top of page
Website Quantum Payments (1).png

Know Your Agent: Visa, Mastercard and Ant Just Built the Trust Layer for Agentic Payments

Writer: Quantum Payments
Quantum Payments
Sep 12
8 min read

On 10 September 2026, Visa, Mastercard and Ant International announced a collaboration that could become the most important infrastructure agreement in agentic payments so far: a shared Know-Your-Agent (KYA) interoperability framework.

The announcement was made through BuildFin.ai, an industry platform convened by the Monetary Authority of Singapore. Its purpose is straightforward: allow an AI agent verified within one payment ecosystem to be recognised by others without repeating the entire registration process.

The stakes are substantial. The companies cited McKinsey projections that AI agents could handle between US$3 trillion and US$5 trillion in global consumer commerce by 2030. But the forecast is not the most important part of the announcement.

The real question is: who gets to vouch for the machine — and who gets paid for doing so?

From Know Your Customer to Know Your Agent

Traditional payments are built around identifiable parties. A customer authorises a transaction, a merchant accepts it, an issuer approves it and the payment network provides the rules, controls and settlement infrastructure.

Agentic payments introduce another participant: software that can search, compare, select and pay on a user’s behalf.

That creates a new trust problem. A payment system must be able to answer:

  • Which agent initiated the action?

  • Who operates or controls that agent?

  • Which customer, business or organisation authorised it?

  • What was the agent allowed to do?

  • Has the agent behaved consistently over time?

  • Who is accountable if the agent makes a mistake?

KYA is designed to provide those answers.

In plain English, it is a way for payment networks, digital wallets, marketplaces and agent platforms to share common trust signals. An agent verified on one network should not need to start from zero when it encounters another.

The proposed framework focuses on three areas:

  1. Cross-network operator traceability — linking each agent to a validated cardholder, business or organisation.

  2. Shared certification requirements — assessing whether an agent meets agreed security and behavioural standards.

  3. Continuous transaction monitoring — using identity and transaction signals to evaluate agent behaviour over time.

The distinction is important: Visa, Mastercard and Ant are not handing over their risk engines to one another. Each network retains its own verification and decisioning processes. They are aligning the signals that make an agent recognisable and accountable across ecosystems.

A central AI-agent identity node surrounded by verification rings and encrypted payment trails

Three rival protocols, one trust layer

The KYA collaboration is also a recognition that the market was heading towards fragmentation.

Over the past year, each participant has introduced its own agentic payments architecture:

  • Visa’s Trusted Agent Protocol

  • Mastercard’s Verifiable Intent

  • Ant International’s Agentic Mobile Protocol

These protocols are strategically valuable because they can determine how agents authenticate, how customer intent is represented and how risk is managed at the point of payment.

But a world where every agent must register separately with every network would quickly become expensive and difficult to scale. It would also create inconsistent trust experiences for merchants and customers.

The timing of the announcement is revealing. On 9 September, Mastercard launched Agent Connect, designed to give merchants a single integration for AI-agent shopping flows spanning product discovery, recommendations, cart construction and payment. Visa and Revolut also announced a live passkey-authenticated agentic payment in France on 7 September, using tokenised credentials and a pre-authorised customer mandate.

These developments show that the race is moving beyond demonstrations. The infrastructure is being assembled now.

The analyst view: this pact buys position, not revenue

It is tempting to read the KYA announcement as a standardisation story. That would miss the strategic tension.

The agreement commoditises the trust layer while each participant still wants its own protocol to become the place where value accumulates.

That is the central paradox.

Interoperability is essential if agentic commerce is to scale. Merchants do not want to build separate integrations for every agent platform, wallet and card network. Consumers do not want to repeatedly authorise the same identity. Payment providers need reliable signals to distinguish legitimate automation from malicious bot activity.

At the same time, the company that controls the most valuable identity, intent, risk and transaction data may control the economics of machine-originated commerce.

KYA therefore gives Visa, Mastercard and Ant a stronger strategic position, but it does not settle the most important commercial question: where will machine-originated dollars actually settle?

Will the agent’s wallet provider control the relationship? Will the card network own the trust decision? Will the merchant’s payment provider determine routing? Or will the winning layer be the one that combines identity, tokenisation, AI fraud detection and transaction optimisation into a single operating model?

The answer will determine more than brand visibility. It will influence authorisation economics, data access, dispute responsibility and the payment success rate for agent-initiated transactions.

For a wider explanation of how this market is developing, see Quantum Payments’ analysis of the US$3 trillion agentic commerce opportunity and the emerging machine wallet.

What KYA means for merchants

Merchants should not treat KYA as a distant standards exercise. It points towards practical changes in how payment acceptance will work.

1. Agent verification becomes a payment requirement

Merchants will need to know whether an automated request comes from a recognised agent operating within an approved mandate or from an untrusted bot.

That means payment infrastructure must be able to receive and interpret agent identity, operator and authorisation signals without treating all automation as suspicious.

2. Intent validation will matter as much as authentication

A verified agent is not automatically a trustworthy transaction.

The payment flow must also establish that the transaction matches the customer’s or business’s intended rules: spend limits, product categories, delivery requirements, timing and account permissions.

This is where tokenised credentials and verifiable intent become important. The goal is to prove not only that an agent is known, but that it is acting within its authority.

3. Fraud engines must adapt

Many existing fraud systems are designed to detect unusual speed, repeated activity, scripted behaviour and automated browsing. Those same signals can describe a legitimate shopping agent.

AI fraud detection for payments will need to distinguish between authorised automation and hostile automation. That requires richer context, including agent reputation, operator identity, historical behaviour, token status and the relationship between the agent and the underlying customer.

