Day 16 April 26, 2026

Agentic Commerce

When the agent is the buyer. Why $0.20 transactions break Visa-economics, how Stripe’s ACP, Google’s AP2, Visa’s TAP, and Coinbase’s x402 each frame the same problem differently, what an Intent Mandate actually is, and what it means that ChatGPT killed Instant Checkout seven weeks ago.

ACP
AP2
Trusted Agent Protocol
x402
Mandates
Stablecoin Rails
Agent Wallets
Curriculum
94.1% complete (Day 16 / 17)
Core Concept

What “Agentic Commerce” Actually Means — and Why Card Networks Are Suddenly Interested

Picture a personal travel agent in 1985 — not a website, the human kind. You tell them “book me Tokyo, week of June 8, under $1,500, prefer JAL.” They go away, query airlines, return with three options, you pick one, and they put it on your AmEx using authority you delegated when you became their client. The agent never has your card. The card network never sees the agent. The merchant ships you a ticket. The audit trail lives in the agent’s file cabinet. Agentic commerce is that, but the agent is software, the discovery happens in seconds, and the “file cabinet’ has to be a cryptographically verifiable signature that survives a courtroom.

Technically, agentic commerce is the set of protocols and trust mechanisms that let an autonomous AI agent (a) discover products on behalf of a principal (the human or business it serves), (b) negotiate or assemble a cart, (c) authorize and execute payment, and (d) receive a receipt — without a human clicking “Buy Now” on a checkout page. The hard problem is not the discovery (that’s just retrieval and ranking from Day 04), nor the assembly (a planning loop from Day 03), nor even the payment rails. The hard problem is authorization: how does a merchant know this agent is allowed to spend this money on this product for this principal, and how does the principal’s bank know to honor the charge if it’s disputed?

Four distinct architectures emerged in the second half of 2025 to answer that question, each from a different power center. Stripe and OpenAI launched the Agentic Commerce Protocol (ACP) in September 2025 — an HTTP/JSON-RPC layer where the merchant stays merchant of record and the agent ferries Shared Payment Tokens. Google (with 60+ partners including Mastercard, PayPal, Amex, Coinbase) launched the Agent Payments Protocol (AP2) in the same week — a payment-rail-agnostic framework built around cryptographically signed Mandates. Visa and Cloudflare introduced the Trusted Agent Protocol (TAP) in October 2025 — a thin shim that lets merchants tell “legitimate agent acting for a customer” from “malicious bot scraping inventory.” And Coinbase shipped x402 in May 2025 — reviving the long-dormant HTTP 402 status code so an API can demand a USDC stablecoin payment in the same handshake that returns the data.

The reason all four shipped within six months: the unit economics of agent transactions are not the unit economics of human ones. Recent on-chain data from x402 shows an average transaction size of around $0.20, against Visa’s human-card average closer to $80. Traditional rails charge ~$0.30 + 2.9% per transaction; on a $0.05 API call that’s a 600–700% effective fee rate. Either the rails get rebuilt or the agent economy never compounds.

The schedule for today references “$0.31 avg tx” — the closest verifiable public figure (Artemis on-chain analytics, March 2026) is ~$0.20 for x402, with sub-dollar tickets the consistent pattern. The point is the order of magnitude: agents transact in cents, not dollars.

The Four Architectures

Four Different Bets on the Same Problem

Each protocol answers “who carries the trust?” differently. Walk the four:

Authorization Layer
Who proves the agent can spend?
Four answers shipped between May and October 2025
ACP
Merchant of record + Shared Payment Token
AP2
Signed Intent + Cart Mandates
TAP
Domain-bound signed handshake
x402
HTTP 402 + USDC payload

AACP — Stripe + OpenAI, September 2025

An open spec maintained at agentic-commerce-protocol/agentic-commerce-protocol (GitHub). The model is conservative: the merchant stays merchant of record, the agent never holds the card, and the integration is one HTTP endpoint plus webhook. The agent ferries a Stripe-issued Shared Payment Token — a single-use token bound to a specific cart and amount. Etsy and ~30 Shopify pilots went live in ChatGPT’s Instant Checkout on Sept 29, 2025; OpenAI quietly removed Instant Checkout from ChatGPT on March 4, 2026 after low usage (only 8% of US adult users tried it in month one).

