American Express Debuts Agentic Commerce Toolkit for AI Transactions – But Validation Process Remains Opaque

American Express Debuts Agentic Commerce Toolkit for AI Transactions – But Validation Process Remains Opaque

American Express (Amex) has unveiled a developer toolkit called Agentic Commerce Experiences (ACE) that allows AI agents to shop and pay on behalf of users—but only within its own payment network, and with a validation process that critics say remains a black box.

Luke Gebb, Amex's EVP and global head of innovation, told VentureBeat that the company believes this closed-loop system is the missing piece in agentic commerce. “Some of what is missing so far is the perspective of a company like ours: We feel that trust and security are critical to advancing this space,” Gebb said. “This is really the first time that an issuer is coming to the table.”

Unlike other card issuers such as Chase or Bank of America, Amex can route transactions through its own American Express Network. Visa and Mastercard are payment networks but do not issue cards themselves, forcing them to work with banks. This gives Amex unique control over the payment layer—a control that ACE leverages.

However, the ACE kit still keeps its verification process under wraps. Amex says agents built with ACE can submit shopping carts and check them against the agent's original intent, but it has not disclosed how this matching works. Raj Ananthanpillai, founder and CEO of identity verification firm Trua, told VentureBeat that despite strides in creating a trust layer, many black boxes remain that could hinder widespread adoption.

Background

Agentic commerce—where AI agents autonomously purchase goods and services—faces fundamental hurdles: trust, control, accountability, validation, and security. Consumers fear rogue agents draining accounts; merchants worry about unpaid items; banks dread chargebacks and fraud.

Current protocols, such as Google's Agent Pay Protocol (AP2), focus on interoperability. Amex's ACE kit goes further by embedding full transaction control in the payment layer, but its closed-loop system means it cannot be independently audited. The company abstracts how it performs validation, describing only that it uses a mix of deterministic checks and semantic evaluation.

What This Means

For consumers, ACE offers potential protection against unauthorized AI purchases—but only if they trust Amex's opaque validation. Merchants may gain confidence in agent-led transactions, yet the lack of transparency could still lead to disputes. Banks face reduced fraud exposure, but only within the Amex ecosystem.

Industry observers warn that without open validation standards, agentic commerce will struggle to scale. “We need verifiable intent and outcome matching that is visible to all parties,” said Ananthanpillai. “Otherwise, we’re just swapping one black box for another.”

Amex’s move is a significant step, but the continued secrecy around its validation process means the dream of fully trustworthy AI shopping remains incomplete. As Gebb acknowledged, “Trust and security are critical to advancing this space.” Yet for now, that trust must be placed in Amex’s hands alone.