Concepts and protocols
What is agentic commerce?
Agentic commerce is AI agents finding, buying and paying on someone's behalf. The main approaches, the protocols behind them, and where control sits.
Agentic commerce means AI agents carrying out purchases and payments on behalf of a person or business: finding a product or service, agreeing a price, and paying, within permissions the person set.
Why is this harder than it sounds?
Payments were built around a person at a checkout: someone types a card number, clicks a button and can be asked to confirm. An agent breaks each of those assumptions. It has no card of its own, it may run while nobody is watching, and a seller can't easily tell an authorised agent from an attacker. So agentic commerce is less about new ways to move money than about new ways to delegate: how a person grants an agent limited, provable permission to spend, and how a seller knows that permission is real.
How do you do it?
- Shopping on someone's behalf. An assistant buys from a merchant for a user. The Agentic Commerce Protocol, developed by Stripe and OpenAI, covers this: coordinating checkout and sharing payment credentials as secure tokens, with the business staying merchant of record.
- Delegated card credentials. Card-based approaches give the agent a scoped credential rather than the card itself, such as a Stripe Shared Payment Token, which Stripe describes as a scoped grant with usage and expiration limits.
- Mandates and verifiable intent. AP2, the Agent Payments Protocol that Google published, records what the user authorised as signed, verifiable credentials it calls mandates, available as an extension for A2A and the Universal Commerce Protocol.
- Machine payments for APIs. For pay-per-call services, protocols such as x402 and MPP let a server answer with a payment request that the agent pays programmatically. Stripe supports both.
- Policy-controlled wallets. An agent requests payments from a wallet the owner controls, through a service that enforces limits, recipients and approvals.
What it does not mean
- It doesn't mean agents with unlimited access to your money. Every serious approach is about narrowing what the agent can do.
- It isn't one standard. ACP, AP2, x402 and MPP solve different parts of the problem and can coexist.
- It isn't only crypto. Card-based and stablecoin-based approaches are both in use.
Common questions
Is agentic commerce live today?
Parts of it are. OpenAI implements ACP in ChatGPT, and Stripe documents machine payments over MPP and x402. Availability varies by country and integration, so check each provider's docs.
Is this financial advice?
No. This page explains the concept and the approaches. It doesn't recommend any product, asset or payment method.
Where does Veyra fit?
Veyra sits in the last category: a permission layer between an agent and a wallet you already own. The agent gets an MCP endpoint with tools like create_payment; Veyra applies your per-payment, daily, auto-approve and ask limits and recipient allowlist, and pays in USDC on Base, either signed by you in your wallet or, on paid plans, from a capped allowance. It does not implement ACP, AP2, x402 or MPP, and it doesn't issue cards.
Sources
Terms on this page
- Shared Payment Token (SPT)A Shared Payment Token (SPT) is a Stripe credential that grants an agent or seller scoped use of a customer's payment method, such as a card, with usage and expiration limits.
- HTTP 402 Payment Required402 Payment Required is an HTTP response status code that RFC 9110 lists as 'reserved for future use', meaning HTTP itself defines no standard behaviour for it.
- Stablecoin (USDC)A stablecoin is a blockchain token designed to keep a stable value relative to a reference asset, usually a currency such as the US dollar; USDC is a dollar stablecoin issued by Circle.
This page explains how agent payments and the related technology work. It is not financial, investment, legal or tax advice. Protocols and provider features change; check the linked official sources before relying on a detail.