The pullback matters — it is not that ACP failed as a protocol, but that the user experience of buying inside chat did not yet beat opening a tab. The protocol survives the product retreat.

BAP2 — Google + 60+ partners, September 2025

An open spec at google-agentic-commerce/AP2. Payment-rail-agnostic by design — cards, stablecoins, RTP, account-to-account — and built around Mandates: tamper-proof, cryptographically signed digital contracts that capture user intent. AP2 distinguishes two flows: real-time / human-present (Intent Mandate from the prompt + Cart Mandate at user approval) and delegated / human-absent (a single richly-scoped Intent Mandate signed up-front, e.g. “buy these tickets the moment they go on sale, up to $400”). Launch partners include Mastercard, PayPal, Amex, Adyen, Coinbase, Etsy, Salesforce, ServiceNow, Worldpay, Revolut, JCB, and Ant International — a coalition that, like A2A’s, is broader than any single vendor’s reach.

CTAP — Visa + Cloudflare, October 2025

Trusted Agent Protocol is a thinner abstraction than ACP or AP2 — it does not specify the payment, only the identity handshake. Merchants today cannot tell apart a malicious scraper, a legitimate price-comparison bot, and an agent acting for a real customer. TAP solves that one problem: the agent presents a signature timestamped and bound to the merchant’s domain plus the specific operation (browse vs. checkout vs. dispute). The reference implementation is visa/trusted-agent-protocol on GitHub, with a sample Cloudflare-Worker proxy. Visa announced “hundreds of secure agent-initiated transactions” in the launch wave and predicts millions by holiday 2026.

Dx402 — Coinbase, May 2025

The most radical of the four: skip the card network entirely. coinbase/x402 revives HTTP’s long-dormant 402 Payment Required status code. A server returns 402 with a payment requirement; the client (often an agent) signs a stablecoin payment payload (USDC by default, multi-chain in v2) and replays the request with an X-PAYMENT header. Settlement is sub-second, atomicity is on-chain, fees are basis-points not cents. Real volume is small — ~$28K daily across ~131K transactions in March 2026 per Artemis — and roughly half is wash trading, but the architectural argument (machine-to-machine commerce needs machine-native rails) is the bet of the next decade.


Mandate Lifecycle (AP2)

The One Diagram That Unlocks AP2

If you only memorize one mechanism from today, make it the AP2 Mandate flow — it is the most architecturally honest of the four protocols and you will see its fingerprints in every payment-network agent product through 2026.

USER
Prompt — “Find me a non-stop SFO→NRT in business class for under $4,800, refundable.”
AGENT
Captures user intent, asks for ambiguities, then issues an Intent Mandate: signed JSON {scope, max_amount, currency, expiry, allowed_merchants, payment_method}
AGENT
Searches, negotiates, returns cart — e.g., JAL JL2 $4,540 incl. tax, refundable fare class ‘J’
USER
Reviews cart → signs Cart Mandate: {cart_id, amount, merchant, timestamp, intent_mandate_ref}
MERCHANT
Verifies both mandates against principal’s public key — intent scope covers the cart, cart not expired, signatures valid
PSP
Processes payment over chosen rail (card / stablecoin / RTP); rail is orthogonal — mandate proves authorization regardless
AUDIT
Both mandates persist as immutable records for chargebacks, regulator audits, and downstream agent disputes

The “human-not-present” flow collapses the first four steps: the user pre-signs an Intent Mandate with rich constraints (“buy concert tickets the moment they go on sale, max $400, only Ticketmaster, expires in 30 days”), and the agent later acts on it without a Cart Mandate signing ceremony. The Intent Mandate is the consent for that bounded space.

