Personal AI agents are starting to make payments on their own, but LLMs are nondeterministic, so there is a real risk of an agent completing a transaction the user never intended. American Express is addressing that with the Intent Intelligence Framework, a set of services that capture and authenticate a customer’s purchase intent before an AI agent can complete a purchase. The company hopes the framework becomes an industry standard for AI-powered transactions. “These new transactions will have a consumer, a merchant, and an agent,” said Hannah Bozian, Amex’s VP of Agentic Partnerships and Strategy, and the goal is to keep agentic transactions “seamless, trusted, and secure” for card members.

The framework is part of Amex’s ACE Developer Toolkit, which launched in April and now includes the intent framework among its services. Amex says it has not yet seen an increase in disputes from agentic commerce, but it wants to be proactive as this type of commerce scales. Recent Amex Trendex data cited by TechCrunch shows 66% of surveyed U.S. consumers are likely to use AI agents to research the best price for a product or service, and 61% are likely to use them for product or service recommendations over the next 12 months.

The flow begins with a Create Intent API that captures the customer’s purchase instructions and any constraints in the request, such as merchant name, maximum budget, and product preferences. How intent is collected may vary by AI model; Amex says the framework is designed to be flexible enough to support all agentic commerce models. The customer then authenticates their identity and signs off on the captured intent. This creates an Intent ID and a Proof of Intent Token containing the customer’s high-level constraints. The token is used to obtain a secure digital payment credential that the agent presents at checkout instead of the customer’s actual credit card number. Amex says the same ID and token can be reused across the transaction lifecycle for fraud prevention and dispute resolution.

Bozian says the approach differs from other implementations because it pulls intent earlier into the transaction flow, mitigating risk before it occurs. She gave the example of a customer asking an agent to buy white Nike sneakers in size 9 while the agent submits a request for Adidas shoes; Amex could prevent the transaction from going through because the agent made a mistake.

Amex also expects agents to become more autonomous in some cases. A consumer might let an agent re-order household staples such as paper towels or toilet paper, where the relevant constraints are price and size rather than merchant or brand. In those scenarios, Amex sees a need for technology in the loop to check that the transaction matches the customer’s intent. Maintaining a log of what was authorized by which agent and when could also help address fraud on behalf of the agent.

Amex is already in discussions with partners across the payments ecosystem to implement ACE, including agentic tokens containing Amex payment credentials, and plans to develop ACE further as the ecosystem evolves. Bozian will discuss the framework in more detail at TechCrunch Disrupt 2026’s Industry Stage on October 13.