XAgent is the open execution market for Agentic Commerce. It turns merchant products, pricing, inventory, order logic, settlement, and fulfillment capability into commerce interfaces that AI agents can read, call, and purchase from. Agents quote, authorize, settle, fulfill, and prove the purchase — machine-to-machine.

What is agentic commerce?

Agentic commerce is a model where AI agents — not humans clicking through web pages — discover services, place orders, and settle payments autonomously. XAgent makes this real through five execution pillars: Quotable, Authorizable, Settleable, Fulfillable, Provable.

How is XAgent different from Stripe or Coinbase Commerce?

Stripe was built for humans clicking checkout buttons. Coinbase Commerce is a crypto payment gateway. XAgent is the protocol-native execution market where AI agents transact: protocols standardize the request, XAgent operationalizes the transaction. The moat is merchant operations — catalog, quote, authorization, order, settlement, delivery, receipt, refund, reconciliation, fraud signal, jurisdiction metadata.

How do AI agents pay merchants on XAgent?

Agents pay using the x402 protocol — an HTTP 402 challenge/response standard. The agent receives a signed payment instruction, authorizes a stablecoin transfer (USDG or USDT) via EIP-3009, and the merchant is paid in seconds with zero gas fee to the buyer (OKX facilitator covers it).

Do I need to write code to list my store?

No. One-click OAuth in your Shopify or WooCommerce admin turns your store into a merchant capability that AI agents can quote, order, pay, settle against, and prove — zero code, zero contracts.

Why agentic payment volume numbers do not agree

Published agentic payment volume figures range from hundreds of millions of transactions to $41,900 a day. They are measuring different things, and some of the traffic is not commerce.

By XAgent Team · 2026-08-11

Published agentic payment volume figures do not agree, and the gap is not a rounding problem. One widely cited number is in the hundreds of millions of transactions. Another, published five months later, is $41,900 a day. Both are real, both are sourced, and stacking them against each other produces nonsense. They cover different chains over different windows, and independent analysts have flagged a large share of the underlying traffic as self-dealing or testing. Anyone deciding whether agentic commerce is commercially real needs to know which number they are holding.

The published agentic payment volume figures

Four figures get quoted most often. Each is stated here with the window and the publication date its source gave it.

Cumulative, through Q1 2026. Chainalysis wrote on June 3, 2026 that "x402 agentic transactions on Base went from near-zero in mid 2025 to well over 100 million cumulative transactions through Q1 2026" (Chainalysis). Cumulative, and explicitly Base only.
Daily snapshot, March 2026. CoinDesk reported on March 11, 2026 that "recent daily snapshots show about 131,000 transactions generating roughly $28,000 in volume" (CoinDesk). No chain scope is given, so this one is not directly comparable to the line above even before the time base differs.
Thirty-day volume, one window, three numbers. Noah Levine of a16z crypto wrote on March 11, 2026: "$24 million reported. $3 million onchain. $1.6 million after removing wash trades." (a16z crypto). These are not three trackers disagreeing. They are one window examined at three depths — what x402.org reported, what Allium Labs' onchain data showed, and what survived a wash-trading filter.
Daily, August 2026. CoinDesk reported on August 3, 2026 that Morgan Stanley, downgrading Circle, was "skeptical about Circle's push into agentic payments, saying transaction volume has fallen to about $41,900 per day, with an implied average transaction size of roughly 24 cents, suggesting limited commercial adoption" (CoinDesk). The analyst was James Faucette; the bank cut its price target to $38 from $106.

Two notes on that fourth one. The words in quotation marks are CoinDesk's, summarising the research note — not Morgan Stanley's own language. And the note as reported does not say which chain, which protocol, or whose measurement the $41,900 covers, or even whether it is Circle-specific rather than market-wide. "Implied" is Morgan Stanley's own qualifier on the 24 cents, and it is worth keeping; how the bank arrived at it is not stated.

Three of these four cannot be compared with each other at all. A cumulative count through Q1 and a daily figure in August are different kinds of object.

Four reasons the figures diverge

Cumulative is not windowed. "Well over 100 million cumulative transactions through Q1 2026" is a running total since inception. "$41,900 per day" is a rate. A total will always look enormous next to a rate, and the comparison carries no information.

