The next stage of AI-powered commerce may not stop at product recommendations. AI agents are increasingly being designed to search, compare and eventually transact on a consumer’s behalf, putting banks and payment networks under pressure to make autonomous purchasing both useful and trustworthy. DBS Bank (Hong Kong) is responding on two fronts: continuing its work with Visa on secure agentic payments while partnering with tech-enablement company Preface to give customers hands-on AI education and tools.
For years, banks have treated artificial intelligence largely as an internal technology—something used for fraud detection, customer service, personalization or operational efficiency. The emergence of agentic commerce changes the equation.
If an AI agent can search for a product, compare prices and make a purchase for a customer, the bank and card network become part of an interaction that may no longer involve a person clicking a conventional checkout button.
DBS Bank (Hong Kong) is preparing for that shift by combining payments experimentation with customer AI education. The bank has announced a partnership with Preface, a Hong Kong-based technology-enablement company, to provide hands-on AI workshops and tools while continuing its collaboration with Visa around AI-enabled payments.
The two initiatives address different sides of the same transition.
DBS’s work with Visa is focused on the infrastructure required when an AI agent acts on a customer’s behalf. The bank was the first issuer in Asia Pacific to pilot Visa Intelligent Commerce, with real-world transactions demonstrating how AI agents can make everyday purchases through issuer-controlled authentication and payment flows.
The Preface partnership focuses on the consumer.
Starting in September 2026, eligible DBS credit-card customers subscribing to designated AI tools will be able to access an AI Experience Workshop through the DBS Credit Cards Subscription Bonus Club. DBS cardholders can also receive discounts of up to 15% on selected Preface AI courses.
The workshops are designed around practical applications rather than abstract AI theory. Topics demonstrated at the launch include prompt engineering, AI assistant creation, computer-using agents and agent orchestrators.
That matters because the adoption challenge is changing. Consumers no longer need only to understand what generative AI can produce. They increasingly need to understand what happens when AI can act.
An AI assistant that drafts an email presents relatively limited financial risk. An AI agent that can access credentials, select a merchant and initiate a payment raises much more consequential questions: What exactly did the consumer authorize? How does the bank know an agent is acting within those permissions? What happens if the agent chooses the wrong product or price? And how can a customer reverse or dispute an action they did not personally initiate?
Visa’s research highlights the gap.
A survey of 14,764 consumers across 14 Asia Pacific markets found that 74% already use AI-powered tools to discover, track or research products. Yet 32% remain reluctant to share personal or payment information with AI systems, while 45% say stronger payment-security assurances would make them more open to AI-powered or agentic commerce.
In other words, AI has already gained considerable ground at the discovery end of commerce. Trust becomes harder to establish at the transaction end.
That is where DBS’s Visa relationship becomes strategically significant.
Visa’s Agentic Ready program provides issuers and ecosystem partners with a framework for testing how AI agents interact with payments, including credential enrollment, tokenization, authentication and transaction monitoring. Visa says more than 50 issuing partners across 10 markets are participating in the program.
The underlying architecture is important because agentic payments require a different security model from traditional card transactions. Instead of authenticating only the person entering card details, payment systems need to establish whether the AI agent is authorized, what the user’s intent was and whether the transaction falls within permitted parameters.
DBS and Visa have already demonstrated the concept through everyday food-and-beverage transactions using DBS/POSB cards. The partners say they are exploring additional use cases including online shopping and travel bookings.
The Preface partnership adds another layer: AI literacy.
DBS is effectively betting that preparing for agentic commerce requires consumers to understand both the opportunities and the boundaries of AI. The workshops can help customers learn how to construct better prompts, customize AI assistants and experiment with systems capable of interacting with digital environments.
That could become increasingly relevant as AI moves from conversational interfaces toward autonomous workflows.
The competitive landscape is evolving quickly. Visa and Mastercard are developing infrastructure for agent-enabled payments, while technology companies such as Google, Microsoft, Amazon and OpenAI are building increasingly capable AI agents. Banks, meanwhile, must decide how those agents will authenticate, access accounts and interact with existing payment rails.
The distinction between a banking app and an AI interface could eventually become less clear. A consumer might tell an agent to find the best flight within a budget, reserve it and pay using an authorized card. The bank’s role would shift from simply processing a transaction to enforcing the customer’s rules around that transaction.
That creates a new category of financial-services infrastructure: trusted agentic commerce.
It also means banks have an incentive to educate customers before autonomous transactions become mainstream. A technically secure system can still fail if users do not understand what they are authorizing or how an AI agent makes decisions.
Deloitte’s 2026 research reinforces the scale of the transition. It found that 29% of consumer businesses in Asia Pacific already report adopting agentic AI, a figure the firm expects to rise to 76% within two years. Deloitte also says the region could account for around two-thirds of the world’s new retail sales over the next five years.
For DBS, therefore, the Preface partnership is not simply an educational promotion attached to a card program. It is part of a broader attempt to prepare customers for an interface shift in commerce.
The important test will be whether consumers trust AI enough to let it cross the line from recommending a purchase to completing one.
DBS and Visa are working on the infrastructure to make that transaction secure. Preface is helping users understand the technology that will sit on the other side of the interface.
Together, those efforts point to a future in which the competitive advantage in digital payments may depend not only on faster transactions, but on how clearly banks can establish consent, control and trust when the customer is no longer the one pressing “Pay.”
Market Landscape
Agentic commerce is moving from experimentation toward early real-world deployment, particularly in Asia Pacific.
Visa’s research shows a significant adoption gap between AI-assisted shopping and AI-enabled purchasing: 74% of consumers surveyed already use AI for product discovery, while 45% say stronger payment-security assurances would increase their willingness to use agentic commerce.
Deloitte similarly expects agentic AI adoption among Asia Pacific consumer businesses to rise sharply, from 29% currently to 76% within two years.
The ecosystem is therefore forming around three layers:
- AI agents: Systems from technology companies that discover, compare and execute tasks.
- Commerce infrastructure: Merchants, marketplaces, banks and payment networks that must support agent-initiated transactions.
- Trust infrastructure: Tokenization, authentication, consent controls, transaction monitoring and mechanisms that keep users in control.
DBS is positioning itself across the second and third layers through its Visa collaboration, while the Preface partnership addresses consumer readiness.
For enterprise financial-services teams, the lesson is that agentic commerce is not simply an AI problem. It is simultaneously a payments, identity, security, governance and customer-experience problem.
Top Insights
- DBS Hong Kong is pairing Visa’s agentic payment infrastructure with Preface AI education, addressing both transaction security and consumer readiness for autonomous commerce.
- Visa research shows 74% of Asia Pacific consumers already use AI for product discovery, but payment security remains a major barrier to agentic adoption.
- DBS’s Visa Intelligent Commerce pilot demonstrates AI agents completing everyday payments through issuer-controlled authentication, moving agentic commerce beyond purely conceptual demonstrations.
- Preface workshops introduce prompt engineering, AI assistants, computer-using agents and orchestration, giving consumers practical skills as AI becomes more autonomous.
- Banks increasingly need to define consent, authentication and transaction controls for AI agents, creating a new competitive layer around trusted digital payments.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
