Personal AI agents are increasingly stepping into the digital marketplace to make transactions on their users’ behalf, but the inherent nondeterministic nature of large language models means these autonomous bots are prone to executing unintended actions. This reality raises a significant operational and financial risk: the possibility of AI misfiring and authorizing unauthorized transactions without direct human oversight at the final checkout screen.
To get ahead of this problem before it escalates into a widespread industry crisis, American Express is introducing a new technological framework designed to capture and authenticate a customer’s exact purchase intent long before an AI agent completes the final transaction.
The core idea is to ensure that a customer’s purchase instructions are reliably carried throughout the entire transaction lifecycle, according to American Express. This continuity spans from the initial product search and intent creation to everything that follows during the checkout and post-purchase phases.
Dubbed the “Intent Intelligence Framework,” the company hopes this proprietary solution will eventually become a widely adopted industry standard for all AI-powered transactions.
These new forms of commerce fundamentally alter the traditional buyer-seller dynamic by introducing a third distinct participant, explained Amex’s Vice President of Agentic Partnerships and Strategy, Hannah Bozian, in an interview. She noted that the company wanted to ensure agentic transactions remain as seamless, trusted, and secure as card members have historically expected traditional transactions to be.
The release of this new framework follows the company’s April launch of Amex’s Agentic Commerce Experiences, commonly referred to as the ACE Developer Toolkit. The toolkit provides a handful of developer services that have now been expanded to include the new intent framework.
Amex noted that it has not yet seen a measurable increase in consumer disputes specifically driven by agentic commerce. However, the legacy financial institution wanted to be proactive in building out this safety infrastructure as this emerging type of commerce scales up across the digital economy. Part of that urgency stems from clear signals that consumers are increasingly open to experimenting with agentic commerce tools for their everyday shopping needs.
In a recent Amex Trendex report, 66% of surveyed U.S. consumers stated they are likely to use AI agents to research the best price for a product or service over the coming months. Furthermore, 61% indicated they are likely to use an AI agent for product or service recommendations over the next 12 months, according to data shared by the company.
To kick off the safety process, Amex’s new intent framework begins with a Create Intent API. This application programming interface is designed to capture the customer’s specific purchase instructions along with any additional constraints they may have included in their request, such as the designated merchant’s name, maximum budget limits, and other nuanced product preferences. How this intent is collected may vary depending on the underlying AI model involved in the user’s request. However, the broader strategic plan is to make the framework flexible enough to support all existing and emerging agentic commerce models, Bozian explained.
Once the parameters are set, the customer authenticates their identity and officially signs off on the purchase intent captured by the AI agent.
Following this authorization, an Intent ID—a unique digital identifier representing the specific intent—is generated, alongside a Proof of Intent Token. This token encapsulates the customer’s high-level constraints and parameters. The token is then utilized to obtain a secure digital payment credential that the AI agent will present at checkout, rather than exposing the customer’s actual credit card number. By doing so, Amex says the purchase remains entirely secure while staying strictly aligned with what the customer originally authorized the agent to execute.
This Intent ID and associated token can also be utilized across various other stages of the transaction process, helping companies effectively manage critical business areas such as proactive fraud prevention and streamlined dispute resolution, according to Amex.
The company’s approach stands out in the broader fintech and payments landscape because it pulls intent much earlier into the transaction flow. This timing is vital because it helps mitigate financial and operational risk well before the risk actually occurs, Bozian emphasized.
To illustrate this vulnerability, she offered a hypothetical scenario: a customer instructs their personal AI agent to purchase a pair of white Nike sneakers in a size 9, but the AI model hallucinates or makes an error, submitting a request to purchase a completely different pair of shoes from Adidas instead. In such a scenario, the Amex framework could successfully intercept and prevent the transaction from going through because the AI agent deviated from the explicit parameters of the user’s intent.
Bozian also pointed out that while many AI agents operating today routinely ask for explicit customer consent before finalizing any purchase, Amex foresees a future ecosystem where agents will act with a much higher degree of autonomy in specific, low-risk contexts. For instance, a consumer might grant their personal agent standing permission to automatically re-order household staples, such as paper towels or toilet paper. In these routine categories, the only real constraints are price thresholds and package sizes, rather than specific merchant loyalty or brand preferences. In these automated scenarios, having specialized technology permanently in the loop performing continuous background checks is essential to ensuring the completed transactions strictly match the customer’s true intent.
Maintaining a comprehensive digital log of what was authorized, by which specific agent, and at what precise time could also help address emerging forms of fraud committed on behalf of AI agents, she noted.
Amex is already actively engaged in discussions with various partners across the broader payments ecosystem regarding the implementation of ACE. This rollout includes the agentic tokens containing Amex’s secure payment credentials, as well as the integration of other modular components housed within the developer kit.
The company stated that it plans to continue developing the ACE infrastructure aggressively as the wider agentic commerce ecosystem matures and changes over the coming years.
The development of the Intent Intelligence Framework will take center stage at upcoming industry events. On October 13 at TechCrunch Disrupt 2026’s Industry Stage, Hannah Bozian, VP of Agentic Partnerships and Strategy at American Express, is scheduled to discuss the company’s new framework in greater detail, exploring what these technological safeguards mean for the future of agentic commerce.
