American Express unveils Intent Intelligence Framework to govern AI agent purchases

American Express has introduced the Intent Intelligence Framework, a system designed to ensure AI agents only execute purchases their users have explicitly authorized. The framework is positioned as a proactive measure to mitigate risks as agentic commerce scales, particularly as consumers express openness to AI-mediated transactions.

The framework’s core component is a Create Intent API, which captures a customer’s purchase instructions and constraints—such as merchant name, maximum budget, and product preferences—before the transaction is initiated. This intent is then authenticated via customer identity verification, after which an Intent ID and a Proof of Intent Token are generated. The token enables secure checkout using a digital payment credential tied to the user’s authorization, rather than exposing the user’s actual credit card number.

Amex’s framework is integrated into the company’s ACE Developer Toolkit, which launched in April as part of the Agentic Commerce Experiences suite. The toolkit includes services that now encompass the intent framework, reflecting Amex’s stated goal of embedding security and trust into agent-driven commerce workflows.

According to Amex, 66% of surveyed U.S. consumers reported they are likely to use AI agents to research the best price for a product or service over the next 12 months, and 61% said they are likely to use agents for product or service recommendations. The company emphasizes that while it has not yet observed a rise in disputes attributable to agentic commerce, it aims to preemptively address potential misuse as adoption grows.

The Intent ID and Proof of Intent Token are designed to support downstream functions such as fraud prevention and dispute resolution by providing a verifiable record of user intent. Amex asserts that its approach differs from others by embedding intent verification earlier in the transaction flow, thereby reducing risk before it materializes. For example, the framework could block a transaction where an agent mistakenly attempts to purchase Adidas shoes when the user instructed a search for white Nike sneakers in size 9.

Amex also anticipates scenarios where agents operate with greater autonomy, such as reordering household staples like paper towels or toilet paper, where constraints are limited to price and size. In such cases, the framework’s verification layer would ensure transactions align with user intent, even as agents act more independently.

The company is in discussions with partners across the payments ecosystem to implement the ACE toolkit, which includes agentic tokens containing Amex payment credentials. Amex plans to continue developing the framework in tandem with the evolving agentic commerce landscape.

Amex’s VP of Agentic Partnerships and Strategy, Hannah Bozian, will discuss the Intent Intelligence Framework in detail at TechCrunch Disrupt 2026 on October 13, highlighting its implications for the future of secure agentic commerce.