Agentic Commerce Overview

A catalog of services that AI agents can pay for autonomously — priced per event/endpoint — across the x402 ecosystem (Coinbase CDP Bazaar + x402scan) and the Apify Store. Pricing, real on-chain settlement volume, liveness, settlement rail, categories, and demand signals, refreshed daily.

Loading…

Priced events by price band

Stacked by source. The agentic economy today is overwhelmingly sub-dollar.

CDP Bazaar (x402) Apify Store x402scan (Merit)

Top service categories

Rough keyword-based categorization across both sources.

Top endpoints by estimated spend

Each bar is a single endpoint's best dollar figure — price × transactions (observed on-chain where available). Color = source. A per-endpoint view; see the services table below for provider/actor/domain totals.

x402 — top endpoints by 30-day unique payers

On-chain settlement makes demand observable. Unique payers are used (not raw call counts) to blunt memecoin-farming noise.

Top services by revenue

One row per service (deduped across sources). Per horizon (30-day / all-time) we show Est $ (price × real transactions, our estimate) beside Obs $ (real settled USDC, ⛓ — ties out to x402scan) so you can see where the estimate diverges from what was actually paid. The Σ columns are the median of the available signals — the overall-ranking column. Click a row to filter the endpoint table to that service. Sortable by any column.

What is what — reading the numbers (sources, timeframes, tie-outs)

The dataset primitive is a priced endpoint (a row); a service (provider / actor / domain) is the top level. For each service we show, per horizon, an Estimate and an Observation side by side:

  • Obs $ (observed) — real USDC settled on-chain, pulled from x402scan's own stats.overall for the service's recipient wallets. It ties out to x402scan (e.g. a service showing $152k / 13.15M tx here reads the same on x402scan). mppscan settlement works the same way. Off-chain sources (Apify) have no observed $.
  • Est $ (estimate) — catalog price × real transactions. Comparing it to Obs $ isolates price accuracy: if Est ≪ Obs, buyers are paying more than the listed price (higher-tier endpoints, price changes); their ratio is roughly listed-price ÷ actually-paid.
  • Σ (median) — the median of the available $ signals (observed + each source's estimate). A robust central number, and the single overall-ranking column.
  • Transaction basis: x402/mppscan use the real on-chain tx count; CDP uses Coinbase's 30-day call count; Apify uses runs. (x402scan's catalog toolCalls counter is NOT settlement — blockrun reads 0 there despite 13M real transactions — so it's not used as a spend basis.)

Unique buyers counts the distinct wallets/accounts that paid (30-day) — not transactions: one buyer can transact many times, so this is much smaller than the tx count (e.g. a service with 13M transactions but ~1,000 buyers is automated/whale traffic, not a broad user base). It's the wash-resistant demand signal — raw transaction counts are trivially inflated, unique payers are not. Sources differ: on-chain unique payers/buyers (CDP, x402scan, mppscan) vs. Apify platform users. ⛓ marks real on-chain settlement.
Loading dataset…