Southeast Asia’s largest bank, DBS Bank, is betting that your next card transaction might not come directly from you—but from an AI agent acting on your behalf.
In a joint pilot with Visa, DBS is testing agent-initiated payments under Visa’s new Visa Intelligent Commerce (VIC) framework. The initiative aims to move “agentic commerce” from whiteboard concept to real-world execution, starting in Singapore and potentially scaling across Asia Pacific.
If it works, it could reshape how consumers interact with banks, cards, and digital payments—especially in markets already comfortable letting AI handle daily tasks.
From Chatbots to Checkout
The timing isn’t accidental.
A Visa-commissioned study shows 77% of Singapore residents already use generative AI tools like chatbots in their daily lives. Even more telling: eight in 10 rely on AI assistance when shopping online. AI isn’t just a novelty feature in Singapore—it’s fast becoming a digital co-pilot.
DBS and Visa are now extending that behavior into payments.
Through a series of live food and beverage transactions, AI-powered agents successfully completed purchases using DBS/POSB credit and debit cards. The payments ran through secure, issuer-controlled flows—meaning DBS maintained oversight of authentication, authorization, and transaction controls.
In plain English: the AI agent can click “buy,” but the bank still controls the rails.
That distinction matters. Agentic commerce only works at scale if banks trust the automation as much as customers do.
What’s Actually New Here?
AI-driven shopping assistants aren’t new. Amazon, Google, and fintech startups have experimented with automated purchasing for years. What’s different here is issuer-level integration.
DBS is the first card issuer in Asia Pacific to pilot real-world agent-initiated commerce transactions tied directly to its card infrastructure under Visa’s new framework. Instead of layering AI on top of payments, the model embeds AI-readiness into the credential itself.
Visa Intelligent Commerce bundles APIs, authentication protocols, and a partner ecosystem designed specifically for AI agents. It introduces concepts like:
- AI-ready credentials
- Advanced authentication controls
- Intent-driven transaction validation
- Consent-based authorization frameworks
Visa’s Trusted Agent Protocol underpins the system, aiming to ensure that transactions are traceable, transparent, and consumer-approved—even when initiated by software rather than a human tapping a screen.
In other words, this isn’t just faster checkout. It’s a rethink of who—or what—can legally and securely initiate a payment.
Why It Matters for Banks and Merchants
For banks, agentic commerce presents both opportunity and risk.
On one hand, AI agents could dramatically reduce friction in routine purchases—subscriptions, travel bookings, grocery reorders, food delivery. On the other, handing purchasing power to software raises obvious questions about fraud, accountability, and customer control.
DBS says it is validating AI-ready credentials and implementing advanced authentication layers to ensure issuer-level oversight. The goal is to balance convenience with resilience—an increasingly important theme as generative AI adoption accelerates.
For merchants, this model could streamline conversion rates. If AI agents can compare prices, evaluate preferences, and execute payments instantly, cart abandonment may shrink. But it also changes the competitive landscape: optimization shifts from appealing to human shoppers to being machine-readable and AI-preferred.
Search engine optimization may soon have a cousin: agent optimization.
Asia Pacific as the Testing Ground
Singapore is a logical launchpad. It’s one of the world’s most digitally mature banking markets, with high card penetration and rapid AI adoption across demographics.
But the implications stretch across Asia Pacific, where super apps, embedded finance, and mobile-first commerce already dominate. If AI agents can safely transact across travel, retail, and services, banks in the region may leapfrog Western markets in mainstream adoption.
The move also positions Visa to reinforce its role as the underlying payments network in an AI-driven commerce era. While fintech challengers experiment with alternative rails and tokenized systems, Visa is effectively upgrading its existing infrastructure to accommodate AI-native use cases.
It’s a defensive and offensive play at the same time.
The Competitive Landscape
Global payments players are racing toward similar territory. Mastercard has explored AI-driven fraud detection and personalization tools. Big Tech firms are embedding AI shopping assistants directly into search and e-commerce platforms. Meanwhile, open banking and real-time payment networks continue to evolve globally.
The difference with DBS and Visa’s pilot is its issuer-led framing. Rather than allowing tech platforms to dominate AI commerce flows, this model keeps banks directly in the loop.
That may prove critical in regions where regulatory oversight and consumer protection are tightly enforced.
What Comes Next?
The current pilot focused on food and beverage transactions, but the roadmap includes broader use cases such as:
- Online retail purchases
- Travel bookings
- Subscription services
- Routine bill payments
As the collaboration expands, the real test will be scale. Can AI agents execute high-frequency transactions without increasing fraud exposure? Will regulators embrace consent-based AI payment frameworks? And perhaps most importantly: will consumers trust bots to spend their money?
Given Singapore’s adoption trends, early indicators suggest they might.
For now, DBS and Visa are laying the groundwork—embedding trust, authentication, and accountability into agent-initiated flows before the model goes mainstream.
If agentic commerce fulfills its promise, the future of payments won’t just be contactless. It will be autonomous.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
