions in the same natural language that they would use to speak to another person. This might be the differences between models of jeans and how to do things like fulfil online orders in the store.

 

Levi’s has also introduced AI into other aspects of its business, such as demand forecasting, and encouraged its employees to use Microsoft Copilot with them creating over 800 AI agents so far.

 

Each one of these agents gives Levi’s insight into the challenges that staff have and where AI can actually make a difference, which helps refine where investment goes.

Retail reporting (Urban Outfitters)

The parent company of Urban Outfitters, URBN, is experimenting with agentic AI to automate its weekly reporting. The system aims to reduce multiple separate reports into a single overview that brings together all that data and identifies actionable elements to speed up decision making.

 

This goes beyond generative AI reporting, where a report is created in response to a prompt and provision of data. The agentic system runs in a proactive way, collecting the data and generating the reports autonomously without needing to be prompted at every step.

 

To maximise accuracy, URBN has developed a universal semantic layer, which is a unified data model, and means that the agentic AI has a single source of truth to work from. This also gives the company a way to trace every output back to where it originally came from.

 

URBN is hoping that insights into things like one product type selling above expectations in one area and under in another will allow it to adjust allocation and respond to trends faster.

AI avatars (Zara)

Zara has been playing around with generative AI for a while, including using it to dress existing photos of human models in new outfits.

 

The aim is to allow Zara to update its online store with new styles faster, compared to booking another photo shoot each time. Zara also reportedly gets approval from models before using AI to edit their photos and provides financial compensation.

 

But Zara is now going a step further with the introduction of generative AI try-on tech in its app.

 

Shoppers can upload a headshot and a full body photo to create a 360-degree virtual avatar of themselves. They can then ‘try on’ any products with AI dressing their avatar in the chosen items.

 

A nice touch is that the tech will suggest items to go with the chosen product. So if you choose a skirt, the AI will pair it with tops, jackets and even shoes to give you an idea of how it would look as an outfit. The AI looks are also saved in the app, so you can go back to them later.

 

It’s an example of how fashion retailers are using AI to tackle one of their biggest headaches – product returns because customers don’t know how clothes will look on or whether they will suit them when buying online.

 

The AI avatar may not be 100% accurate but customers may find it beneficial to see items on a version of themselves rather than a model with completely different proportions.

 

Photo courtesy of Revolve

AI applications for physical stores (Revolve)

Revolve is one of the most all-in on AI fashion brands and very open about it.

 

A data first business from the start, Revolve is incorporating AI into almost every part of its business from design to user experience to internal reporting. For example, on the e-commerce side, Revolve has developed a new internal programme that uses AI to improve product metadata. 

 

Usually, retailers have to manually tag the attributes of a product so that they can appear in results when customers search for specific product types or for keywords. Revolve is using AI to ‘look’ at images and tag them so that they appear in more relevant searches – eg an item doesn’t have to be specifically listed under ‘work outfit’ to appear in a search for those keywords if it has relevant attributes.

 

What’s most interesting though is that Revolve has started to move into physical retail, opening two stores in the US since 2024. It also plans to open further stores to drive growth.

 

And the company wants to apply its same data and AI powered approach to its real world stores.

 

Revolve is actively developing tools to give it access to the sort of reporting that it is used to from its website where every change is measurable in terms of its impact on sales and user experience. It sees AI as an opportunity to understand more about what’s happening in the store and how to optimise it, such as where to put a display.

 

There aren’t many physical store operators who wouldn’t like more insight into how their stores convert so if Revolve can prove AI’s capability in this area, many are likely to follow their example. 

 

What do these use cases mean for sourcing?

Practical examples of AI usage like these give us insights into how the fashion industry is evolving. And this impacts sourcing as much as every other part of the business.

 

If generative AI avatars help to reduce online returns then brands and retailers may start to adjust how much they order of certain products. Likewise, insights into demand for individual items across specific physical stores may help with trend predictions and associated sourcing.

 

AI-supported reporting also could help brands with predictions and more accurate ordering, which in turn may reduce waste from unsold clothing. And employee AI assistants could have their own part to play in helping guide shopper decisions to cut the number of returns.

 

While the conversation around ethical and genuinely useful AI usage needs to continue, these real life examples show that the tech has potential as a tool for reducing returns and overordering, which can help retailers make better sourcing decisions and ultimately produce less waste. And that’s a discussion worth having.