// Simplified AP2 Cart Mandate, signed by principal { "mandate_type": "cart", "intent_ref": "int_8f3a...", "cart_id": "crt_jal_jl2_20260601", "amount": { "value": 4540.00, "currency": "USD" }, "merchant": "jal.com", "payment_method": "vc_4242_4242", "issued_at": "2026-04-26T09:14:22Z", "expires_at": "2026-04-26T09:24:22Z", "signature": "MEUCIQDvJ..." // ECDSA over the canonical JSON }
The 10-minute expires_at is the source of the runtime bugs that the Zero-Trust Verification paper (below) attacks — under retries and concurrency, a mandate can be replayed within its window unless the merchant tracks consume-once nonces.

Papers to Know

Three Papers That Will Show Up in Engineer Conversations

arXiv preprint 2026 Security
Zero-Trust Runtime Verification for Agentic Payment Protocols: Mitigating Replay and Context-Binding Failures in AP2
Qianlong Lan, Anuj Kaul, Shaun Jones, Stephanie Westrum · arXiv:2602.06345 · submitted Feb 2026

The first published security analysis of AP2’s mandate lifecycle. The paper shows that while AP2 has signature, expiration, and binding semantics at the spec level, real implementations break those assumptions under retries, concurrency, and orchestration. The proposed runtime layer enforces context-binding and consume-once nonces, mitigates replay attacks at ~3.8 ms verification latency, and sustains 10,000 tx/s.

Why it matters: when AP2 hits production at scale, this is the threat model the implementing PSPs (Adyen, Worldpay, Stripe) will be reviewed against. Speak this paper’s language and you sound like someone who deploys, not someone who reads launch posts.
arxiv.org/abs/2602.06345
arXiv preprint 2026 Web3 / TEE
A402: Bridging Web 3.0 Payments and Web 2.0 Services with Atomic Service Channels
Yue Li, Lei Wang, Kaixuan Wang, Zhiqiang Yang, Ke Wang, Zhi Guan, Jianbo Gao · arXiv:2603.01179 · March 2026

A formal critique of x402 from the cryptography side. The authors show x402 does not enforce end-to-end atomicity — the client can pay and never receive the service, or receive the service and never pay. A402 introduces Atomic Service Channels (ASCs) using TEE-assisted adaptor signatures so that payment finalizes if and only if the service is correctly executed and result delivered. Reports orders-of-magnitude better performance and on-chain cost than x402.

Why it matters: x402’s missing atomicity is the most cited reason enterprises hesitate. A402-style fixes are how stablecoin rails earn enterprise trust by 2027. Knowing this paper signals “I have read the criticism, not just the launch post.”
arxiv.org/abs/2603.01179
arXiv preprint 2025 Survey
Inter-Agent Trust Models: A Comparative Study of Brief, Claim, Proof, Stake, Reputation and Constraint in Agentic Web Protocol Design—A2A, AP2, ERC-8004, and Beyond
arXiv:2511.03434 · November 2025

The first cross-protocol comparison framework. Six trust primitives — Brief (third-party verifiable claims), Claim (self-proclaimed identity, e.g. AgentCard), Proof (cryptographic / ZK / TEE), Stake (bonded collateral), Reputation, and Constraint (sandboxing) — mapped onto A2A, AP2, and Ethereum’s ERC-8004. The thesis: no single mechanism suffices; high-impact actions need Proof + Stake, augmented by Brief for discovery and Reputation for social signal.

