Agentic Commerce Goes Live: DBS Bank and Visa Enable AI A...

Agentic Commerce Goes Live: DBS Bank and Visa Enable AI Agents to Make Purchases

AI agents are no longer just recommending products—they're now buying them. Singapore's DBS Bank has become the first issuer in Asia Pacific to pilot Visa Intelligent Commerce, successfully completing real-world food and beverage transactions where AI agents executed credit card purchases on behalf of customers.

This isn't a proof-of-concept in a lab. These are actual transactions through existing payment infrastructure, marking a significant step toward mainstream agentic commerce.

What Is Agentic Commerce?

Agentic commerce refers to AI-powered software agents that can autonomously search for products, compare options, negotiate terms, and complete transactions on behalf of consumers—within predefined parameters set by the user.

Unlike traditional e-commerce where humans click through checkout, agentic systems handle the entire transaction flow:

1. User defines preferences (budget, brands, delivery requirements)

2. AI agent searches across merchants

3. Agent selects optimal option based on criteria

4. Transaction executes using pre-approved payment credentials

5. User receives confirmation

The consumer sets the rules. The AI handles execution.

DBS Bank and Visa: What Happened

In late February 2026, DBS Bank and Visa announced their collaboration under Visa Intelligent Commerce (VIC), completing a series of real-world food and beverage transactions using DBS/POSB credit and debit cards.

Ananya Sen, Group Head of Regional Consumer Products at DBS Bank:

> "AI agents are unlocking a new phase in digital payments, where routine transactions can be completed efficiently and reliably, helping customers save time and simplify everyday tasks. Our collaboration with Visa shows how agent-led payments can be deployed securely and safely at scale."

The pilot validates three critical components for agentic commerce:

•AI-ready credentials: Secure, tokenized card details that trusted AI agents can use to make purchases safely
•Advanced authentication: Multi-layer verification ensuring transaction legitimacy
•Intent-driven transaction controls: Safeguards ensuring purchases align with user-defined parameters

Why Singapore? The Adoption Numbers

A Visa-commissioned study reveals why Singapore is the launchpad:

•77% of Singapore residents already use generative AI tools like chatbots daily
•8 in 10 Singapore consumers rely on AI assistance when shopping online
•Singapore ranks among top adopters of AI shopping tools globally

The infrastructure and consumer readiness align. Singaporeans have integrated AI into their digital habits—extending that to payments is a natural progression.

The Bigger Picture: Who Else Is Moving

DBS and Visa aren't alone. The agentic commerce race is accelerating across major players:

Mastercard

Completed fully authenticated agentic transactions in January 2026 with:

•Commonwealth Bank of Australia (cinema ticket purchases)
•Westpac (hotel reservations)
•Santander (cross-border tests announced March 2026)

Mastercard's "Agent Pay" infrastructure enables AI agents to pay securely on behalf of users and businesses.

Retail Giants

•Walmart partnered with Google to enable purchases through the Gemini app
•Target announced collaboration with OpenAI for agent-mediated checkout

Payment Infrastructure

•Stripe launched an API for agentic payments in September 2025
•FIS developed "Smart Basket" technology to optimize agent-driven transactions in real time

Mladen Vladic, Head of Product for Payment Networks at FIS:

> "We are seeing something different with agentic commerce where you have the largest brands in the marketplace announcing first... This is a transformational inflection point in the industry, not only in this country, but globally."

The Market Opportunity

The numbers are staggering:

•$1 trillion in US retail revenue projected from agent-mediated purchasing by 2030 (just 4 years away)
•81% of US consumers expect to use agentic AI tools to shop, according to Boston Consulting Group
•Visa predicts millions of consumers will use AI agents for purchases by the 2026 holiday season

This isn't speculative—it's infrastructure being built now for adoption at scale.

How It Actually Works

The technical implementation maintains existing payment rails while changing who initiates:

1. Authorization: User pre-approves spending parameters (max amounts, merchant categories, frequency)

2. Tokenization: Card credentials are converted to AI-ready tokens, not raw card numbers

3. Agent Execution: AI agent identifies purchase opportunity matching criteria

4. Authentication: Transaction passes through standard Visa security checks

5. Completion: Purchase executes, user receives notification

The key innovation: the familiar plumbing of payments remains intact. Networks, issuers, acquirers, fraud controls—all standard. What changes is who pulls the lever.

Consumer Concerns: Security and Trust

Despite high AI adoption for shopping, consumers remain cautious about AI handling money directly.

A Visa survey of 14,700 consumers across 14 APAC markets found:

•32% reluctant to share personal or payment information with AI systems
•26% unsure if AI recommendations align with their best interests
•45% would be more open to agentic commerce with stronger payment security assurances

Interestingly, affluent households (monthly income $8,000+) showed the most caution at 39%, compared to 29% for lower-income groups. Digitally mature markets like Australia (38%), New Zealand (37%), and Singapore (34%) displayed above-average concern.

Emerging markets are most open: India and Vietnam lead at 42% willing to use AI for online purchases, while Singapore, Japan, and New Zealand hover around 14-16%.

What This Means for You

For Consumers

•Gradual shift from manual checkout to AI-assisted purchasing
•More control over spending parameters (set budgets, brand preferences)
•Reduced time on routine purchases (groceries, subscriptions, repeat orders)
•New considerations around data sharing and AI permissions

For Businesses

•Need to optimize for AI discovery, not just human browsing
•Agent-friendly checkout flows become competitive advantage
•Customer acquisition may shift from targeting humans to targeting AI decision criteria
•Dispute and fraud frameworks will need updating for agent-initiated transactions

For Developers

•New APIs for agentic payments (Stripe, Visa, Mastercard)
•Opportunity to build AI shopping agents across verticals
•Authentication and consent management become critical UX challenges

What's Next

DBS and Visa are expanding beyond food and beverage to explore:

•Online shopping
•Travel bookings
•Subscription management
•Cross-border payments

The pilot's success signals that major banks are no longer experimenting—they're preparing for production-scale deployment.

FAQs

Can AI agents spend without limits?

No. Users set parameters including maximum amounts, merchant categories, and frequency. Agents operate within these constraints.

What if an AI agent makes an unauthorized purchase?

Standard dispute and fraud protection frameworks apply. Tokenization and authentication layers provide transaction trails.

Which banks offer agentic commerce?

Currently in pilot phases. DBS Bank (Singapore), Commonwealth Bank (Australia), Westpac, and Santander have announced active programs.

Do I need a special card or app?

Initially, yes. Banks are rolling out agentic features through specific programs. Mainstream integration will follow pilot success.

Is my data safe with AI shopping agents?

Banks and payment networks are implementing consent-driven data sharing. However, consumers should review specific privacy policies before enabling agentic features.

Bottom Line

Agentic commerce has moved from concept to live transactions. DBS Bank and Visa's pilot proves that AI agents can execute real purchases through existing payment infrastructure with appropriate security controls.

The shift from AI recommending products to AI buying them represents a fundamental change in how commerce operates. With $1 trillion in projected US retail volume by 2030 and major banks, payment networks, and retailers racing to build infrastructure, the question isn't if agentic commerce will scale—it's how quickly.

For consumers and businesses alike, now is the time to understand the technology, set appropriate boundaries, and prepare for a future where your AI agent might handle your next purchase.

Sources:

Agentic Commerce Goes Live: DBS Bank and Visa Enable AI Agents to Make Purchases