How Singapore's 3 Largest Banks Are Using AI in 2026

DBS, OCBC and UOB each report more than 300 AI use cases in production. The clearest results come from credit memos, wealth onboarding and risk surveillance.

You have probably sat through a presentation in which a vendor or an internal team announces that the organisation now runs "hundreds of AI use cases". The number sounds impressive until you ask what those use cases do, who uses them and what changed as a result. Singapore's three largest homegrown banks – DBS, OCBC and UOB – have each crossed the 300-use-case threshold, and each has disclosed a different mix of results this year. The clearest picture emerges not from the totals but from what each bank has put into production and where the impact is already measurable.

DBS puts numbers on its AI push

DBS reports more than 430 AI use cases across over 2,000 AI models. Its 2025 annual report estimates that its data analytics and AI and machine-learning initiatives delivered about S$1 billion in economic value in 2025. The newer effort is agentic AI, software that carries out a chain of tasks instead of answering a single question. In August, DBS gave around 1,500 relationship and credit risk managers worldwide a tool that works through more than 70 such tasks to draft a corporate credit memo, the document that makes the case for lending to a company. It grew out of a pilot with 150 staff, and bankers and credit risk managers still make the assessment and approval decisions.

Late July saw DBS say its DBS Joy and digibot assistants, which now carry generative and agentic AI features, serve more than 10 million customers across Singapore, Hong Kong and Taiwan. Some of those features let users finish a banking task without leaving the conversation. Ahead of its March AGM, DBS told shareholders in written answers that it is redesigning processes, expanding collaboration between employees and AI, and reskilling staff as part of changes to its operating model.

OCBC's most visible AI launches this year are in wealth

OCBC's 2024 annual report counts more than 300 AI use cases, though its most prominent 2026 launches have centred on wealth management. OCBC WoW, launched on 1 July, is built around two AI avatars, Wendy and Wayne, which clients can talk or type to about their portfolios using answers drawn from OCBC's investment research. The beta is open to staff and invited Premier Private Client customers with at least S$1.

Bank of Singapore followed later that month with HELIOS, an agentic AI platform that handles much of the information gathering and due diligence before a relationship manager takes on a new client. OCBC is targeting a median of 15 business days to open a private banking account, compared with an industry median of about six weeks, provided clients submit their documents promptly. HELIOS builds on Bank of Singapore's Source of Wealth Assistant, launched in October 2025 to draft reports tracing a client's source of wealth. Once documents are uploaded, the tool cuts preparation time from 10 days to one hour, and relationship managers review the draft before it is finalised.

The bank also uses AI to train its wealth advisors. Participants in a six-month GenAI programme for 900 Singapore advisors booked twice as many weekly client appointments as untrained peers, Computer Weekly reported. OCBC also reported a 50% revenue uplift against their previous three months, although both figures came from early in the programme. Beyond wealth management, the bank screens payments for scams in under 15 milliseconds, while an Asian Banker award write-up said personalised engagement drove 40% of sales.

UOB's clearest published AI results come from risk management

More than 300 AI and analytics use cases are now in production at UOB, one of its executives said, as QA Financial reported. An August investor presentation shows where they are heading, from wealth advisory in private banking to contact centres and software development. It also describes a programme called AURA that is building agentic AI for anti-money laundering and know-your-customer checks, credit, collections and relationship manager support. The same deck says more than 35,000 staff have been enabled with Microsoft Copilot, and UOB says it has built more than 30 domain-specific chatbots on its Build-Your-Own-Bot platform.

Project Magnet, a surveillance tool in UOB Private Bank that pairs machine learning with rules-based checks, has raised true hits – the alerts that turn out to be genuine risks – by 148% and improved exception rates by 130%, UOB reports. The system draws on more than 3,100 data points to spot early signs that a borrower is struggling to repay, The Digital Banker wrote. Through the TMRW app, the bank served close to 180 million personalised insight cards to more than 2.5 million customers in 2023, per a Cloudera case study, and it runs a model that predicts wait times at its branches.

What this means for L&D and risk leaders

On 7 October, MAS issued its Guidelines on AI Risk Management for all regulated financial institutions, which take effect on 7 October 2027, with some requirements phased in until October 2028. MAS also plans to consult the industry in 2027 on whether agentic AI needs further guidance, OpenGov Asia reported. The guidelines follow a busy year for the regulator, which finished the second phase of Project MindForge in March and published an AI Risk Management Toolkit built with 24 financial institutions and other industry partners. The Monetary Authority of Singapore and industry partners introduced the SAFR framework for agentic finance in July, and an AI-focused cyber and technology risk task force followed with the Association of Banks in Singapore later that month. Back in August, MAS announced S$220 million over three years under FSTI 4.0, which includes support for wider AI adoption in financial services.

Speaking at a DBS event in May, Deputy Prime Minister Gan Kim Yong said slowing AI adoption would weaken Singapore's competitiveness, while urging companies to think about what new roles could be created and how existing workers could be retrained. From October 2027 onwards, the banks' AI deployments will fall under those supervisory expectations. The question for learning and development leaders in financial services is whether their organisations are building the capability to meet those expectations at the same pace they are deploying the technology. The three banks have shown that the use cases are already in production. The next test is whether the people who work alongside them are ready for the supervision that follows.

Sources:

• Fintech News Singapore, 14 October 2026