The RealReal Opens Its AI Shopping Agent to All 45 Million Members
Ask TRR turns everyday requests into luxury resale recommendations. Product-page guidance is planned next, but the rollout announcement gives no measured pilot results.
Loading page…
Ask TRR turns everyday requests into luxury resale recommendations. Product-page guidance is planned next, but the rollout announcement gives no measured pilot results.
Listen to this story
The RealReal’s catalog contains more than 1.5 million one-off pieces and changes daily, making conventional keyword search a poor fit for shoppers who know what they want but not how to name it. On October 8, 2026, the retailer expanded Ask TRR, built with Google Cloud, from a pilot to all 45 million members. It has not published measured results from the pilot, and planned product-page recommendations have no launch date.
Ask TRR uses Gemini Enterprise for Customer Experience to turn everyday-language requests into product recommendations.
CEO Rati Sahi Levesque says the recommendations draw on 15 years of The RealReal’s proprietary data and luxury expertise.
The company described the pilot as successful but provided no figures on its effect on purchases or product discovery.
Shoppers on The RealReal can now describe what they want rather than start with an exact designer name or search term. On October 8, 2026, the luxury resale retailer announced that Ask TRR, its Google Cloud-powered AI shopping agent, was moving from a pilot to all 45 million members.
The retailer’s inventory makes product discovery unusually time-sensitive. Each item is a single piece: when it sells, that listing’s inventory is gone, while new items arrive daily. CEO Rati Sahi Levesque says the site carries more than 1.5 million unique pieces every day.
The company frames that as a mismatch with conventional keyword search, which it describes as built for retailers stocking many units of the same product. A resale shopper might know the silhouette, era or occasion they have in mind without knowing the designer or wording needed to locate it.
In the retailer’s account, the consequence is not simply a frustrating search. A suitable piece can remain buried in the catalog or sell before a shopper finds it. Ask TRR is intended to bridge that gap between an idea for a purchase and the available assortment.
The RealReal calls the pilot successful, but the announcement gives no measured results showing how it affected purchases or product discovery. The concrete change is access: the agent is now launching across the retailer’s membership rather than remaining a pilot.
Built with Gemini Enterprise for Customer Experience, Ask TRR accepts requests in everyday language and returns product recommendations. The announcement offers “What luxury bag should I buy?” as an example. The company says those suggestions are tailored to the shopper’s intent and preferences.
Levesque points to 15 years of proprietary data and luxury expertise as the foundation for that personalization. Her stated aim is specificity whether a customer begins with a designer, an occasion or only an idea—not a requirement to arrive with the right catalog keyword.
In a luxury boutique, the best sales associates remember what you love and know what just came in. The RealReal is bringing that level of service online, to every shopper who visits.
Darshan Kantak, vice president, product, Applied AI, Google Cloud
The RealReal also plans AI-driven recommendations on product description pages. That would put contextual guidance alongside an item a shopper is already considering, extending the experience beyond asking the agent to find candidates in the wider catalog.
That addition remains a future development, with no launch date specified. For now, the announced rollout concerns conversational recommendations across the changing luxury inventory; guidance embedded in individual product pages is the next proposed step.
Loading discussion...
Join the conversation
Explain which approach would help you find something unexpected.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.