Skip to content
News on crypto, fintech and digital infrastructure

News Payments

Mastercard Completes AI Agent Payment Pilot with Antom in Singapore

Mastercard completed its Agent Pay for Machines pilot with Antom in Singapore on November 19, 2026, testing business and consumer AI agent transactions.

Rating 4.6 80 votes
3D render of an AI payment card and data streams representing the Mastercard AP4M pilot in Singapore.

Mastercard announced on November 19, 2026, that it had completed a pilot of its Agent Pay for Machines framework with Antom, the merchant payments platform from Ant International, in Singapore. The tests included both business and consumer AI agent transactions using defined authorization and payment controls.

Pilot Scope and Transaction Mechanics

The AP4M pilot focused on business scenarios where an AI agent received credentialing, policy controls, and payment authorization to buy digital services. Spending stayed within preset limits and approved activity categories. Antom handled merchant-side acceptance, allowing the agent to complete purchases while merchants received clear identification of the represented business.

A separate live consumer transaction showed an AI agent buying coffee after the user had pre-captured their intent. This run used Mastercard Agent Pay with Verifiable Intent, which generated an auditable record of the authorized action. Antom again supported the acceptance side of the flow.

Each transaction under AP4M produces a verifiable record. The system relies on network tokenization and authentication to enforce controls at the point of payment. Merchants gain visibility into the business or consumer behind the agent without needing to manage agent credentials themselves.

Program Background and Technical Components

Mastercard introduced AP4M in June 2026. The framework lets businesses assign credentials to AI agents, set purchase categories and spending caps, and receive transaction-level audit trails. Verifiable Intent adds an open, standards-based layer that merchants can use to confirm authorizations independently of any specific protocol.

Isometric infographic of Mastercard AP4M technical framework showing credentials, policy limits, and settlement.
AP4M framework architecture: agent credentialing, spending limits, and verified execution.

The Singapore pilot builds on an earlier AP4M demonstration at Cloudflare Connect and on Mastercard’s prior agentic commerce tests in the city-state, including ride-booking integrations with DBS and UOB completed earlier in 2026. Those earlier runs established basic patterns for intent capture and settlement that the November tests extended to merchant acceptance.

Antom’s role centered on enabling merchants to recognize and accept payments from agents operating under defined permissions. The combination of Mastercard’s authorization layer and Antom’s acceptance infrastructure allowed both sides of the transaction to operate within existing payment rails while adding agent-specific safeguards.

Next Steps for Merchant Adoption

Mastercard and Ant International stated the collaboration will help merchants accept AI-agent purchases that stay inside approved permissions. The companies expect the tested controls to support new commerce models that rely on autonomous agents while maintaining identity, authorization, and spending oversight.

Joyce Bo, EVP Core Payments, Asia Pacific at Mastercard, noted that AI agents can already execute payments and that the remaining requirement is building trust through practical safeguards. The Singapore results are intended to move agentic commerce from demonstration toward broader operational use across the region.

Frequently asked questions

What was tested in the Mastercard AP4M pilot with Antom?

The pilot tested AI agents making purchases under business spending limits and approved categories, plus a consumer transaction for a coffee purchase using pre-captured intent. Both used credentialing and authorization controls on the Mastercard network with Antom handling merchant acceptance.

When did Mastercard complete the Singapore AI agent payment pilot?

Mastercard announced completion of the pilot on November 19, 2026. The tests took place in Singapore and covered both business and consumer use cases.

How does Verifiable Intent work in these transactions?

Verifiable Intent creates an auditable record of the consumer or business authorization for each agent action. Merchants can confirm the authorization independently of the underlying protocol before accepting payment.

What controls does AP4M apply to AI agent spending?

AP4M lets businesses set spending limits and approved purchase categories for each agent. Every transaction stays within those limits and produces a record that identifies the represented business to the merchant.

Comments

No comments yet. Be the first!

Leave a comment