EMVCo Publishes Draft Agentic Payments Framework: What Intent Services Actually Solves
EMVCo, run by six card networks, has released a draft framework for card-based agentic payments. A breakdown of its core concept Intent Services and how it differs from Verifiable Intent, AP2, ACP and UCP.
Key Takeaways
- EMVCo, the body behind EMV chip specifications and 3-D Secure, has released a draft framework for card-based agentic payments and is collecting comments until 30 September 2026
- Its core concept, Intent Services, defines a shared state layer where the purchasing authority a consumer grants to an AI agent can be registered, referenced and updated by multiple participants
- It does not replace the cryptographic proof of authorization provided by Verifiable Intent or AP2. EMVCo positions it as a common coordination point that complements them
EMVCo Releases a Draft Framework for Agentic Payments
EMVCo has released a draft framework to help promote secure, interoperable and scalable card-based agentic payments.
www.businesswire.comOn 1 September 2026, EMVCo published a draft framework for card-based agentic payments. The document is titled "EMV® Agentic Payments: Framework for Specifications v1.0." As the name suggests, it is not a specification itself but a foundation for future specification development. Comments are open until 30 September 2026.
Junya Tanaka, EMVCo Executive Committee Chair, framed the release this way.
Card-based agentic payments require a globally interoperable foundation that consumers, merchants and issuers can all trust. This publication marks an important step forward by outlining a consistent approach for establishing and communicating intent alongside payment-related information.
EMVCo also confirmed that a dedicated Agentic Payments Task Force is already in place. The task force itself was reported at the Board of Advisors meeting in April 2026, and this draft is its first published output.
Why It Matters That EMVCo Is the One Moving
Announcements about agentic payments have arrived at a relentless pace over the past year. What sets this one apart is the source.
EMVCo is a technical body jointly operated by six card networks: Visa, Mastercard, American Express, JCB, Discover and UnionPay. It began in the late 1990s with Europay, Mastercard and Visa, with JCB, American Express, China UnionPay and Discover joining later. The EMV chip specification, EMV 3-D Secure (3DS) for online authentication, and EMV Payment Tokenisation all sit under its management. If your team has ever spent a quarter on 3-D Secure 2.0 compliance, this is where that specification came from.
There is a difference between one network announcing its own mechanism and six of them starting to build consensus at the same table. The latter carries weight with merchants and issuers that a single-network initiative does not.
Looking back, EMVCo had already signalled its intent to work on agentic payments in November 2025. That earlier statement stayed at the level of "we will examine how EMV 3DS, EMV Payment Tokenisation and EMV SRC can be extended." Roughly ten months later, a concrete framework has emerged.
The Idea Behind Intent Services
The core concept in the draft is Intent Services. In an explainer by Clinton Allen, chair of the Agentic Payments Task Force, it is described as a shared, interoperable layer that lets payment participants register, reference, retrieve and manage consumer-authorised intent before, during and after a transaction. The draft covers ecosystem roles, the data fields used to register intent, how lifecycle and state information is maintained, and how authorised parties retrieve intent-related data.
What deserves attention is that EMVCo chose managing the state of intent rather than proving intent as its problem to solve.
Proving intent already has several answers. Mastercard Verifiable Intent uses an SD-JWT delegation chain to lock down who authorized what and within what bounds, and AP2 attacks the same problem with signed Mandates. For a one-off purchase, that is sufficient. The consumer approves, the approval persists as a signature, and the merchant and issuer verify it.
The difficulty appears when intent spans time. EMVCo cites three examples: recurring purchases, cumulative budgets and post-transaction activities. Instructions like "restock this supplement monthly" or "spend up to a set amount in this category this month" are not settled by a single signature. A budget balance shrinks with each purchase and grows back on a return. Recurring purchase conditions change mid-stream.
In those scenarios, a static signed artifact is not enough. If the agent, the merchant, the PSP and the issuer each decide separately how much budget remains or whether a recurring arrangement is still valid, their answers diverge. What EMVCo is trying to supply is a place to look it up. The announcement describes it as a common coordination point that complements the cryptographic assurance provided by existing solutions such as Verifiable Intent.
Cryptographic signatures answer whether an intent is genuine. Intent Services answers what is left of that intent right now. The roles do not overlap, which is why complement is the operative word.
How It Differs From What Already Exists
This is where most readers will focus. Agentic payments is crowded with similarly named initiatives, and whether the EMVCo draft competes with them or completes them is a fair question.
