Artificial intelligence (AI) is no longer an abstract idea for the banking industry. It is transforming the way institutions manage risk, design products and engage customers. The discussion has moved beyond curiosity about what AI can do to the challenge of how to implement it safely, responsibly and at scale. During the Singapore AI Sunset Cruise hosted by TAB Global in collaboration with Huawei, senior executives from banks, regulators and technology firms across Asia Pacific met for an evening of candid dialogue on AI as an enabler of future banking. The dialogue brought together leaders from institutions across the world including AEON Bank, Alliance Bank, Bank Inter, Bank of Singapore, Central Bank of Kenya, Citibank Singapore, GXS Bank, OCBC, Sathapana Bank, Techcombank and Vietinbank, all reflecting on how AI is reshaping the foundation of financial services. The conversation revealed a shared recognition that banks are entering a new phase of transformation—moving from automation for efficiency to intelligent collaboration between humans and machines. Yet, the speed and direction of this shift differ widely across institutions. What follows are the key themes and insights from the discussion, as participants explored how AI is transforming financial services, what organisational change is required and what divides the banks leading this shift from those lagging behind. AI becomes the next platform for enterprise reinvention Emmanuel Daniel, founder and chairman of TAB Global, opened the session by challenging banks to look beyond superficial use cases and reimagine their structures. “It’s interesting to see how traditional banks are thinking about AI,” he said. “Some have never used agentic AI before and are surprised by how autonomous these systems have become. You can even build your own anti-money-laundering models—so the question now is, where does it sit and who’s in charge?” Daniel framed AI as an outside-in transformation, where value increasingly flows from customer data outside the bank rather than internal databases. “The end user is defining their own financial relationships,” he remarked. “If you’re still building apps, you are already outdated. Customers will soon define their own intelligent agents that talk directly to whoever they want.” His comments set the stage for a conversation about AI as an infrastructure, not a feature—a shift from experimentation to architectural change. Participants discussed how enterprise systems must evolve to allow employees and customers alike to interact with AI agents that execute complex financial tasks autonomously. The consensus was that banks must first dismantle outdated hierarchies that slow innovation. Several speakers noted that data silos, legacy systems and traditional IT structures remain barriers to enterprise-wide adoption. The first step, they agreed, is cultural: enabling experimentation without fear of disruption. Daniel summarised the sentiment succinctly: “The customer is already using AI to manage their life. Banks have to catch up.” Huawei calls for long-term vision over short-term hype Representing Jason Cao, CEO, Huawei Digital Finance, urged banks to adopt a long-term view of AI’s value. “We should not underestimate the long-term value AI can bring,” he said, “but we should also not overestimate its short-term benefits.” He cautioned against treating AI as a quick efficiency solution or a marketing label. “If we think of it as another tool, it will tumble. The real power of AI is structural—it rebuilds end-to-end business processes, including human–machine collaboration models.” Jason predicted that the next decade will divide the industry. “In the future, there will be only two types of banks—AI banks and others,” he said. “Those that embed intelligence into every part of their organisation will outperform those that keep AI at the edges.” He drew a parallel between the rise of AI assistants and the mobile revolution. “The last ten years were about the mobile app—a unified portal for all services. The next ten will be about the super-assistant. It will know what you want, see what you see and act for you.” The message resonated among participants: AI should not be viewed as a discrete initiative but as a new operating layer underpinning every business process. The banks that succeed will be those that build for adaptability, not merely for automation. Central banks and regulators see AI as an intelligent assistant For June Lomaria, digital payment policy analyst from the Central Bank of Kenya, AI represents a pragmatic balance between potential and responsibility. “AI is an assistant that can do anything for you,” she said. “It can process a lot of things on your behalf. As a bank, we’re not yet there, but we’re having discussions, and once we do, we’ll let the world know.” Her description of AI as an intelligent assistant echoed across the table, reflecting how central banks and regulators view AI not as a competitor to human judgement but as an enabler of augmented supervision—from detecting anomalies in transaction data to improving risk assessments. Participants noted that public institutions face their own constraints. For regulators, the issue is not whether to use AI but how to govern its adoption responsibly. The need for explainability and transparency remains a central concern. Lomaria concluded that AI’s evolution in finance must be guided by human oversight. “We want the technology to assist, not replace,” she said, reinforcing the idea that AI’s greatest potential lies in collaboration rather than substitution. Digital challengers seek to AI-proof their institutions Representing one of Singapore’s new digital banks, Vincent Mok, group chief risk officer, from GXS Bank described how starting without legacy systems offers a unique advantage. “Our focus will be to AI-proof banking,” he said. “We start from zero, so we can design processes that are resilient to change rather than constrained by it.” Daniel noted that this clean-slate advantage allows digital banks to leapfrog traditional models. “The funny thing with your bank,” he told Mok, “is that you don’t have the legacy that others do—you can start clean.” Mok explained that building an AI-proof