The Dawn of Agentic Commerce: EMVCo Sets the Stage for AI-Driven Payments
As artificial intelligence shifts from a tool for information retrieval to a proactive participant in commerce, the global payments infrastructure is undergoing a fundamental transformation. The emergence of "agentic payments"—transactions initiated by AI agents acting on behalf of consumers—is challenging the traditional, human-centric architecture of card-based networks.
To bridge the gap between AI autonomy and financial security, EMVCo, the global standards body for card payments, is spearheading a technical framework designed to standardize how intent, authority, and agent identification are communicated across the payment ecosystem. This development marks a critical step toward ensuring that as consumers delegate their spending power to machines, the financial networks that support them remain secure, transparent, and interoperable.
The Paradigm Shift: From Manual Transactions to Autonomous Intent
In the conventional payment model, a transaction is a discrete, synchronous event. A human consumer selects an item, initiates a checkout, and provides authentication—usually in the form of a card swipe, dip, or digital wallet tap. The merchant, issuer, and payment network receive this signal and process it in real-time.
However, the rise of AI agents introduces a temporal and logical disconnect. An AI agent might be instructed by a user to manage recurring subscriptions, monitor for price drops on specific goods, or operate within a strict budget across a diverse array of merchants. Consequently, the actual "authorization" of a payment may occur long before the transaction itself takes place.
Current payment systems were not built to exchange information regarding the intent behind a transaction. They are designed to verify that a credential is valid and that the account has sufficient funds. They lack a standardized way to signal that a transaction is the result of an autonomous agent operating under a specific set of user-defined constraints.
Chronology of the EMVCo Framework Development
The industry’s recognition of this "agentic gap" has accelerated throughout 2024. EMVCo, which manages the specifications underpinning billions of global card payments, has moved quickly to address the risks and opportunities inherent in this shift.
- Early 2024: As generative AI tools gained traction, financial institutions began reporting "ghost" transactions or unexpected behavior from AI-powered shopping assistants. The need for a standardized "handshake" between agents and banks became apparent.
- September 1, 2024: EMVCo issued a formal request for public comment regarding its Agentic Payments — Framework for Specifications. This document outlined a "common foundation" for card-based agentic payments, focusing on consumer intent, agent identification, and transaction signaling.
- September 30, 2024: The public consultation period officially closed. The EMVCo Agentic Payments Task Force began the complex process of aggregating industry feedback from stakeholders, including major card networks, fintech innovators, and global banks.
- Post-September 2024 (Current Phase): The Task Force is currently in the collation and review phase. The findings from this period will dictate whether EMVCo pursues entirely new technical specifications or updates existing protocols, such as EMV 3-D Secure or Payment Tokenization.
Supporting Data: The Consumer Appetite for AI Delegation
The drive toward agentic payments is not merely a technological push; it is a response to changing consumer behavior. According to data from PYMNTS Intelligence, the "AI-powered shopper" is already a reality, though their comfort levels vary by task.
The June report, “Global Digital Shopping Index: The AI-Powered Shopper Has Arrived,” revealed that 56% of consumers are eager to use AI agents to search and compare products. When it comes to financial delegation, the numbers are significant:
- 37% of consumers are comfortable allowing an AI agent to authorize payments on their behalf.
- 36% are open to predictive automatic buying—where an agent anticipates needs and executes purchases without manual intervention.
- 35% are willing to grant AI agents direct access to their saved payment credentials.
Furthermore, the shift is already impacting the retail funnel. An August report from PYMNTS Intelligence, titled “The 50 Million Consumer Migration,” found that 19% of U.S. consumers now begin their retail research journey with an AI. The impact is measurable: 43% of these users found better pricing, 27% switched brands based on AI recommendations, and 26% switched products entirely. As consumers grow more accustomed to AI-driven discovery, the transition to AI-driven execution (payment) is the logical next step.
Official Perspectives: The Role of Standards
EMVCo’s approach is defined by neutrality. The organization aims to build the "pipes" through which information flows without dictating the "business rules" that banks or merchants choose to apply.
Oliver Manahan, director of engagement and operations for EMVCo, emphasized that the framework is strictly technical. "It does not define the business rules; in other words, how a product is used," Manahan told PYMNTS.
Instead, the framework proposes the creation of "Intent Services." These services act as a shared layer where participants can register, reference, retrieve, and manage consumer-authorized intent. This is critical for the lifecycle of a transaction. For instance, if an AI agent makes a purchase that a consumer later disputes, the issuer needs a way to verify what the agent was actually authorized to do. By creating a repository of intent, EMVCo aims to provide the necessary audit trail for dispute resolution.
Regarding the identification of these agents, EMVCo is exploring "Know Your Agent" (KYA) capabilities. Similar to "Know Your Customer" (KYC) regulations, KYA would allow banks and merchants to verify the identity of the software agent involved in a transaction, ensuring that the agent is legitimate and operating within an accountability chain.
Implications: A New Era of Risk and Transparency
The implications of this framework are profound for every participant in the payment ecosystem.
For Issuers and Networks
The primary benefit for issuers is improved risk assessment. Currently, an agentic transaction might look like a standard "Card Not Present" (CNP) transaction. By signaling that an agent was involved, issuers can adjust their fraud detection models to account for the specific risk profiles of AI-automated shopping.
For Merchants
For merchants, the framework offers a path to higher conversion rates. If a merchant can verify that an agent has the authority to complete a transaction, the friction associated with extra authentication steps can be reduced. However, merchants must also prepare to interface with these "Intent Services" to ensure they are compliant with the consumer’s delegated limits.
For Consumers
The most significant impact on the consumer will be the balance between convenience and control. As consumers delegate more authority to AI, the risk of "over-spending" or unauthorized agent behavior increases. The EMVCo framework attempts to solve this by embedding the "rules of engagement" into the payment flow. If an agent tries to purchase an item that violates the consumer’s budget, the network will be able to block the transaction at the point of origin because the "Intent Service" will signal that the action falls outside of the authorized scope.
The Road Ahead: Technical Evolution
As the Agentic Payments Task Force synthesizes the feedback from the September consultation, the industry awaits the next series of formal announcements. The framework is expected to influence several established EMV technologies:
- EMV 3-D Secure: Enhancing this protocol to incorporate agent-specific metadata during the authentication process.
- Payment Tokenization: Creating tokens that are not just tied to a card, but also to specific agent authorizations.
- Digital Payment Credentials: Evolving the credential to include the "Intent" layer as a core component of the transaction data packet.
The challenge, as Manahan noted, is ensuring that the ecosystem remains scalable. "EMVCo does recognize that intent must be continuously managed from a lifecycle perspective," he stated.
The successful implementation of these standards will determine whether the next decade of commerce is characterized by a seamless, automated flow of value or a period of fragmentation and uncertainty. By moving proactively to establish these technical guardrails, EMVCo is attempting to ensure that even as the agents grow more intelligent, the payments they execute remain as reliable and secure as the traditional methods they are beginning to replace.
As the industry pivots toward this AI-first future, the focus will shift from "who is paying" to "what was intended." In the world of agentic payments, the accuracy of that intent will be the most valuable currency of all.
