American Express has published a new framework designed to help card issuers verify purchases made by artificial intelligence-powered shopping agents, marking another step forward as the financial industry adapts to automated commerce.

The payments giant released its Intent Intelligence framework, outlining a standardized method to record a shopper’s purchase instructions before AI agents buy products or services on their behalf. Card issuers can utilize these records to help detect fraud, approve payments, and resolve disputes when transactions do not go as planned. For retailers, the system provides a clear mechanism to determine whether an ordering mistake originated with an autonomous AI agent rather than the merchant or the customer.

While American Express has published the technical specifications to encourage broader industry adoption, the company confirmed that Intent Intelligence does not currently feature live integrations. Instead, the move is intended to allow other payment processors and technology providers to build upon its approach, fostering a common ecosystem standard for agentic commerce rather than introducing immediate consumer-facing capabilities.

"Providing authorized purchase intent before an agent transacts can give issuers greater confidence that an agent is acting as a customer intended, while also helping issuers proactively manage risk and reduce the chance of costly disputes across the payments ecosystem," said Luke Gebb, head of global innovation at American Express, in the company’s announcement.

The rollout arrives as major financial networks and technology companies rush to develop tools for agentic commerce. In this emerging landscape, advanced AI agents are increasingly capable of searching for products, comparing prices, and executing transactions independently on behalf of consumers.

Despite the rapid technological progress, consumers remain cautious about relinquishing control over their spending to algorithms. A survey of 1,000 U.S. adults released in August by payments firm NMI revealed that 58% of respondents worried an AI shopping agent might purchase the wrong item, size, color, or brand. Furthermore, 60% expressed concern that an automated agent could spend more money than intended.

Although American Express initially introduced Intent Intelligence as part of a broader suite in April, the recent publication makes its underlying technical framework publicly accessible for other organizations to integrate. The stated goal is to help shape a unified industry approach for handling intent as agentic commerce continues to mature.

What the New American Express Framework Means for Retailers

For merchants, the rise of AI-driven purchasing introduces novel operational and customer service challenges. A retailer might fulfill an order precisely as it was received from an AI agent, only to face a subsequent chargeback or transaction dispute because the agent ultimately purchased an item the consumer never actually requested.

This is where documented records of a shopper’s original instructions, referred to as purchase intent, become vital. By preserving clear evidence of what the customer authorized, the framework helps establish accountability by showing whether the AI agent accurately followed the user’s directives.

Intent Intelligence is one of several services American Express originally introduced through its Agentic Commerce Experiences developer kit. The company noted that the framework differentiates itself from current transaction methods by transmitting a customer’s authorized purchase intent to the card issuer much earlier in the transaction lifecycle, well before an agent executes the final payment.

The framework supports a wide variety of customizable purchase controls. Shoppers can establish strict spending caps, expiration dates, and merchant preferences. They can also define specific operational triggers, such as instructing an agent to complete a purchase only when an out-of-stock item returns to inventory or when a specific promotional event like Cyber Monday begins.

In addition to basic retail purchases, the framework accommodates more sophisticated, multi-step transactions. For instance, a consumer could authorize an agent to purchase travel insurance only if it successfully secures the accompanying flight tickets. It can also manage cross-merchant transactions, such as simultaneously acquiring shoes and matching socks from entirely different online retailers.

How American Express Intent Intelligence Framework Works

The Intent Intelligence system is designed to operate seamlessly across various agentic commerce architectures, regardless of which entity initially captures the consumer’s instructions. Depending on how a specific AI agent manages its payment workflows, it can route the instructions directly to a card issuer, or alternatively pass them through a digital wallet provider or another authorized intermediary.

Beyond verifying transactions, the framework serves as a foundational component of American Express’s broader strategy to protect consumers and merchants from AI-related errors. When the company unveiled its developer kit, it outlined plans for "Agent Purchase Protection." This proposed coverage would apply to eligible transactions executed by registered AI agents that utilize the developer kit and successfully transmit authenticated purchase instructions. For merchants, the ultimate objective is to minimize costly chargebacks and disputes.

However, American Express confirmed that these protections are not yet live, as there are currently no active ACE integrations utilizing Intent Intelligence—a mandatory requirement for transactions to qualify. In the interim, card members continue to rely on the standard protections and benefits already associated with their eligible American Express cards.

This evolving landscape also introduces complexity regarding consumer perceptions of liability. In the NMI survey, 34% of consumers indicated they would blame the retailer if an AI tool executed a purchase prematurely, while 24% said they would hold their payment provider responsible. Meanwhile, 62% stated they would place primary blame on the AI provider itself.

Where American Express Fits in Agentic Commerce

American Express is far from alone in developing infrastructure to make AI-powered shopping more secure and reliable. Traditional financial institutions and major technology companies are actively establishing competing standards and protocols.

In early 2025, Visa and Mastercard both launched dedicated offerings allowing developers to connect AI agents directly to their global payment networks. Visa’s Intelligent Commerce program provides developers with application programming interfaces and tools to embed payment capabilities within AI agents, backed by partnerships with major technology firms including OpenAI, Microsoft, Anthropic, Stripe, and Samsung. Concurrently, Mastercard introduced Agent Pay to facilitate automated consumer and business-to-business transactions through conversational AI platforms.

Simultaneously, technology developers have forged their own pathways for agentic checkout. OpenAI and Stripe co-developed the Agentic Commerce Protocol to let AI agents securely interface with merchants. Google introduced its Agent Payments Protocol, utilizing digitally signed records to verify shopper authorization with early backing from major commerce platforms like Shopify, Salesforce, Cloudflare, and Etsy.

Google subsequently expanded its footprint with the Universal Commerce Protocol, an open standard developed alongside Shopify and other industry partners to streamline checkout within AI shopping environments. Major retailers, including Newegg, Wayfair, Walmart, and Macy’s, have begun integrating the protocol to allow direct purchases through AI-driven search interfaces.