J.Crew launches Style Advisor, an AI stylist built with Claude, only in the United States
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
Style Advisor, the new assistant on JCrew.com, suggests whole looks based on what the shopper describes, whether a job interview or a wedding by the sea, and lets them adjust the results. It runs on Claude: the stylist is Sonnet 5.5, while Haiku 5.5 speeds up the short tasks. For now it is only in the United States, with men's and women's pieces.
J.Crew now has an AI stylist on its website, and it runs on Claude, Anthropic's models.
The fashion brand rolled out Style Advisor on Wednesday, November 11, an assistant on JCrew.com built for buying the whole outfit rather than one piece at a time. For now it only works in the United States, with women's and men's clothing, and on the computer version of the website; the mobile app will get it later, according to Fortune.
Shoppers say what they are looking for, such as something to wear to a beach wedding or to their first job interview, and the assistant shows them complete looks taken from the brand's own photo shoots, which they can keep adjusting. They can also start from a single item and ask for the rest of the outfit.
According to Anthropic, J.Crew's team developed it with Claude Code, Anthropic's AI coding tool. It used Claude Fable 5.1 to decide how the assistant should reason, put together a prototype with Claude Opus 5.5, and its engineers then took it to production. In each conversation, Sonnet 5.5 plays the stylist and Haiku 5.5 takes care of short tasks so the reply arrives sooner.
Over time, the brand wants to tailor its suggestions to each shopper's tastes and earlier purchases, and bring the assistant to more countries and product lines.
Why it matters · analysis and opinion
What stands out is not that a store has a chat, but that it sells outfits instead of single products, something a sales assistant in a physical store used to do. By limiting suggestions to photos from its own shoots, the brand keeps the assistant from inventing combinations it cannot deliver and preserves its style. The split between a frontier model to build it and faster, more efficient ones to run it is also a clue for any company that wants an assistant that does not keep people waiting. It is still unclear whether customers buy more this way: J.Crew has not shared results, and personalization based on purchase history is still a plan.
Official source: Anthropic · Written with the help of AI: how we make the news