The short answer is that each one works on a different layer.
| Initiative | Owner | Layer it addresses | Role in agentic payments |
|---|---|---|---|
| EMV Agentic Payments Framework (draft) | EMVCo | Shared intent state | A coordination point where participants register, reference and update the same intent |
| Verifiable Intent | Mastercard and Google | Proof of authorization | Records the delegation and its constraints as a tamper-resistant signed artifact |
| Agent Pay | Mastercard | Card credentials | Issues agent-specific tokens that plug into issuer-side decisioning |
| Trusted Agent Protocol | Visa | Agent identity | Uses request signatures to prove an agent's legitimacy to the merchant |
| AP2 | Authorization handoff | Links intent, cart and payment method through signed Mandates | |
| ACP and UCP | OpenAI and Stripe / Google | Discovery and checkout | Connects product discovery, cart and order placement from the conversational surface |
Visa Intelligent Commerce and its core Trusted Agent Protocol handle proving an agent's identity to the merchant. Without that, the agent is simply blocked as a bot. Mastercard Agent Pay takes a different route, issuing tokens specific to agent transactions so the traffic stays on existing card rails while remaining identifiable to the issuer.
ACP and UCP sit at a different altitude entirely. They connect the commerce journey from product discovery through cart construction to order placement, with payment as one component. The EMVCo draft scopes itself to card payments, so the two do not collide directly.
Does it compete with Verifiable Intent? EMVCo does not say so. It names Verifiable Intent explicitly and describes its own work as complementary. Verifiable Intent, for its part, is built on open standards from the FIDO Alliance, EMVCo, IETF and W3C, so both have been looking at the same standards base from the start.
The other point worth noting is that EMVCo positions this draft as a starting point for extending other EMV technologies. It states that the framework will help identify enhancements to EMV 3DS, EMV Payment Tokenisation, EMV SRC and the newer EMV Digital Payment Credential (DPC). If the 3DS authentication flow is revisited with agent transactions in mind, that lands directly on merchant implementations.
Listed as possible future work are Know Your Agent (KYA) and Agentic Transaction Indicators. The former would cover identifying an agent and communicating its attributes, the latter would signal that a transaction involved an agent acting for a consumer. Both are described only as capabilities EMVCo may consider. Timing and specification shape are undisclosed.
Reasons Not to Get Ahead of This
Read only the proponents' side and this looks like clean progress. Some reservations are warranted.
Eric Grover, principal at the payments consultancy Intrepid Ventures, told Digital Transactions that while he welcomes the effort, he flagged how slow and political the process tends to be. Reaching technical consensus across competing payment networks while simultaneously coordinating with the FIDO Alliance, W3C and the OpenID Foundation is, in his words, notoriously slow. Anyone who remembers how long merchant adoption of EMV 3DS 2.0 took will not find that exaggerated.
There is also scepticism about agentic payments themselves. In the same article, Cliff Gray of Lakeland Scientific argues that handing purchasing authority to software deserves considerable caution. A trust layer being technically available is not the same as consumers actually delegating their spending.
Ironically, EMVCo itself acknowledges the risk of fragmentation. Its explainer notes that the agentic payments ecosystem comprises various existing and emerging standards and protocols, that this poses a fragmentation risk, and that this is precisely why it collaborates with other bodies. The flip side is that EMVCo's own entry may add one more fragment. Because the framework is scoped to card payments, it cannot cover account-to-account transfers or stablecoin settlement, so merchants will be tracking several frameworks in parallel.
Several conditions remain undisclosed: when a formal specification would land, who operates Intent Services and under what model, and whether support will be optional for merchants or effectively mandatory.
What Ecommerce Operators Should Watch
In practical terms, there is no implementation work to start today. The draft is in a comment period, and no API or technical requirement exists yet.
What is worth watching is where this framework will connect. EMV 3DS, Payment Tokenisation and SRC, the technologies EMVCo named, are already in use by most ecommerce operators through their payment providers. If agentic payment support arrives as extensions to those, the work will surface not as adopting a new standard but as upgrading the payment stack you already run. It is worth checking whether your payment provider participates in EMVCo as an Associate or Subscriber.
The second implication is for merchants running subscriptions. The scenarios Intent Services targets are exactly the world of recurring billing and cumulative budgets. When agent-initiated recurring purchases actually arrive, how they connect to existing subscription infrastructure will be the design decision that demands the most attention. If you are evaluating agentic commerce readiness, keeping this layer in view within the broader agentic payments picture will speed up later calls.
Submitting a comment is also an option. The draft is available from EMVCo's resource page and the deadline is 30 September. When a specification hardens without merchant operational reality reflected in it, the implementation cost tends to land on merchants.
Conclusion
The EMVCo draft signals that the agentic payments conversation is moving one layer deeper, from proof toward state management. Purchases that are not settled by a single approval were never solvable with signatures alone.
That said, this is still only a framework. The distance from six networks agreeing to merchants and issuers implementing is long, and the market may move faster than the process. What changes after 30 September is the next thing to watch.