Why it matters: this is the rosetta stone — one mental model for comparing every agent-trust protocol that ships next. When Susan asks “is this AP2 or A2A or something else,” this paper’s vocabulary gives the precise answer.
arxiv.org/abs/2511.03434

GitHub Pulse

Five Repos to Watch — The Plumbing Layer Is Where Power Concentrates

coinbase/x402 Hot ~5.7K★
HTTP-native payment protocol; reference implementation, multi-chain SDK, Cloudflare facilitator.
Architecturally the most radical — bypasses card networks entirely. Watch the v2 multi-chain rollout.
google-agentic-commerce/AP2 Spec ~2.6K★
Agent Payments Protocol reference implementation, sample Android wallet, mandate schemas.
Will likely become the “de facto” mandate spec because of the partner breadth (60+ at launch).
agentic-commerce-protocol/
agentic-commerce-protocol
Spec ~1.2K★
Stripe + OpenAI’s ACP spec, examples, and merchant adapters.
Survives ChatGPT’s Instant Checkout retreat — the protocol is independent of the product surface.
visa/trusted-agent-protocol New actively maintained
TAP spec + sample agent registry, merchant backend, CDN proxy, frontend, and TAP agent. Built with Cloudflare.
The bot-vs-agent identity layer underneath all four protocols. Strategic regardless of which checkout standard wins.
stripe/ai actively maintained
Stripe’s reference SDKs and cookbooks for building AI agents and ACP-compliant merchant endpoints.
The path Stripe is signaling for any merchant that wants to be addressable by an agent.
bitrefill/awesome-agentic-payments actively maintained
Curated list of protocols, SDKs, facilitators, and tools across the agentic-payments stack.
Use as a single-page situational awareness map of what shipped this quarter.

Community Pulse

What Practitioners Are Actually Saying

PC
Patrick Collison
@patrickc · Stripe CEO · Sept 29, 2025

“We have three cool announcements today: (1) @OpenAI is launching commerce in ChatGPT. Their new Instant Checkout is powered by @stripe. (2) We’re releasing the Agentic Commerce Protocol, codeveloped by Stripe and OpenAI. (3) @stripe is launching an API for agentic payments.”

Source: x.com/patrickc/status/1972716417280860391. The triple-announcement pattern matters — Stripe is signalling that ACP is not a feature of Instant Checkout but the foundation Stripe is betting on regardless of which surface wins.

DA
Dario Amodei
Anthropic CEO · March 2026 interview

Amodei has called the recent wave of agentic-commerce announcements “kind of frothy,” suggesting many of them are framed as breakthroughs but are still upstream of revenue.

Paraphrase from a Fortune interview cycle. The Anthropic posture (visible in Project Deal — the in-house Claude marketplace experiment that closed 186 deals on $4K total) is to treat agent commerce as a research surface first, a product second.

G
Google Cloud team
AP2 launch blog · Sept 2025

Google has argued that the right mental model for AP2 is the “mandate” from the legal world — a tamper-proof signed contract proving the principal authorized this agent for this scope, with the rail (card, stablecoin, RTP) deliberately left as a free variable.

The framing is strategically smart for Google: by making the rail orthogonal, AP2 can sit underneath both Stripe’s card flows and Coinbase’s stablecoin flows without picking a winner.

A
Anthropic · Project Deal team
anthropic.com/features/project-deal · Dec 2025

From the Project Deal writeup, Anthropic has noted that when both buyer and seller use equally capable models, deals price fairly — but when one side runs a smarter model, the smarter agent silently wins, mirroring real-world information asymmetry and pointing at a new class of agent-vs-agent fairness problems.

186 deals across $4K of in-office trades is small, but the qualitative finding (smarter agent extracts surplus) is the question every CGO selling B2B agents needs to answer for their enterprise buyers.


