AI has been talked about in fashion for years - usually in the context of trend forecasting or supply chain optimisation, which are useful but largely invisible to shoppers. What is actually changing at the consumer level is more interesting, and more recent. Here is what has shifted and where the biggest gaps still remain.
Personalised Discovery
The most visible AI in fashion retail right now is personalised feed curation. Every major fashion platform - ASOS, Zalando, and various social commerce layers on Instagram and TikTok - uses machine learning to surface products based on what you have browsed, saved, or purchased before.
The better implementations go beyond simple "you bought this so you might like this" logic. They build a model of your aesthetic preferences over time, understand your price sensitivity, and can surface new brands or product types that fit your style even if you have never bought from them before. The result is a shopping feed that feels curated rather than generic.
The limitation is that this personalisation is still largely reactive. It reflects your past behaviour more than your evolving taste. If you want to shift your aesthetic in a new direction, you typically have to actively signal that change rather than expecting the system to anticipate it. Personalisation based on purchase history will always lag slightly behind what you actually want next.
Visual Search
Visual search - the ability to photograph something and find similar items for sale - has become a standard feature on most major shopping apps. Pinterest Lens, Google Lens, and the built-in visual search tools on platforms like ASOS and Farfetch let you find comparable items from a photograph.
This is particularly useful for specific things you have seen on social media or in real life and want to find at a different price point, in a different colourway, or from a specific brand. It cuts out the frustrating process of trying to describe something in keywords. "Oversized washed denim jacket with distressed collar" is harder to search than just pointing a camera at it.
The technology is not perfect - it tends to match on broad silhouette and colour before fine details - but it has reached a level of reliability where it genuinely saves time in the right situations.
AI Size and Fit Guidance
One of the most persistent problems in online fashion is that size labels vary significantly between brands, and often between styles within the same brand. A medium from one label may fit very differently to a medium from another, and product descriptions rarely give you the specific measurements you actually need to make a confident decision.
AI has been applied to this in a few ways. Some tools use manually entered body measurements to recommend a specific size for a specific garment. Others use purchase and return history - if you consistently buy a medium and return it for a large from a particular type of brand, the system can infer your preference. Services like True Fit and Fit Analytics have been providing this kind of size recommendation to retailers for several years, often embedded invisibly in the product page.
This is one of the more impactful areas because accurate size guidance directly reduces returns - which are costly for retailers, inconvenient for shoppers, and have a measurable environmental footprint. It is also one of the more difficult problems because fit is subjective: some people prefer a looser fit in a garment that others would want fitted.
Virtual Try-On and Reducing Returns
The UK's online fashion return rate is consistently among the highest in Europe. Estimates typically put it between 30 and 50 per cent for clothing, with fit and appearance being the most commonly cited reasons. When something does not look right or does not fit the way you expected, it goes back.
Virtual try-on addresses this directly. If you can see how a garment looks on a body that resembles yours before you buy it, you have more information with which to make a better decision. You are less likely to impulse-buy something that looks good in an editorial shot but will not work for you specifically. And you are better placed to evaluate whether two items will actually look good together as an outfit, not just as individual pieces.
Apps that let you try on clothes from multiple brands before purchasing - like My Styles, which renders full outfits on a mannequin matched to your body type - are part of a broader shift toward making online shopping more visually confident. The technology handles some garment types better than others, and it cannot replicate the tactile experience of trying something on in person. But the directional logic is sound, and the quality of rendering has improved considerably.
What Is Still Missing
AI has not yet solved the tactile problem. Fashion shopping is partly a sensory experience - the weight of a fabric, the way a collar sits, the quality of stitching, the feel of a zip. No visual tool can replicate that. What the best tools can do is give you better visual information before you commit, and reduce the number of "not quite right" purchases you make.
The other significant gap is cross-platform integration. Most AI tools are still siloed within a single retailer or app ecosystem. Your browsing history on ASOS does not inform your experience on Zalando; your saved items on one platform do not travel with you to another. The more open vision - where your wardrobe, your purchase history, and your style preferences are portable across every retailer you use - remains largely unrealised.
Those gaps aside, the experience of shopping for clothes online in 2026 is meaningfully better than it was five years ago. The best tools combine personalised discovery, visual try-on, and size guidance in a way that reduces uncertainty at every stage of the process. The direction of travel is clear, even if the destination is still some way off.