trending
The race to define the next phase of digital commerce via agentic payments has moved decisively into the banking mainstream.
Agentic payments through bank pilots
Visa and Mastercard have enlisted leading Asia-Pacific institutions to trial agentic payment technologies — systems in which AI-powered agents execute transactions on behalf of consumers within secure, consent-driven frameworks.
In Singapore, DBS Bank has become the first bank in the region to pilot Visa Intelligent Commerce, a suite of APIs designed to allow artificial intelligence agents to initiate and complete card payments under issuer control.
The move reflects rapid shifts in consumer behaviour: more than three quarters of Singapore residents already use generative AI tools in daily life, while eight in ten reportedly rely on AI assistance when shopping online.
DBS and Visa have conducted live food and beverage transactions demonstrating how AI agents can securely process debit and credit card payments.
The flows remain fully authenticated and governed by the issuing bank, preserving customer visibility and consent.
The partners now intend to extend testing into more complex use cases, including e-commerce and travel bookings.
From Assisted Checkout to Autonomous Commerce
Across the Tasman, Mastercard has partnered with Westpac to complete what it describes as New Zealand’s first agentic transaction using its Agent Pay framework.
A Westpac-issued debit card was used to purchase cinema tickets, with every participant in the transaction chain — issuer, acquirer and merchant — able to identify that an AI agent conducted the payment.
Mastercard has also showcased similar capabilities in India in collaboration with Axis Bank and RBL Bank, executing authenticated agent-driven payments within a large language model environment.
Collectively, these pilots signal a shift beyond today’s “click-to-pay” paradigm.
Agentic payments envisage AI systems that can select payment methods, route transactions, reconcile data and even optimise checkout timing based on contextual cues.
Rather than merely facilitating payment, the agent becomes an active participant in commerce.
Security, Governance and Infrastructure Compatibility
For the card networks, the strategic priority is ensuring that agentic systems integrate seamlessly with existing payment rails, tokenisation frameworks and regulatory obligations.
The pilots emphasise issuer-controlled authentication, transparent audit trails and explicit cardholder consent — critical safeguards as automation deepens.
By embedding banks directly into early-stage trials, Visa and Mastercard are stress-testing interoperability with core banking systems, fraud monitoring engines and identity infrastructure.
The objective is not disruption for its own sake, but controlled evolution.
If successful, agentic payments could extend digital wallet functionality into a more proactive model — one that anticipates needs, reduces checkout friction and personalises payment choices in real time.
For consumers, this promises convenience. For banks and merchants, it offers potential gains in conversion rates, operational efficiency and data insight.
The broader implication is clear: AI is no longer confined to recommendation engines or customer service chatbots.
It is beginning to operate inside the payment flow itself.
In that transition lies both competitive advantage and systemic responsibility — a balance the industry is now carefully navigating.
Comments