Platform Deep-Dive

How Each Stack Treats Agentic Commerce Today

PlatformPostureRecent move
Claude (Anthropic)Research-first; Project Deal as in-house experiment; Computer Use can drive any web checkout flow todayMar 2026 Computer Use research preview; Apr 2026 ended subscription access for 3rd-party agent frameworks — signals desire for direct economic relationship with end users
OpenClaw 2.1Hub-and-spoke skill orchestration; commerce skills are user-installed plugins; supports both ACP and AP2 mandates via skill manifestsAdapter skills shipped for ACP and AP2 in Q1 2026; sandbox model fits TAP’s domain-bound signature requirements naturally
Codex / GPT-5.4Native ACP since Sept 2025; Instant Checkout product retired Mar 4, 2026 but the protocol stays; agentic SDK exposes Shared Payment TokensApril 2026 reset is “ACP keeps shipping; we’re rethinking the consumer surface” — merchants stay integrated
Gemini 3.1Reference platform for AP2; Vertex AI agent SDK exposes mandate signing as a first-class primitive; Workspace agents pre-wiredCloud Next 2026 made signed Cart Mandates the default for any Vertex agent that touches payment APIs

Three implementation details to know:

1OpenClaw skill manifests can declare a payment_protocol field

A skill that wants to act commercially must declare payment_protocol: acp, ap2, or x402 in its SKILL.md. The hub denies execution if the user has not granted that protocol’s consent — this is the same gate as filesystem or network access from Day 14, just extended to money.

2AP2 mandates are JWS-signed JSON, verifiable with public JWKS

The AP2 spec uses standard JOSE (JWS / JWK / JWKS), which means existing OAuth and API-gateway infrastructure can verify mandates without bespoke crypto code. This is why incumbent PSPs adopted AP2 quickly — their existing key-rotation and signing primitives carry over.

3x402 + AP2 are not exclusive

You can run an AP2 Mandate as the authorization layer and settle the actual payment over x402’s stablecoin rail. The Lightspark and Stellar integrations published in late 2025 do exactly this. The protocols compose — AP2 owns “is this allowed,” x402 owns “move the funds.”


Vocabulary

Engineer-Level Definitions

Agentic Commerce
Commerce in which a software agent — not the principal — performs discovery, cart assembly, and payment authorization on behalf of a human or business principal.
Avoid: equating it with “AI shopping assistants” (which only recommend — agents transact)
e.g. “The new SKU is addressable to agentic commerce because the merchant supports ACP.”
Mandate
A cryptographically signed declaration of a principal’s authorization scope. AP2 distinguishes Intent Mandate (what the agent may do) from Cart Mandate (this specific purchase is approved).
Avoid: confusing with OAuth token — mandates encode intent + scope + amount + counterparty, not just identity
e.g. “The Cart Mandate’s expiry is 10 minutes — we need consume-once enforcement.”
Shared Payment Token (SPT)
In ACP, a single-use Stripe-issued token bound to a specific cart and amount, ferried by the agent so it never holds the underlying card.
Avoid: treating it as a wallet credential — SPTs are per-transaction and disposable
e.g. “The agent never sees the PAN, only the SPT for this $42 cart.”
Merchant of Record (MoR)
The legal entity that takes the customer payment and bears chargeback / tax / regulatory liability. ACP keeps the merchant as MoR; some agent-wallet designs flip this to the agent operator.
e.g. “Etsy stays MoR in ChatGPT Instant Checkout — OpenAI doesn’t want chargeback exposure.”
x402 / HTTP 402
An HTTP-native payment protocol reviving the 402 status code. A server returns 402 with payment requirements; the client signs a stablecoin payload (USDC) and replays with an X-PAYMENT header.
Avoid: thinking of it as “just crypto” — the protocol is rail-agnostic and aims to support fiat too
e.g. “We could meter our embeddings API per call with x402 instead of monthly tiers.”
Trusted Agent Protocol (TAP)
Visa+Cloudflare protocol that lets a merchant cryptographically distinguish a legitimate AI agent acting for a customer from a malicious bot, by domain-bound signed handshake.
Avoid: confusing with payment auth — TAP is the identity layer, not the money-movement layer
e.g. “Without TAP, our WAF blocks the agent traffic as scraping.”
Atomic Service Channel (ASC)
From the A402 paper: a payment channel that binds service execution and payment finalization atomically using TEE-assisted adaptor signatures. Plugs the “pay and never receive” gap in x402.
e.g. “ASCs are how stablecoin rails earn enterprise trust — you don’t lose funds on a bad merchant.”
Consume-once Nonce
A merchant-side enforcement mechanism (proposed in arXiv:2602.06345) that prevents replay of a valid mandate within its expiry window by tracking single-use nonces.
e.g. “Without consume-once nonces, a 10-minute Cart Mandate is replayable 10,000 times under retries.”
Agent Wallet
A funding source held in an agent’s name (vs. a principal’s account). Agent wallets are common in x402 deployments — the agent is the economic counterparty itself, not just a delegate.
Avoid: assuming all agentic commerce uses agent wallets — ACP and AP2 deliberately avoid them in v1
e.g. “Our research-agent fleet has $50 of USDC each in dedicated agent wallets.”
Human-Present vs. Human-Absent
AP2’s two transaction modes. Human-present requires a Cart Mandate signing per purchase. Human-absent uses a richly-scoped Intent Mandate signed up-front to bound future autonomous purchases.
e.g. “The concert ticket flow is human-absent — we sign Intent Mandate at 9pm for tickets dropping at midnight.”
Mandate Replay Attack
Reusing a still-valid signed mandate to re-execute a purchase. The attack surface that the AP2 zero-trust runtime paper formalized in Feb 2026.
e.g. “Our middleware deduplicates by intent_ref+cart_id to block replay.”
Agent-of-Record
Emerging concept — the registered agent (with verifiable credentials) that a principal has officially designated to transact on their behalf. Visa’s Trusted Agent Registry is one early implementation.
e.g. “The bank only honors charges from the customer’s registered agent-of-record.”

