Vietnam’s AI moment is here

Vietnam processes up to 100 million digital banking transactions every day. Over 90% of banking services already happen through digital channels. More than half of the country’s banks are increasing IT budgets to keep pace with rising customer expectations.

At a recent leadership roundtable hosted by the Vietnam Banks Association (VNBA) and the State Bank of Vietnam, executives aligned on one clear point: the next wave of growth will come from AI that moves beyond pilots and into execution.

The fundamentals are there. A young, digitally native population. Mobile adoption rates among the highest in ASEAN. A national mandate pushing digitalization forward. A CIO Vietnam report shows that 60% of IT leaders see generative AI as the top driver of 2026 budget growth, yet fewer than 8% of organizations have integrated AI at an enterprise level. The gap between ambition and execution is real.

The question is no longer whether AI matters. It is how Vietnam’s banks can turn scattered projects into a unified engine for growth.

From digital access to the AI “factory”

Over the last decade, banks in Vietnam have done the hard work of digitizing access. eKYC, mobile apps, chatbots, and online forms are now standard. These were the right first moves. But they are no longer enough.

AI represents a different kind of shift. It is not only a tool for automation. It is a new kind of “factory” inside the bank - an engine that can generate and test thousands of ideas for new products, pricing, risk models, and journeys in parallel.

In simple terms, that AI factory should be able to:

1. Sense need: Read real-time data and spot emerging patterns in a customer’s financial life.
2. Generate options: Propose next best actions, products, or interventions that fit that context.
3. Deliver value: Turn those ideas into real offers, journeys, or actions at the right moment.

In theory, this is how banks move from reactive servicing to proactive, empathic banking.

In practice, most of these AI factories are stuck.

They are starved for fuel because data is scattered across legacy cores, card systems, wealth platforms, and marketing tools. And they fail the last-mile test of trust, because AI recommendations often appear as “black box” decisions without clear context or explanation.

The missing nervous system: data that can keep up with AI

This is where architecture becomes a board-level topic.

Vietnam’s banks are investing heavily in data platforms, lakes, and governance programs. But the roundtable discussions made it clear: technology alone will not fix the data problem if the operating model stays the same.

To unlock AI, banks need a data “nervous system” that:

• Lets business domains like lending, cards, and SME banking own and serve their data.
• Keeps information accurate, clean, and near real time.
• Makes data discoverable and usable across the bank, not trapped in one team.

Global leaders are moving toward this kind of distributed model, often referred to as a data mesh. The label is less important than the principle: AI needs high-quality, domain-owned data products, not another monolithic warehouse that updates once a day.

Without this nervous system, agentic AI will keep running on stale, incomplete views of the customer. With it, AI can finally see and act on a true, 360-degree picture of a customer or business in the moment.

CLV as the north star metric

If the data layer is the nervous system and the AI factory is the brain, the next question is simple: what is this brain optimizing for?

Historically, banks have organized around products. Each line of business fights for its own P&L. Success is measured product by product: more cards, more loans, more accounts.

In an AI-powered world, that model breaks down. Recommendations, journeys, and actions cannot be optimized in isolation. They need a single, clear objective.

Customer Lifetime Value (CLV) provides that objective. It shifts the bank from:

• Product centric to plan centric
• Transaction optimization to lifetime value orchestration
• “How do we sell this product?” to “What is the next best action to improve this customer’s financial health and deepen our relationship?”

Sometimes the right action is to sell a product. Sometimes it is to warn a customer about emerging risk, recommend a lower-cost option, or even highlight a better offer the bank cannot match.

This is counter-intuitive in the short term. But in markets where switching costs are low and information is abundant, long-term trust is the only sustainable way to protect CLV. AI at scale makes that strategy operational.

Augmented, not automated: keeping humans in the loop

There is another critical design choice at the heart of AI in banking: automation versus augmentation.

• Automation replaces humans with “black box” systems. It is faster, but opaque.
• Augmentation gives humans a “glass box” copilot. It explains, suggests, and leaves the final decision to the banker or the customer.

