Google is laying the groundwork for a future where artificial intelligence (AI) agents can handle online payments on behalf of users, and its new protocol has support from more than 60 other companies.

In a Sept. 16 blog post, Google introduced the Agent Payments Protocol (AP2). The open standard that lets AI agents securely complete online transactions without requiring direct shopper involvement.

Google said AP2 “establishes a payment-agnostic framework for users, merchants, and payments providers to transact with confidence across all types of payment methods.”

At launch, other payments and technology companies have pledged support for the protocol, including Mastercard, PayPal, American Express and Coinbase. Salesforce, Shopify, Cloudflare and Etsy are also among the early adopters.

“With the rise of AI-driven commerce, trust and accountability are more important than ever,” said Luke Gebb, executive vice president of Amex Digital Labs at American Express, in a statement. “American Express is excited to contribute to the creation of AP2 as a protocol intended to protect customers and enable participation in the next generation of digital payments.”

Why Google created a new payments protocol for AI contexts

AI agents — or software programs that can act autonomously on a user’s behalf — can already perform a wide range of tasks, such as coordinating bookings or managing subscriptions.

In ecommerce, they’re increasingly being used to assist with product discovery, comparisons, and even automated checkout, as seen in AI shopping experiences Google has already showcased.

This emerging category — often referred to as agentic commerce — centers around large language model-powered agents that can interpret shopper intent and take action.

However, while the technology offers new levels of convenience, it also raises questions around user authorization, transaction authenticity and accountability, according to Google.

Most existing payment systems assume a human is present to approve each purchase. But as AI agents take on more of the shopping journey, Google argues that the industry needs a shared playbook for more secure, verifiable transactions.

That’s where AP2 comes in. The open standard “provides a trusted foundation to fuel a new era of AI-driven commerce,” the company said.

Google said the protocol establishes a common language for secure transactions between agents and merchants, helping avoid a fragmented ecosystem. It supports multiple payment methods – including cards, stablecoins, and real-time bank transfers — to create a consistent, scalable experience. The company added that AP2 also gives financial institutions the clarity needed to manage risk.

How AP2 works

At the heart of AP2 is a system of digital contracts, known as mandates.

These tamper-proof, cryptographically signed records serve as verifiable proof that a user authorized a transaction, Google said.

Each is backed by verifiable credentials (VCs) and forms part of an audit trail.

In real-time scenarios, where a user is actively shopping, a query such as “Find me white running shoes” creates what’s known as an “Intent Mandate,” which captures the request’s context. After the agent presents a cart, the shopper approves the purchase by signing a “Cart Mandate,” locking in the item details and price, Google said.

Delegated shopping works similarly, but without the user present. A shopper might instruct their agent to “buy concert tickets under $100 when they go on sale.” The Intent Mandate outlines conditions like price limits and timing. Once those are met, the agent automatically generates a Cart Mandate and completes the purchase.

In both cases, the chain of mandates links shopper intent to payment execution.

“This complete sequence — from intent, to cart, to payment — creates a non-repudiable audit trail that answers the critical questions of authorization and authenticity, providing a clear foundation for accountability,” Google explained.

Examples of AP2 in action

Google offered several examples of how AP2 could reshape ecommerce.

A shopper looking for a green winter jacket, for example, agrees to pay up to 20% more for that color. The agent monitors availability and completes the purchase as soon as it finds a match.

In another case, a customer shopping for a bike at a specific merchant shares a trip timeline with their agent. The agent relays this to the retailer’s own agent, which responds with a personalized, time-sensitive bundle — including a helmet and rack — at a discount, Google outlined.

More complex workflows are also possible. A user planning a weekend trip, for example, instructs their agent to book airfare and a hotel with a $700 total budget. The agent coordinates across airlines and hotel systems, finds a match, and executes both bookings at once, the company said.

Expanding to crypto and enterprise

Google said AP2 is designed to support a broad range of payment types, including stablecoins and cryptocurrencies.

To accelerate support for web3 ecosystems, it launched a protocol extension called A2A x402, built in collaboration with Coinbase, MetaMask, the Ethereum Foundation and others. The goal is to enable secure, agent-led crypto payments, Google assessed.

Several partners are now building AP2-compatible agents, which will be featured in Google’s AI Agent Marketplace. Developers can access the protocol and contribute to its evolution via the public GitHub repository.

In the enterprise space, Google said AP2 could be used to automate tasks like cloud service procurement or scaling software licenses based on real-time usage.

Part of a bigger AI shopping push

AP2 is one part of Google’s broader push into agentic commerce.

As reported by Digital Commerce 360 in May, the company is developing an agentic checkout feature that can automate parts of the shopping journey based on user-defined parameters.

Other players are racing into the space as well.

Perplexity, an AI-powered answer engine, has been rolling out a new AI browser called Comet as a competitor to Google’s Chrome. The browser includes agents that can browse, compare products, and complete purchases. It also integrated checkout tech from Firmly.ai earlier this year and partnered with PayPal to enable in-search transactions.

According to Reuters, OpenAI is reportedly developing a browser with similar agentic capabilities. The ChatGPT maker has also worked with retailers like eBay and Etsy to test its agentic commerce tool, Operator. Amazon, meanwhile, has rolled out its own agentic shopping solution called “Buy for Me.”

While the space is still evolving, many of these efforts could eventually support AP2 or similar standards as the industry works toward secure, scalable frameworks for agent-led shopping.

Do you rank in our databases? 

Submit your data and we’ll see where you fit in our next ranking update.

Sign up

Stay on top of the latest developments in the online retail industry. Sign up for a complimentary subscription to Digital Commerce 360 Retail News. Follow us on LinkedIn, X (formerly Twitter), Facebook and YouTube. Be the first to know when Digital Commerce 360 publishes news content.