4. AI payment routing will become more valuable

An agent may compare prices, delivery times and payment options across multiple merchants. Merchants, in turn, will need intelligent routing to optimise approval rates, costs and authentication requirements.

AI payment routing could help select the most effective gateway, acquirer, token or network for a given transaction — while keeping the experience invisible to the customer.

5. Reconciliation must remain unified

Agentic commerce will not remove the need for accurate settlement, refunds, chargebacks, tax records and financial reporting. It will increase the number of systems involved.

Merchants should prioritise a unified commerce and reconciliation model that connects online payments, in-person transactions, wallets, agent channels and back-office accounting.

Three rival AI payment protocols connecting through a luminous interoperability mesh

A practical merchant checklist

Businesses preparing for agentic payments should ask their payment provider:

  • Can the platform identify and authenticate recognised AI agents?

  • Can it capture the agent, operator and customer relationship?

  • Does it support tokenised credentials and verifiable intent?

  • Can its fraud models distinguish approved agents from malicious bots?

  • Does it support AI payment routing across gateways and acquirers?

  • Can merchants expose structured product, pricing, inventory and fulfilment data?

  • Are agent transactions visible in reporting, reconciliation and dispute workflows?

  • Can the same controls work across e-commerce, mobile, terminals and SoftPOS?

  • How are permissions revoked if an agent changes behaviour?

  • Who carries liability when an agent acts outside its intended mandate?

The answers will matter well before the US$3–5 trillion forecast becomes reality.

Why Australia should pay attention

Australia is unlikely to be isolated from this shift. Australian retailers, marketplaces, travel companies, logistics businesses and subscription providers already operate across digital wallets, card networks, mobile experiences and physical points of sale.

The regional significance of Singapore’s BuildFin.ai and the MAS safeguards work is particularly relevant. It signals that Asia-Pacific regulators and payment leaders are trying to establish responsible operating principles before agentic commerce becomes mainstream.

For Australian merchants, the immediate priority is not choosing a particular agent protocol. It is ensuring that the underlying payment stack is flexible enough to support several protocols without creating several disconnected operating environments.

That means investing in strong identity controls, tokenisation, structured commerce data, adaptive fraud decisioning, smart routing and unified reconciliation.

The trust layer is only the beginning

Visa, Mastercard and Ant International have made an important concession: agentic payments will not scale through isolated proprietary systems.

But interoperability does not eliminate competition. It moves the competition to a more valuable layer: the ability to verify intent, manage risk, optimise acceptance and connect every machine-originated transaction back to a real customer, business or organisation.

That is why the KYA framework matters.

It is not simply a technical standard. It is a strategic attempt to define the accountability layer for the machine economy.

The winners will not necessarily be the companies with the loudest agentic commerce forecasts. They will be the businesses that can make autonomous transactions feel as secure, observable and financially manageable as human-led payments.

For merchants, the message is clear: agentic payments are becoming less about whether an AI can click “buy” and more about whether the entire commerce stack can trust what happens next.

For a foundation in the subject, read Agentic Payments Explained and Quantum Payments’ analysis of the UPI Unified Agent Protocol.

Sources

SEO metadata

Meta title: Know Your Agent: The Trust Layer for Agentic Payments

Meta description: Visa, Mastercard and Ant International are aligning on Know-Your-Agent interoperability. Here is what KYA means for merchants, AI payments and fraud.

Daily handover to Sonny

Wix CMS slug:/blog/know-your-agent-trust-layer-agentic-payments

Publication target: Saturday, 12 September 2026, 7:00am AEST

LinkedIn post 1 : 8:08am AEST

Angle: News-led strategic analysis : the real contest is ownership of the machine trust layer, not the headline commerce forecast.

Visa, Mastercard and Ant International are aligning on a shared Know-Your-Agent framework for agentic payments.

The headline number is US$3–5 trillion in AI-orchestrated commerce by 2030.

But the bigger question is: who gets to vouch for the machine?

KYA aims to let an AI agent verified in one payment ecosystem be recognised across others, while each network retains its own risk decisioning.

That could reduce duplicated onboarding and help merchants distinguish trusted automation from malicious bot activity.

Strategically, however, the agreement does something more interesting: it commoditises the trust layer while the networks continue competing to control where machine-originated value accrues.

Relevant tags: Visa, Mastercard, Ant International, Monetary Authority of Singapore, Quantum Payments

Visual concept: Hero image showing three neon payment-network pathways converging through a glowing trust gateway.

LinkedIn post 2 : 3:23pm AEST

Angle: Merchant operator : practical preparation for agentic payments and AI-driven acceptance.

Agentic payments will not be ready for merchants simply because an AI agent can complete checkout.

The payment stack must also be able to:

  • Verify the agent and its operator

  • Validate customer intent and spending permissions

  • Accept tokenised credentials

  • Avoid blocking legitimate automation as bot fraud

  • Optimise routing and payment success rate

  • Feed transactions into unified reconciliation

  • Support online, in-person and omnichannel commerce

Visa, Mastercard and Ant International’s new Know-Your-Agent collaboration is a signal that these capabilities are moving from experimentation towards infrastructure.

For merchants, the priority is not picking one protocol. It is building a flexible payment foundation that can support trusted agents across multiple networks and channels.

Relevant tags: Visa, Mastercard, Ant International, Quantum Payments, e-commerce leaders, retail leaders, payments professionals

Visual concept: Merchant-focused neon network showing an AI agent token travelling through fraud detection, routing and reconciliation gates.

 
 
bottom of page