Search the Watch Market Opens AI Chat on Live Secondary Listings
On 24 August 2026 WatchPro reported that Search the Watch Market launched AI Market Chat, a tool that answers pricing and valuation questions from live secondary listings rather than from general-purpose chatbot memory. The index covers hundreds of thousands of ads from major platforms, independent dealers, private sellers, and auction houses.
What the chat is allowed to see
Users can ask whether an asking price is fair, what a private sale might realise, and how long a reference has been taking to sell. The model mixes live asks, sold and unsold auction results, price history, and days-to-sell, and is described as showing the underlying listings used for each answer. Broader prompts are in scope: which brands are holding value, or how a watch has moved versus the S&P 500. The product is designed to reply “I don’t know” when the data cannot support a number.
Founder Rusty Green told WatchPro that people were already asking ChatGPT what a watch was worth and receiving figures “with nothing to back them up,” sometimes off by thousands because the model was “remembering the market rather than looking at live data.” Market Chat, he said, frames the firm’s existing live prices as a conversation so users skip building search filters.
- Data: live listings, dealer and private ads, auction sold/unsold
- Outputs: fair-ask checks, private-sale estimates, days-to-sell, comparables shown
- Refusal: “I don’t know” when evidence is thin
- Free tier: first two questions, no account
- Individual plan: $15/week, $52/month, or $450/year (also quoted as £12 / £41 / £356)
- MCP key: Individual subscribers can pipe live data into ChatGPT, Claude, or Gemini
The pricing is a SaaS overlay on a listings database. Two free queries are a trial. Unlimited use sits behind the Individual subscription. Connecting an MCP key to other chatbots is a distribution tactic: the live book travels into tools collectors already use.
Valuation tools and parts markets
Secondary-market chat does not authenticate a watch. Super-fakes and incomplete provenance, in the news the same week, still require human inspection. An AI that cites listings can still be wrong if those listings are wrong. The improvement claimed versus generic LLMs is recency and citation, not infallibility.
Aftermarket builders rarely need an S&P 500 overlay. They need to know whether a case, crystal, and movement actually fit and whether a vendor’s photos match the SKU. A parts quality guide is that due diligence at component level. Once a case is closed, a price chat cannot replace a pressure test; a water-resistance testing note remains the physical check.
Dealers gain a faster way to quote without opening five marketplaces. Collectors gain a conversational wrapper on data they could already filter by hand. The industry question is whether “I don’t know” is used often enough to stay credible when the book is thin.
FAQ
Does the chat use the same training cut-off as ChatGPT? WatchPro says it queries live listings and auction results on Search the Watch Market’s own index, and can refuse when data is insufficient.
What does it cost after two questions? Individual subscription at $15 per week, $52 per month, or $450 per year, with an MCP key for other assistants included on that plan.
Can it tell a fake from a genuine watch? No such claim was made. The tool prices from public listings. Authentication remains a separate inspection problem.