At the roundtable, Vietnamese banks were clear. Given the regulatory, cultural, and trust context, AI must be framed as augmentation first.

For bank employees, this looks like:

• Relationship managers seeing proactive insights on upcoming liquidity events.
• Branch staff guided through complex servicing steps with AI-generated checklists.
• Credit teams reviewing suggested decisions that come with explanations and simulated outcomes.

For customers, it looks like:

• Helpful prompts that clearly explain why a recommendation is being made.
• Clear links between actions and stated goals, like saving for a home or smoothing cash flow.
• Control over what the AI can do on their behalf and where human approval is needed.
  
  

This “glass box” approach is how banks in Vietnam can solve the last-mile trust problem. It is also how they can align with regulators who are rightly cautious about opaque decisioning.

Three questions shaping Vietnam’s AI agenda

The roundtable surfaced three practical questions that now sit at the center of Vietnam’s AI journey.

1. How do banks move from chatbots to agentic AI?

Chatbots handle FAQs and simple tasks. Agentic AI executes full workflows.

In Vietnam, early agentic AI use cases include:

• Digital coworkers for contact center and branch staff, providing context, guidance, and suggested next steps.
• AI-driven engagement engines that orchestrate hyper-personalized campaigns and journeys.
• Autonomous workflows for credit approvals, KYC, and transaction handling, with humans reviewing edge cases.

The shift is from “answering questions” to “solving problems” across channels.

2. How do banks govern AI with speed and discipline?

AI can only scale at the speed of governance.

Banks at the roundtable highlighted three pillars of responsible scaling:

• Governance councils that approve, monitor, and refresh AI models.
• Human oversight as a non-negotiable principle for critical decisions.
• Phased deployment, starting with small, high-impact use cases where value and risk can be measured clearly.

Vietnam’s central bank is already working on a national AI governance framework. Banks that invest now in internal structures, documentation, and monitoring will be ready to move when that framework solidifies.

3. How can AI improve journeys without eroding trust?

AI should make experiences feel smoother and more human, not more robotic.

That requires aligning four elements:

1. Data architecture that gives AI a clean, unified view of the customer.
2. Journey design that embeds AI at key decision points instead of bolting it on.
3. Transparency about why an action is recommended or taken.
4. Choice and control for customers and staff.

Done poorly, AI feels like a “creepy” recommendation engine. Done well, it feels like a trusted adviser that knows when to speak up and when to stay in the background.

Quick fact: Backbase's AI-native platform unifies systems and orchestrates journeys at scale, with agentic AI that acts autonomously to drive banking outcomes.

The takeaway for banks

1. Treat AI, data, and digital as one strategy
AI pilots, data programs, and digital channels are not separate projects. They are components of a single machine that must be designed together.

2. Start with high-impact, controlled use cases
Pick journeys where value is clear, risk is manageable, and data is available. Prove impact. Then scale.

3. Build a data nervous system, not another data silo
Empower domains to own their data products. Focus on quality, timeliness, and usability.

4. Use CLV as the north star
Optimize for the lifetime health of the relationship, not one-off product pushes. AI makes this measurable and actionable.

5. Anchor AI in augmentation and transparency
Keep humans in the loop. Design “glass box” experiences that earn trust with every interaction.

6. Modernize progressively
Progressive modernization - replacing fragmented systems step by step while keeping the bank running - is the most realistic path to an AI-ready architecture.

Vietnam's next leap is within reach

Vietnam’s banks have already proven they can move fast on digital. The next step is harder and more strategic: turning AI, data, and digital experience into a single, unified system that works at the scale of the institution, not just in pockets.

With strong governance, a modern data foundation, and a clear focus on customer lifetime value, Vietnam has the potential to leapfrog regional peers and help define what AI-powered banking looks like in practice.

The technology is ready. The ideas are mature. The question now is execution: which banks will assemble the machine first, and use it to build the next decade of growth?