Glossary

What is Agent Payments Protocol (AP2)?

The Agent Payments Protocol (AP2) is an open, payment-agnostic standard, announced by Google in September 2025, that lets an AI agent prove a real user authorized a specific purchase. It uses cryptographically signed digital contracts called mandates to create a tamper-proof audit trail between agents, merchants and payment networks.

Last updated:

AP2 answers the question every merchant and card network asks about agentic purchasing: how do we know the human actually agreed to this? The protocol's core object is the mandate — a cryptographically signed record of what the user authorized (item, price, conditions), which travels with the transaction and settles disputes about consent and liability.

AP2 launched with 60+ partners including Mastercard, PayPal, American Express and Coinbase, and extends the A2A protocol rather than competing with it. The v0.2.0 release (April 2026) added "Human Not Present" flows, letting an agent complete a pre-authorized purchase the moment conditions are met. Google has announced it is donating AP2 to the FIDO Alliance so the standard remains platform-neutral (as of August 2026).

AP2 governs the payment leg of an agentic transaction. Everything before it — discovery, qualification, structured Q&A between a buyer's agent and a vendor — is the layer AgentDoor works at. The two are complementary: a deal surfaced and scoped through an AgentDoor-style front door can hand off to AP2-governed rails when money actually moves.

Sources: [1] [2]

Frequently asked questions

How is AP2 different from the A2A protocol?
A2A standardises how independent agents talk to each other; AP2 extends it with the payment-specific layer — signed mandates proving user authorization. A purchasing flow can use A2A for the conversation and AP2 for the transaction.

Skip the discovery calls.

Describe what you need and let AgentDoor's agents interview the vendors for you — you get a decision-ready shortlist, not six meetings.

We'll reach out at launch. No spam, ever.