How AI Shopping Agents Are Changing Luxury Resale Discovery

AI shopping agents are beginning to help resale shoppers describe what they want in natural language and find relevant pieces in changing, one-of-a-kind inventories. The RealReal’s Ask TRR, expanded to all 45 million members in October 2026, is a current example of discovery and recommendations—not autonomous purchasing. ^[1]

• Ask TRR uses Google Cloud’s Gemini Enterprise to recommend luxury resale items based on shoppers’ questions and preferences. ^[1]
• Because a one-of-one item is gone once sold, resale recommendations need to account for what is available at that moment. ^[1]
• The RealReal’s reported search trends illustrate specific shopper interests, but do not show that Ask TRR caused those trends. ^[2]
• An availability-first recommendation framework can connect a shopper’s intent to a current listing, explain the match, and respond clearly if the item sells. This is editorial analysis, not a confirmed Ask TRR feature.

What does Ask TRR do for luxury resale shoppers?

Ask TRR is a natural-language shopping agent designed to help shoppers explore The RealReal’s changing luxury assortment. ^[1]

On October 8, 2026, The RealReal announced that Ask TRR, built with Google Cloud’s Gemini Enterprise, was expanding after a pilot to all 45 million members. ^[1] Shoppers can ask questions such as “What luxury bag should I buy?” and receive recommendations based on their intent and preferences. ^[1]

The distinction between discovery and purchasing matters: the announcement describes recommendations, not an agent completing a purchase or checkout on a shopper’s behalf. ^[1] The RealReal also said it plans to add AI-driven recommendations to product-description pages, which is a future plan rather than a capability the announcement says is already in place. ^[1]

The company says its site carries more than 1.5 million unique pieces every day. ^[1] That scale helps explain why conversational discovery could be useful: a shopper can express a broad request in their own words rather than begin with a specific product name. Whether recommendations make the search easier in practice is a separate question; the announcement does not report conversion, revenue, or satisfaction results. ^[1]

Why is resale discovery different when every item is one of a kind?

One-of-one resale inventory means that each individual item is unique and, once sold, that exact piece is no longer available. ^[1]

The RealReal says new inventory arrives every day and that once an item sells, “it’s gone for good.” ^[1] This makes resale discovery different from recommending replenishable stock: an agent needs to connect a shopper’s intent to an item that is currently available, not simply suggest a product type that might be restocked.

Consider an illustrative request for a vintage designer bag for a particular occasion. A useful agent would surface relevant available pieces and explain how each matches the shopper’s stated purpose or preferences. This example describes a possible approach, not a confirmed Ask TRR feature.

An availability-first framework for resale recommendations can be expressed in four steps:

1. Capture intent. Understand the shopper’s stated style, purpose, and preferences.
2. Match to current inventory. Recommend items that are available when the shopper is searching.
3. Explain the fit. Connect each recommendation to the shopper’s request rather than offering an unexplained result.
4. Handle changes transparently. If an item sells, say so and offer relevant alternatives where possible.

This is an editorial framework for evaluating resale-agent experiences, not a description of Ask TRR’s confirmed functions. The practical challenge is to keep a recommendation useful even when a particular listing disappears.

What do resale search trends reveal about the intent agents may need to understand?

The RealReal’s 2026 Resale Report offers company-reported examples of the specific interests that resale discovery tools may need to interpret. ^[2]

The company says its report draws on 15 years of proprietary data across more than 50 million items sold and a community of 44 million members. ^[2] The RealReal reports that demand for vintage on its marketplace was up 432% since 2020, decade-specific saved searches rose 135% year over year, and searches for 2000s fashion increased 93%. ^[2] These are company marketplace figures, not independent estimates of the resale market or evidence that AI caused the changes.

Designer-era searches offer another illustration of nuanced intent: The RealReal reported increases in saved searches for three specific designer-and-era combinations. ^[2]

• Phoebe Philo Celine: saved searches increased 650%. ^[2]
• Nicolas Ghesquière Balenciaga: saved searches increased 400%. ^[2]
• Marc Jacobs Louis Vuitton: saved searches increased 375%. ^[2]

These reported increases describe shopper activity on The RealReal; they do not show that Ask TRR generated the searches or caused a change in demand. ^[2] The figures do, however, illustrate why shoppers may want to express a preference more precisely than “find me a bag.”

What should shoppers and resale retailers expect from this shift?

For shoppers, a conversational agent can provide another way to express preferences and begin discovery, but recommendations should be treated as suggestions to investigate rather than a substitute for checking the current listing.

Before deciding, shoppers can review the item details on the listing and confirm that the piece remains available. This is particularly relevant when the recommendation concerns a unique item that may sell before the shopper returns to it. ^[1]

For retailers building or assessing resale discovery tools, an availability-first checklist offers a practical starting point:

• Keep item availability current in the experience that produces recommendations.
• Make each unique item’s attributes usable for matching against shopper requests.
• Show how a recommendation relates to the shopper’s stated style, purpose, or preferences.
• Communicate clearly if a recommended item is no longer available.

McKinsey’s 2026 research provides wider luxury-shopping context, but its figures should be read as directional. McKinsey reports that 85% of surveyed luxury consumers use a multipurpose AI assistant to support shopping decisions, 74% have used visual search, and 68% of surveyed luxury merchants cite more personalized experiences as a primary source of excitement about agentic commerce. ^[3] McKinsey says its research included a consumer survey of 300 people and a merchant survey of 90 people, with luxury-segment percentages referring to 31 respondents; it cautions that the findings are directional indicators captured as a snapshot in time. ^[3]

The near-term test for resale agents is therefore not whether they can make a recommendation once, but whether they can keep discovery relevant as individual listings disappear. The RealReal’s announcement describes a current discovery and recommendation experience and a separate plan for product-page recommendations; it does not provide measured business outcomes from the rollout. ^[1]

Frequently asked questions

How can shoppers check the condition and authenticity of a luxury resale item?

Review the specific listing’s condition information and item details before deciding, and ask the retailer for clarification if something is unclear. Do not assume that an AI recommendation itself verifies a listing’s condition or authenticity.

Can an AI shopping agent compare resale listings across multiple marketplaces?

That depends on the agent and the marketplaces it can access. The Ask TRR announcement describes discovery across The RealReal’s assortment; it does not say that Ask TRR compares listings across multiple marketplaces. ^[1]

How should shoppers set a budget before using an AI shopping agent?

Set a maximum spend before searching, then include the budget in the request if the agent accepts that kind of preference. Check the listing and any applicable costs yourself before deciding; a recommendation is not a commitment to buy.