Inside Rapaport’s Agentic AI Tool for Diamond Inventory, Pricing, and Demand Signals

The Core Problem: Market Intelligence at Human Speed

Diamond traders have always operated in an information-dense environment. Pricing benchmarks, inventory turnover rates, buyer search patterns, and trading volumes all feed into daily decisions — what to buy, what to hold, what to discount, and when to move.

The bottleneck is not data availability. Rapaport has long provided pricing lists, market reports, and marketplace access. The bottleneck is synthesis speed. Translating raw data into a specific, actionable recommendation for a specific trader, in a specific market moment, requires analyst-level judgment applied continuously.

That is precisely the gap the agentic AI tool targets.

What the AI Agent Actually Does

Mano described the tool in direct terms: “It’s like you have an analyst inside Rapaport that has access to all the data and it’s working for you, 1,000 times faster than a strong analyst.”

The agent connects to four primary data streams:

• Inventory data — what stones are listed, by whom, and for how long
• Pricing data — current market rates and historical benchmarks
• Trading patterns — transaction behavior across the marketplace
• Search data — what buyers are actively looking for

From these inputs, the agent answers operational questions that traders ask daily: Which diamonds should I buy right now? How should I price this parcel? Which goods in my inventory have been sitting too long and need repricing?

From Queries to Complex Scenarios

The tool is designed to handle more than simple lookups. Mano gave a concrete example: a jeweler helping a customer upgrade a diamond based on the stone they already own and their available budget. That scenario requires cross-referencing current inventory, trade-in value logic, buyer preference signals, and pricing margins — simultaneously.

This is the defining characteristic of an agentic system. It does not return a list of results. It reasons through a scenario and returns a recommendation.

Where This Fits in the Broader Platform Revamp

The AI agent does not exist in isolation. Rapaport is restructuring its entire product surface around more intelligent, context-aware tooling.

Search Algorithm Overhaul

RapNet, now rebranded as Rapaport Trade, is changing how search results are ranked. The lowest-priced stone will no longer automatically appear first. Instead, a “best match” algorithm surfaces the most relevant diamonds based on a range of factors — mirroring the logic used by Amazon or Booking.com.

Chief Revenue Officer Jeremy Werblowsky made the reasoning explicit: buyers do not actually want the cheapest option first. They want the most relevant option. Ranking purely by price distorts the market and disadvantages sellers who offer quality at fair — not rock-bottom — prices.

The platform will also group near-identical stones from the same seller into single listings, reducing visual clutter and improving the overall search experience for buyers.

SellerIQ: Demand Visibility for Sellers

One of the more practically useful new features is SellerIQ, a tool that shows sellers which of their listed diamonds are attracting buyer attention. Werblowsky compared it to LinkedIn’s “who viewed your profile” feature — a simple but powerful signal that tells sellers where latent demand exists before a transaction occurs.

For inventory management, this is significant. A seller can identify which stones are generating interest but not converting, then adjust pricing or presentation accordingly — rather than waiting for a sale to fail before acting.

Rapaport Polaris and the Intelligence Layer

Rapaport Polaris is a new dashboard consolidating pricing data, inventory analytics, and broader market intelligence into a single view. Combined with the rebranded Rapaport Intelligence Report, it positions the platform as an active market intelligence layer rather than a passive data repository.

The integrated chat function — designed to consolidate business conversations and eventually connect to payments — adds a transactional dimension. The goal is to reduce fragmented communication across WhatsApp and email, keeping deal flow traceable and secure within the platform.

The Agentic AI Pattern Applied to a Vertical Market

What Rapaport is building follows a pattern increasingly visible across B2B verticals: take a proprietary dataset, layer a reasoning agent on top, and deliver domain-specific intelligence that generic AI tools cannot replicate.

The competitive moat is not the AI model itself. It is the data. Rapaport’s decades of pricing benchmarks, trading records, and marketplace behavior give the agent context that no external tool can access. A general-purpose AI assistant cannot tell a diamond dealer which SI2 rounds in the 1.50–1.80 carat range are underpriced relative to current search demand. Rapaport’s agent, in principle, can.

This is the practical value proposition of vertical AI tools: specificity derived from proprietary data, not from model architecture.

Actionable Takeaways for Jewelry Trade Professionals

For traders, dealers, and jewelers evaluating whether to engage with Rapaport’s evolving platform, several practical considerations emerge.

Monitor the beta program. Mano explicitly called for beta testers. Early access to the AI agent means early influence over its development and early competitive advantage in applying its outputs.

Treat SellerIQ as an inventory management signal. Attention data — knowing which stones buyers are viewing — is a leading indicator. Use it to make pricing decisions before inventory ages, not after.

Reassess search strategy on Rapaport Trade. With the algorithm shifting from price-first to relevance-first, listings optimized purely for lowest price will lose visibility. Quality presentation, accurate grading data, and competitive-but-not-desperate pricing become more important.

Integrate Polaris into regular market review. A consolidated dashboard reduces the time spent aggregating data manually. If the tool delivers on its promise, it should replace several separate research steps with a single weekly review.

A Market Positioning Note

Martin Rapaport’s address at the same breakfast offered a strategic frame worth noting. His argument — that natural diamonds are Veblen goods, and that the industry should orient toward affluent buyers rather than compete on accessibility — shapes the context in which this AI tooling is being built.

The platform is not being designed to help traders find the cheapest stone for a budget-conscious buyer. It is being designed to help sophisticated sellers serve sophisticated buyers more efficiently. The AI agent, the best-match algorithm, and the SellerIQ demand signals all point in the same direction: toward quality, relevance, and margin — not volume and discount.

The Broader Signal

Rapaport’s AI pilot is a precise case study in how domain-specific agentic tools differ from general AI assistants. The value is not conversational fluency. It is the combination of structured proprietary data, domain-calibrated reasoning, and workflow integration — delivered at a speed no human analyst can match consistently.

For AI observers, this is the pattern to watch across every data-rich vertical: the organizations that own the data and build the agent on top of it will define the intelligence layer of their industry. Rapaport is making that move deliberately, and the diamond trade will be a useful benchmark for how quickly agentic tools can shift from internal pilot to operational standard.