As the AI world reshapes everything around us, its impact is not limited to how we work or shop—it is fundamentally transforming the payments landscape as well. We are rapidly evolving from using AI to assist in payments—through orchestration layers, intelligent decisioning, and API-driven integrations—toward a new paradigm of agentic commerce.
In this new model, AI agents do not merely support user actions; they are increasingly capable of understanding intent, making decisions within defined boundaries, and executing transactions on behalf of users. This marks a shift from human-initiated payments to delegated, machine-executed commerce, where the role of the user transitions from operator to supervisor.
At the center of this transformation are global card schemes—particularly Visa and Mastercard—quietly building the infrastructure for a world where machines can transact with the same trust and reliability as humans. This evolution is also redefining a fundamental question in payments—not just whether a transaction is authorized, but whether it accurately reflects user intent.
It is worth noting that several of these initiatives are still in early or pilot phases. Visa Agentic Ready launched with a first cohort of European issuers, and Mastercard Verifiable Intent is still being defined in collaboration with industry bodies. The infrastructure described here represents the direction of travel—not a fully deployed reality.
What is emerging across these initiatives are four distinct yet complementary layers: Trust — can the agent be verified? Intent — what exactly was approved? Execution — can the payment be completed securely? And Experience — how does the interaction actually happen? Each scheme addresses one or more of these layers, and together they sketch the architecture of agentic commerce.
Let’s explore how each scheme addresses these layers.
1. Visa Trusted Agent Protocol — Trust at the Protocol Layer
Visa’s approach starts at the foundation: can the agent itself be trusted?
The Trusted Agent Protocol focuses on establishing this trust by introducing agent-level verification using cryptographic assertions and structured data elements exchanged with merchants.
At a high level, this approach enables agents to present verifiable context along with transaction requests, including:
- Agent Intent – A structured signal indicating that the agent is acting in a defined capacity, such as discovering product details or initiating a purchase on behalf of a user.
- Consumer Recognition – Data elements that help merchants identify whether the underlying consumer has an existing relationship, enabling continuity in experience and risk assessment.
- Payment Context – The ability for agents to carry payment-related information or tokens, aligned with the merchant’s preferred checkout and authorization mechanisms.
- This model effectively pushes trust establishment down to the protocol layer, ensuring that every agent-initiated interaction is identifiable, attributable, and verifiable at the network edge.
2. Visa Agentic Ready — Enabling Agent-Driven Experiences
Launched initially in Europe with a first cohort of issuing partners—including Barclays, HSBC UK, Banco Santander, and Revolut—Visa Agentic Ready is designed to prepare the payments ecosystem for the transition to agentic commerce.
In its current phase, the initiative focuses on issuer readiness, providing a structured framework for banks to test and validate agent-initiated transactions. This allows issuers to explore how such transactions can operate securely and at scale within controlled, production-like environments.
At its core, Visa Agentic Ready builds on Visa’s existing trust capabilities—bringing together tokenization, identity, risk management, and control frameworks—to evaluate how agent-initiated payments can be enabled across channels and use cases.
A key objective of this effort is to extend familiar consumer protections into AI-driven interactions. By leveraging mechanisms such as tokenization and biometric authentication, the model ensures that agent-initiated transactions remain strongly linked to a real user, with explicit consent and control embedded at critical decision points.
This positions Visa Agentic Ready as a crucial bridge between today’s payment experiences and a future where agents can seamlessly orchestrate and execute transactions on behalf of users—without compromising trust, security, or control.
While branded here as an experience layer, Visa Agentic Ready is perhaps better understood as the connective tissue between existing payment infrastructure and an agentic future — less about the end-user interaction itself, and more about ensuring the underlying systems are ready to support it.
3. Mastercard Agent Pay — Enabling Secure and Scalable Agent Execution
Mastercard Agent Pay is positioned as an execution-focused capability designed to support secure and scalable payments in the emerging landscape of agentic commerce.
At its core, it builds on a remote commerce tokenization framework, enabling AI agents to initiate and complete transactions using tokenized payment credentials rather than exposing sensitive card data. This ensures that agent-initiated payments remain aligned with existing security and acceptance models.
The solution brings together several of Mastercard’s core capabilities to support agent-mediated transactions:
- Network Tokenization (MDES) – Leveraging the Mastercard Digital Enablement Service to provision secure, domain-controlled payment tokens for agent usage
- Strong Authentication – Incorporating mechanisms such as payment passkeys to authenticate users and bind agent actions to verified identities
- Fraud and Cybersecurity Controls – Extending existing risk and fraud detection systems to scenarios where transactions are initiated by agents rather than humans
- Agent Pay Acceptance Framework – Providing merchants with the necessary constructs to recognize, validate, and process agent-initiated transactions within their existing checkout and authorization flows
- This approach focuses on execution and acceptance, ensuring that once trust and intent are established, agents can reliably complete transactions across merchant ecosystems using Mastercard’s global network.
In doing so, Mastercard Agent Pay acts as a critical layer that bridges AI-driven decisioning with real-world payment execution—at scale, and within established security boundaries.
4. Mastercard Verifiable Intent — The Intent and Trust Layer
Mastercard Verifiable Intent introduces an open, standards-based approach to establishing trust in agentic commerce, focused specifically on capturing and proving user intent.
Designed to be protocol-agnostic, it creates a tamper-resistant record of what a user has authorized when an AI agent acts on their behalf. This establishes a shared, verifiable source of truth across the ecosystem—providing cryptographic proof of authorization that can be relied upon by consumers, merchants, and issuers alike.
Rather than focusing solely on identity or execution, Verifiable Intent addresses a more fundamental question:
Did the agent act exactly as the user intended?
To support real-world adoption, Verifiable Intent is expected to integrate with Mastercard Agent Pay through intent-related APIs, enabling intent capture, validation, and linkage to downstream payment execution flows. This creates a cohesive model where intent verification and transaction execution are tightly coupled but independently verifiable.
In parallel, Mastercard is working with industry bodies to help define broader standards for AI-driven and conversational commerce, ensuring interoperability and consistency as agentic commerce ecosystems evolve.
Together, this positions Verifiable Intent as a critical layer in the stack—ensuring that in a world of autonomous agents, every transaction is not only authorized, but also provably aligned with user intent.
The Real Shift: Payments Are Becoming Intent Systems
What these approaches collectively reveal is a fundamental architectural shift in how payments are designed. Payments are no longer a single-step process of authorization. Instead, they are being decomposed into four distinct layers:
- Trust (Identity) — Can the agent be trusted? (Visa Protocol)
- Intent (Authorization Context) — What was actually approved? (Mastercard Verifiable Intent)
- Execution (Transaction Processing) — Can the payment be completed securely? (Mastercard Agent Pay)
- Experience (Orchestration) — How does the interaction happen? (Visa Agentic Ready)
This separation is critical, because in a world where machines act on behalf of humans, authorization alone is no longer sufficient—intent must be captured, preserved, and proven.
In the coming years, the winners in payments will not be defined solely by scale or speed—but by their ability to establish trust in autonomous interactions. As AI agents become participants in commerce, payments will evolve from simple transaction rails into systems of intent, trust, and verification.
The payment schemes that will define the next decade are not waiting for agentic commerce to arrive — they are building the rails for it now. What this infrastructure ultimately enables is a world where trust is cryptographic, intent is auditable, and execution is seamless. The shift is already underway — and the institutions that move early to understand and implement these frameworks will be best positioned to lead in it. If you want to explore more
If you’re exploring what agentic payments means for your business — whether that’s evaluating scheme readiness, understanding implementation pathways, or designing for an AI-driven commerce experience:
