Personal AI agents are increasingly performing transactions on behalf of users, but given the non-deterministic nature of large-scale language models, AI bots tend to do things they aren’t supposed to do. In this case, there is a high possibility that the AI will malfunction and conduct fraudulent transactions.
To solve this problem, American Express is introducing a new framework designed to help AI agents understand and authenticate a customer’s purchase intent before completing a transaction.
AmEx says it needs a way to ensure that a customer’s purchase instructions are executed throughout the transaction lifecycle, from purchase to all subsequent transactions.
The company calls it the Intent Intelligence Framework and hopes it will become the industry standard for AI-powered trading.
“These new transactions require consumers, merchants, and agents,” explained Hannah Bozian, AmEx’s vice president of agent partnerships and strategy, in an interview with TechCrunch. He said the company wants to ensure that agent transactions continue to be “the seamless, reliable and secure method cardholders expect.”
The framework release follows the company’s launch of Amex’s Agentic Commerce Experiences (ACE Developer Toolkit) in April, which includes a handful of services that include an intent framework.
AmEx said it has not yet seen an increase in disputes from proxy transactions, but it wants to be proactive in building out the technology as this type of business expands. One reason for this is that consumers have shown a willingness to try agent commerce.
In a recent report from Amex Trendex, 66% of U.S. consumers surveyed said they are likely to use an AI agent to find the best price for a product or service, and 61% are likely to use an AI agent to make product or service recommendations in the next 12 months, the company told TechCrunch.
First, Amex’s new intent framework starts with the Create Intent API. It captures the customer’s purchase instructions and any other constraints that may be included in the request, such as the seller’s name, maximum budget, and other product preferences. How this intent is collected may vary depending on the AI model involved, but Bojian said the plan is to make the framework flexible enough to support all agent commerce models.
The customer then authenticates their identity and signs off on the intent retrieved by the agent.
Next, an intent ID, a unique identifier representing the intent, is created along with a proof of intent token containing the customer’s high-level constraints. This token is used to obtain secure digital payment credentials that agents use at checkout in place of the customer’s actual credit card number. That way, AmEx says, purchases remain secure while consistent with what the customer has authorized the agent to do.
The intent ID and token can also be used in different parts of the transaction process, helping businesses manage areas such as fraud prevention and dispute resolution, Amex said.
“Our approach differs from other implementations in that it captures intent early in the transaction flow, and this is important because it helps mitigate some risk before it occurs,” Bojian said. She gave an example where a customer asked an agent to buy size 9 white Nike sneakers, but the agent sent a request to buy shoes from Adidas. In these cases, Amex may block the transaction due to the agent’s mistake.
Bojian also points out that while many agents currently seek customer consent before making a purchase, AmEx foresees a future where agents may act more autonomously in some cases. For example, a consumer might authorize an agent to reorder household essentials such as paper towels or toilet paper. In this category, the only real constraints are price and size, not seller or brand. In such a scenario, it makes sense to introduce technology within the loop that performs additional checks to ensure that the transaction matches the customer’s intent.
Logs of which agents authorized what and when could also help address fraud on behalf of agents, she noted.
AmEx is already in discussions with partners across the payments ecosystem to implement ACE, including agent tokens containing AmEx payment credentials and other parts of the developer kit.
The company said it plans to further actively develop ACE as the agent commerce ecosystem evolves.
On the Industry Stage at TechCrunch Disrupt 2026 on October 13th, Hannah Bozian, vice president of agent partnerships and strategy at American Express, will explain more about the company’s new Intent Intelligence Framework and what it means for the world of agentic commerce.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
