TLDR;
The RealReal has opened its Ask TRR shopping agent to all 45 million members after a pilot. Built with Google Cloud's Gemini Enterprise, it lets shoppers describe a style, occasion or product need in normal language and receive recommendations from a fast-changing luxury resale catalogue. The scale is significant, but the company has not published usage, conversion or revenue results, so brands should treat this as a production rollout rather than proof of commercial success.
What happened
The RealReal and Google Cloud announced the wider rollout on 8 October. Ask TRR is now available to the retailer's full member base after what the companies call a successful pilot. The agent is designed for product discovery, using conversational questions and stated preferences to recommend items from more than one million unique luxury products. The RealReal also says it plans to add AI recommendations to product description pages, although it gave no launch date.
The catalogue makes this a useful test case. Most retailers can sell several units of the same product and rely on a stable identifier, structured attributes and accumulated engagement signals. The RealReal's items are largely one of one. Once an item sells it disappears, while new inventory arrives continuously. A shopper may know the era, shape or occasion they want without knowing a designer name or the exact vocabulary used in the listing.
Ask TRR tries to bridge that gap by translating an open-ended request into catalogue matches. The retailer says its advantage comes from 15 years of proprietary data and more than 50 million items sold. Those are company claims, and the launch announcement provides no detail on the retrieval architecture, recommendation quality thresholds or how personalisation is measured. It also gives no pilot figures for adoption, conversion, order value or customer satisfaction.
Google positions the underlying Gemini Enterprise for Customer Experience suite as a way to connect discovery, recommendations and commerce. Its retail material says the shopping agent can use real-time availability, build a cart and streamline purchase. That describes the product's intended scope, not a verified result for Ask TRR. The evidence today is that a large retailer has moved a branded conversational shopping experience from pilot to broad availability.
Why it matters
This rollout shows where first-party AI discovery can be strongest. The agent sits on the retailer's own property and can use catalogue data, business rules and customer context that a general assistant may not have. It can also handle needs that keyword search expresses poorly, such as finding an understated bag for work or a particular decade's silhouette within a strict budget.
The hard part is not producing fluent answers. It is keeping recommendations grounded in inventory that can disappear between retrieval and checkout. Price, condition, size, authenticity information and availability all need to remain consistent across the catalogue index, product page and order system. A persuasive recommendation for a sold or misdescribed one-off item is worse than an empty result because it spends customer trust.
The member number also needs careful reading. Forty-five million people having access does not mean 45 million people use the agent. Enterprise teams should separate rollout reach from active users, completed conversations and incremental sales. Without a holdout group or a clean comparison against search and filters, it is easy to count assisted orders that would have happened anyway.
How your brand can benefit / be affected
Brands with complicated catalogues should start with a narrow discovery problem where conversational input adds real value. Examples include compatibility questions, bundles, occasions, room dimensions or products with many overlapping attributes. Give the agent live inventory and policy data, then make it show the relevant facts behind each recommendation instead of relying on generic copy.
Instrument the funnel at the conversation level. Track the request, eligible result set, products shown, product views, adds to basket, purchases and returns. Compare that journey with existing search and recommendation experiences. Google Cloud documentation for AI Commerce Search includes catalogue and user-event setup because relevance depends on both product data and behavioural signals. Those inputs need the same ownership and quality controls as any other production merchandising system.
Keep human and operational safeguards close. High-value resale items may need extra authentication detail, condition checks or customer support before purchase. Give shoppers an easy route to inspect the listing and ask for help. The RealReal's rollout is a credible sign that conversational product discovery is becoming a normal retail interface. The business case still has to be proven with incremental margin, lower search failure and better customer outcomes.
News date: 8 October 2026. Editorial review: 10 October 2026.