Expert Questions

Five Questions That Signal You Have Read Past the Launch Posts

Q1
If AP2’s Cart Mandate has a 10-minute expiry but the merchant retries on 5xx, what stops the same mandate from being charged twice within the window? Does anyone actually implement consume-once nonces in production today, or is it still just the runtime-verification paper?
Q2
ChatGPT pulled Instant Checkout in March after low usage. Is the right read “ACP failed” or “the buy-in-chat surface failed but the protocol survives” — and what data would let us tell those apart?
Q3
x402 daily volume is ~$28K and arguably half wash trades. At what real volume does the “machine-native rails” thesis become defensible? Is it a chicken-and-egg problem with agent supply, or is the merchant friction the actual blocker?
Q4
In the Anthropic Project Deal experiment, the smarter agent silently extracted surplus from the dumber one. How does AP2 or ACP price-anchor against that — or is intent-policing inherently outside the payment protocol’s scope and a job for the agent framework instead?
Q5
Visa’s TAP and AP2’s mandates seem layered — TAP for “is this an agent we trust to talk to,” AP2 for “is this transaction authorized.” Where does the bank’s own KYC live in that stack? Does the agent need to inherit the principal’s KYC, or is the agent a separate KYC’d entity?

CGO Lens

What Today Means for botlearn.ai

Five Strategic Implications


Curriculum Tracker

17-Day Path to Peer-Level Agent Conversations

Day 01Full Agent Stack
Day 02Memory Architecture
Day 03Planning & Tool Use
Day 04RAG Deep Dive
Day 05Agent Frameworks
Day 06Benchmarks & Eval
Day 07Multi-Agent Systems
Day 08Computer Use Agents
Day 09Code Agents
Day 10Long-Horizon Tasks
Day 11Agent Safety
Day 12Agent Economics
Day 13Research Frontiers
Day 14OpenClaw Deep-Dive
Day 15A2A Protocols
Day 16Agentic Commerce — today
Day 17Synthesis — building your agent strategy