institution means designing systems flexible enough to accommodate continuous learning, data flows and evolving compliance requirements. Instead of adapting AI to old workflows, GXS aims to create dynamic, modular frameworks where AI becomes a natural part of decision-making. The discussion highlighted a growing divide between banks that retrofit AI into existing architectures and those that are building from scratch. Digital-first institutions can deploy AI across credit, risk and customer engagement simultaneously, rather than through pilot projects. This capability, participants agreed, positions challengers as potential leaders of the next generation of AI-native banks—institutions that are adaptive by design rather than by necessity. Enterprise transformation and the democratisation of AI Celine Le Cotonnec, chief data and innovation officer, Bank of Singapore, likened the rise of AI to the leap from pen and paper to computers—an “enterprise-wide revolution” that would require a complete rethink of how banks operate. “It needs to be democratised so that everyone can use it,” she said. “We need to rethink the way we’re working. Potentially, a lot of the processes or platforms we’ve invested in will have to be scrapped.” To illustrate her point, she shared a recent example from a workshop with Microsoft and a research provider. Traditionally, she explained, her team would spend weeks structuring data, building databases and creating interfaces. “Now, all of this can be done by an agent on the Microsoft server,” she said. “I can connect one agent to another and get the output I need. I don’t need my legacy systems or IT development cycle anymore. Things can be done much faster by the user themselves.” Le Cotonnec argued that the challenge is not technology but empowerment—how to enable everyone in the organisation to build, train and use AI agents responsibly. “It’s about changing roles,” she said. “How do you empower everyone in the organisation to build their own agent? And how do you reposition IT from being the builder to being the governance body?” Her comments reframed the debate on digital transformation. In her view, IT must evolve from gatekeeper to guardian, providing the governance, security and guardrails that keep enterprise-wide AI innovation aligned with regulatory and ethical standards. This shift, she noted, requires redefining accountability and control in a decentralised innovation environment. Her remarks resonated across the table. Several participants agreed that the real test of AI maturity is not adoption but governance—how to democratise innovation while ensuring it remains responsible, explainable and secure. From augmented to authentic intelligence in enterprise transformation Santosh Mahendiran, chief data and analytics officer at Techcombank, described the journey from augmented intelligence to authentic intelligence. “The transition from augmented to authentic is where identification and decision-making converge,” he explained. Techcombank has already deployed AI agents and voice bots in underwriting and customer service, delivering measurable results. “We measure around $45 million in impact per year,” Mahendiran said. “But the real challenge is scale.” He argued that banks need to re-engineer their business processes rather than simply automate them. “We have shortened the loan cycle and automated low-level approvals, but do we even need an underwriting department that passes files from front line to credit to risk? That’s what we haven’t solved yet.” The challenge, he added, is cultural as much as technical. “We have started to democratise AI functions across the organisation, but for now it’s still mainly a productivity tool. We want to be much more before we can declare success.” His remarks underscored a critical insight: adopting AI requires not just technology but an organisational redesign that empowers employees to innovate within an AI-augmented framework. Emerging markets demonstrate agility without legacy Nak Pechkorsa, chief technology and information officer of Cambodia’s Sathapana Bank provided an emerging-market perspective. “I want AI to be the intelligent assistant that helps us make decisions faster and more accurately,” he said. He described how banks in Cambodia are using external data—such as behavioural and social data—to enhance know-your-customer (KYC) and credit evaluation processes. Without legacy systems or deeply entrenched data hierarchies, these institutions can experiment with AI more freely. Pechkorsa emphasised that this lack of legacy is not a disadvantage but a catalyst for innovation. “We can build without the burden of what came before,” he noted, suggesting that developing economies may become testbeds for new AI-driven models. Daniel agreed: “A lot of the most exciting innovations might start from countries without legacy and work their way back.” The discussion reinforced that emerging markets can serve as innovation laboratories, pioneering models that established institutions later adopt at scale. The path to AI-driven transformation The Singapore AI Sunset Cruise revealed how the financial industry is moving from theory to practice in its approach to AI. While some banks still experiment at the edges, others are beginning to restructure entire organisations around AI capabilities. Participants agreed that AI’s real promise lies in collaboration—between humans and machines, banks and technology partners, and regulators and innovators. It is not merely an efficiency tool but a foundation for redefining the economics of banking. The consensus among speakers like Huawei’s Cao, Bank of Singapore’s Le Cotonnec and Techcombank’s Mahendiran was that the industry stands at an inflection point. Banks must move from incremental improvements to enterprise reinvention, embedding AI into governance, risk, product design and culture. As Daniel concluded, “The customer is already defining their financial relationships. The real race is not about who builds the best AI model, but who builds the most adaptive organisation around it.” In the end, as Jason put it, “there will only be two kinds of banks—AI banks and others.” The choice, and the transformation, has already begun.
