Some AI Recommend A Piece of Garment. This Picks Out Entire Outfits.

Italy-based AI startup Ustyle said it has developed an engine that can generate entire outfits from a brand’s product catalog, with each garment and accessory shoppable and personalized.

The idea is to recommend an outfit to shoppers, Ustyle said, a top-to-bottom styling advice that many would typically expect when they shop in-person. This is a shopping experience that e-commerce struggles to replicate, leaving consumers to shop online one item at a time.

If a brand wanted to provide complete outfits for every shopper online, that would most likely require “an unworkable volume of photoshoots and an impossible number of combinations,” the startup said in a statement.

Ustyle’s engine provides a workaround to the otherwise gargantuan task of re-shooting as many outfits as there are combinations in a collection. Instead of reshooting outfits, Ustyle said the engine processes each garment once, then layers real garment assets on an AI model to compose limitless, correctly fitted looks from a brand’s catalogue.

“We help brands sell style instead of stuff,” said Amirreza Pourassadi, Ustyle co-founder and CEO. “We are looking forward to working with fashion businesses that want to increase their income, and with investors active in this domain who want to collaborate.”

With over 1.27 million outfit views so far, Ustyle’s technology shows promise. One of Ustyle’s early clients to try the technology was Mialuis, a Turin-based artisanal leather brand, which attributed 33 percent of its total revenue in a 60-day pilot to the engine.

Prior to Ustyle, Mialus would often make suggestions based on new arrivals, as opposed to user behavior or preference, according to a case study published on Ustyle’s website. Ustyle said its engine was then integrated in the company’s e-commerce operations as a lightweight plug-in.

“Instead of relying on generic ‘new arrivals,’ Ustyle surfaces relevant combinations based on context and styling logic, making recommendations feel more engaging and tailored, Ustyle said.

About a quarter of the sales was rated with an attribution score that ranged from 80 to 100, which showed that shoppers used Ustyle outfits before a sale. This, Ustyle said, is a “clear evidence that the AI styling experience is actively driving purchasing decisions.”

Ustyle is supported by Accelera Portugal, Flywheel Fund, CTE Next, and the Innovative Enterprises Incubator of the Politecnico di Torino.

It begs further study to find out how much shoppers are willing to trust fashion recommendations from AI. A recent Vogue business survey with over a hundred respondents found 11 percent only use AI chatbots for personalized style recommendations. A former Sourcing Journal reporter also sought to find out for herself, getting both AI and human styling recommendations before siding with the latter.