← Go back

Guide // MandateGuard

What is the EMVCo draft framework for agentic payments?

By the Ingenious AI team · Published 2026-11-14 · Reviewed 2026-11-14

▸ Run the free live demo See pricing

Free covers 1 agent and 1 policy with the full authorize flow, signed receipts and the approval inbox — no card required to start.

Short answer

The EMVCo draft framework for agentic payments (draft 1, September 2026) defines how agent mandates should work as shared infrastructure: a registered object with a principal, an agent, merchant scope, cumulative and per-period budgets, an authority window and a managed lifecycle, with transactions verifiable against it. MandateGuard's interoperable intent service is aligned with that draft — the same anatomy, implemented as working software — and claims alignment only, never certification. The practical takeaway for buyers: judge draft-aligned systems by live drills over their shared state, not by standards language.

What is the EMVCo agentic payments framework?

The EMVCo draft framework for agentic payments is a draft specification (draft 1, dated September 2026 in MandateGuard's materials) describing how mandates — scoped, verifiable spending permissions — should work when AI agents transact. MandateGuard's interoperable intent service is aligned with that draft's model: registered intents with principal, agent, merchant scope, cumulative and per-period budgets, recurring authority windows and shared state. Alignment means the model matches the draft's shape; it is not a certification by EMVCo or anyone else.

Why payments bodies are writing agent rules at all

Card payments work because every party — issuer, merchant, network — shares definitions of authorisation, liability and proof. Agent payments break the old definitions: the buyer is software acting for a principal, the instruction may be a goal rather than an itemised order, and the moment of consent is a configuration made days earlier. Frameworks like EMVCo's draft exist to rebuild shared definitions for this world: what counts as the principal's mandate, how an agent's authority is scoped and evidenced, and how a merchant or issuer-side checker confirms that a given transaction sat inside a real permission. Until such definitions settle, every agent-commerce relationship improvises them bilaterally — which is why a registered, verifiable intent is valuable right now.

The draft's core concept: the mandate as a first-class object

The conceptual heart of the draft approach is that the mandate is an object, not a message. It has an issuer and a principal. It names its agent. It scopes its merchants. It bounds value cumulatively and per period, and time through an authority window. It has a lifecycle — active, suspended, revoked — that authorised parties manage and all parties can observe. And transactions against it are evidenced in a way that survives the moment: receipts that a checker can verify against the mandate rather than against the agent's assurances. If that anatomy sounds like a description of MandateGuard's intent service, that is the alignment claim, stated exactly: the same object, with the same fields and lifecycle, implemented as working software.

What draft alignment means in MandateGuard

Concretely, alignment shows up in five product facts. Intents are registered with principal, agent, merchant scope, cumulative and per-period budgets and a recurring authority window. Authorisation against an intent decrements its budget atomically and links the receipt into the intent's own hash-chained event log. Post-transaction events — refunds, adjustments, chargeback notes — maintain the shared figures. Role-scoped keys separate issuers (write) from merchants and agents (read and verify). And cross-party receipt verification checks hash, signature, workspace chain and intent linkage. Those are implementable, testable behaviours, which is the right way to honour a draft: build its model faithfully enough that if the final framework shifts, the mapping work is small.

What alignment does not mean

It does not mean certification, endorsement or compliance status, and MandateGuard's materials are careful to say the alignment is with a draft — draft 1, September 2026 — not a final standard. That care is itself a signal worth reading correctly. Agent-payments marketing is filling with certification-flavoured language for standards that have not finished existing. A vendor that distinguishes alignment from certification on its own pages is telling you how it will handle other claims — prices, counts, capabilities — and the fleet-wide rule behind MandateGuard's copy is that claims must be checkable against the product. When you evaluate any vendor in this space, apply the same test: ask which of their standards words correspond to a drill you can run today.

Why the framework matters even if you never register an intent

Most MandateGuard users govern their own agents with policies, not shared intents. The framework still matters to them for two reasons. First, direction: mandates are becoming the standard unit of agent spending authority, so skills and records built around mandates (receipts, budgets, scoped authority) appreciate rather than strand. Second, counterparties: as merchants and issuers adopt mandate verification, an organisation whose agents already operate under registered, verifiable mandates will onboard faster and be trusted sooner. Building your internal practice on the draft's anatomy — principal, scope, budgets, window, lifecycle — is therefore cheap future-proofing even for purely internal fleets.

Reading drafts like an operator

Operators should read emerging frameworks the way engineers read RFCs: for the model, the invariants and the vocabulary, not for badges. Ask of any draft-aligned system: which invariants does the draft demand (scoped authority, verifiable evidence, lifecycle control), and can I watch this system uphold them in a live drill? MandateGuard invites exactly that drill through its registry and demo surfaces. Standards will keep moving — drafts become finals, versions turn — and the systems that age well are the ones whose state is honest and exportable, because migration is a data problem long before it is a branding one. Keep your mandates registered, chained and verified, and whichever final shape the framework takes, your history arrives intact.

Five actions to take now

  1. Treat EMVCo references as draft alignment until the framework finalises
  2. Build mandates with the framework's anatomy: principal, agent, scope, budgets, window
  3. Keep your intent records verifiable — alignment means little without checkable state
  4. Watch the framework's evolution at your policy review, not in your marketing
  5. Never claim certification you do not hold, in either direction of a deal

Worked example: evaluating alignment claims honestly

An issuer evaluating MandateGuard for a mandate programme asks the certification question directly. The honest answer structures the whole evaluation: the intent service is aligned with the EMVCo draft framework — its intents carry the draft's anatomy of principal, agent, merchant scope, budgets and authority window, with role-scoped verification — and no certification exists to claim, because the framework is a draft. The issuer's team then evaluates the thing that can actually be evaluated today: the working state machine.

They register a test intent, run authorisations, append a refund, suspend and resume, and have their merchant partner verify receipts with a read-only key. The draft alignment told them the model would feel familiar; the live drill told them it works. That ordering — standards as orientation, working state as evidence — is how agent-payment infrastructure should be bought while the standards are still drying.

Frequently asked questions

Is MandateGuard certified under the framework?

No — and it does not claim to be. The framework is a draft; MandateGuard claims alignment of its intent model with the draft, nothing more.

Who is EMVCo in payments?

EMVCo is the payments industry body behind the EMV specifications used by card networks globally. Its move into agentic payments signals mandates becoming mainstream infrastructure.

Does alignment affect AP2-style intents?

The intent model is described as AP2-style (mandate-centred, after Google's Agent Payments Protocol) and EMVCo-draft-aligned. Both references describe the same registered-intent anatomy.

Should I wait for the final framework before using intents?

Mandates solve present problems. Using a draft-aligned, working system now, with honest claims, beats deferring controls until standards finalise.

Where can I see the alignment in the product?

In the intent service itself: registration fields, role-scoped keys, verification and the registry UI are the draft's concepts made operational.

Keep reading

Put the policy in force: spend envelopes · intent registry · approval inbox · how the proof works · all MandateGuard guides