Fashion platforms including Musinsa and W Concept are rolling out artificial intelligence-powered image search services, accelerating efforts to upgrade their search technology. The idea is to help shoppers find identical or similar clothing from a single photo, even when they do not know the product name or brand, and to carry that discovery through to a purchase. Competition is intensifying beyond search convenience alone, moving toward using AI to read customer tastes and recommend tailored products.
Musinsa will introduce an AI-based image search service called "Musinsa Lens" starting on the 28th, according to industry sources on the 23rd. The company recently formalized the launch by adding clauses to the integrated terms of service covering Musinsa Store, 29CM and SoldOut that define the image search service, set usage standards and address the sharing of search results. Musinsa Lens lets users attach a photo they already have or upload an image taken on the spot, then uses AI analysis to find identical or similar products. It also offers a function to share search results via a link.
W Concept also added image search last month when it began beta testing "W Assistant," a conversational AI shopping agent. The service, currently available to customers at its Best tier or higher, lets shoppers attach a photo of the style they want and then searches roughly 1.67 million product images to recommend items with similar colors, shapes and styles.
The service applies generative AI-based multimodal embedding and vector search technology, allowing users to explore preferences that are difficult to describe in words by using images instead. Multimodal embedding is an AI technique that gathers different types of data, such as text and images, into a single high-dimensional space so that context and meaning across the data can be understood at once. W Concept plans to verify service quality first with heavy users before expanding to all customers as a full service.
Fashion platforms are turning to image search because keyword search alone has limits in capturing specific customer preferences. With visual elements such as color, silhouette, fabric and pattern weighing heavily on purchase decisions for fashion items, demand is growing for ways to find a desired product from a photo alone, without knowing the product name or brand, according to industry analysis. Image search shortens the browsing process and links a style a customer has spotted directly to products on the platform.
Some companies in the industry have already launched such services and are refining them. SSG.com introduced an image search service called "SSG Lens" in 2018 and applied an in-house multimodal AI model in 2024. The key change was to draw on text information such as product names, brands and detailed features alongside product images. Where the earlier version focused on finding products that looked similar to an image, the upgraded service lets users add search terms such as color or design to the image search results to narrow down products that match their conditions more precisely.
Zigzag, operated by Kakao Style, began a pilot of "ZigZag Lens" in May 2023 and launched it formally in September that year. When a user uploads an outfit photo from social media or takes a picture of a specific item, the service recommends products with similar shapes, colors and patterns, ranging from clothing to accessories such as bags and shoes. Built on proprietary machine learning technology, the service becomes more refined as search data accumulates. According to Zigzag, the number of ZigZag Lens users rose about 40% last year from a year earlier, while product clicks on image search results pages increased about 70% and the number of items saved to wish lists roughly doubled.
"The larger a fashion platform's range of products and brands, the more search technology that helps customers quickly find what they want determines purchase conversion rates and ease of use," an industry official said. "The ability to understand customer intent through image and AI search and to connect that to actual product recommendations and purchases is becoming an essential competitive edge."