Measurement is itself early. Levine's three figures — $24M reported, $3M onchain, $1.6M after wash trades — describe the same thirty days. The gap between them is not a market movement; it is the distance between what a dashboard reports, what the chain shows, and what survives filtering. Levine's own reading of that gap: it "tells you how early-stage even the measurement infrastructure is."

A large share has been flagged as gamed. The filter behind that third number was built by Lucas Shin, an analyst at Artemis Analytics, who per Levine's account "built a wash trading filter for x402 volume, flagging wallets that repeatedly transacted with themselves or cycled funds between addresses." Cryptopolitan, reporting the same analysis on March 12, 2026, gave the split: "as of December, about 48% of transactions and 81% of transaction volume were directly linked to gamed volume", with the caveat that "the numbers could even be higher" (Cryptopolitan). Two limits on that pair of percentages: they are December 2025 figures, and Cryptopolitan is the only outlet carrying them — CoinDesk's account of the same work says only "roughly half". Artemis's own framing was blunter. Posted in February 2026 and quoted by CoinDesk: "The x402 'agent payments' boom is still mostly a mirage."

Spikes have been infrastructure, not demand. CoinDesk noted one day in February 2026 "that logged 3.8 million transactions and roughly $2 million in volume", and — citing onchain analysts at Artemis — that "much of that spike was due to infrastructure testing and experimental use."

There is also a composition point that the headline counts obscure. Chainalysis attributes much of the earlier surge to a single speculative mechanic: "By Q4 2025, transaction volumes surged, driven in large part by meme coin activity, particularly PING", whose pay-to-mint design generated "over 150,000 transactions in its first month." Minting a meme coin is a payment. It is not commerce.

What the figures do agree on

Strip out the incompatible comparisons and a consistent picture survives.

Ticket sizes are very small. Whether you take CoinDesk's March snapshot or Morgan Stanley's August estimate, the per-transaction amounts describe API calls, data access and gated content — not retail baskets. That matches what we said when explaining the protocol itself: most x402 transactions today are API-level micropayments, not full e-commerce orders with inventory, shipping and fulfillment.

Settlement has consolidated. In its Q2 2026 results, released August 5, 2026, Circle said the Agent Stack it launched in May 2026 is "currently home to 900+ paid services", with "99.3% of x402 agent-payment volume settling in USDC" (Circle).

And the supply side is well ahead of the demand side. PYMNTS's read of the same call was that "management did not present agent commerce as an immediate revenue opportunity." Nine hundred services can be paid. The paying is not happening at scale yet.

The metric that would actually mean something

Transaction count is a poor proxy for agentic commerce because it is cheap to manufacture. A wallet cycling funds against itself produces transactions. So does a load test. Both land in the headline number, which is why the trackers disagree and why an analyst can build a filter and remove half of it.

A completed commercial transaction is much harder to fake. It requires a quote the merchant honored, a mandate the purchase was checked against, an order created in a real system, a fulfillment state that moved, and a proof record tying those together. That chain is the agentic commerce sequence, and no part of it is produced by a wallet talking to itself.

So the number worth tracking is not how many x402 payments cleared. It is how many of them ended in an order somebody shipped. On that measure the published figures tell us very little, because almost nobody is reporting it — which is itself the most useful thing the current data says. The protocol layer works. The settlement rails work. The merchant operations layer that turns a cleared payment into a shipped order is the part that has not been built out, and it is the part no volume dashboard measures.

What's next

The signal to watch is composition, not size. If average ticket sizes start moving from cents toward tens of dollars, that is agents buying goods rather than API calls, and it will show up before any headline count does. If the wash-trading share falls without the total collapsing, that is real traffic replacing manufactured traffic.

Until then, treat any single agentic payment volume figure as a claim about a specific chain, over a specific window, measured by a specific tracker with a specific view on what counts. Ask which one before you quote it.

If your business should be able to turn an agent's payment into an order you can actually prove you shipped, list your store on XAgent and let the open execution market handle the quote, the mandate check, the settlement and the record.