DBS unlocks S$1 billion in AI value in 2025

[SINGAPORE] DBS notched up a one-third increase in economic value from its artificial intelligence (AI) initiatives to hit a record S$1 billion in 2025, according to the bank’s chief data and transformation officer Nimish Panchmatia.

DBS has “achieved significant progress” in such initiatives, he told The Business Times.

The 2025 figure is a jump from the S$750 million in economic value delivered in 2024, and underscores a broader trend across the three local banks – DBS , OCBC and UOB – for AI to play a crucial role in Singapore’s financial sector.

The banks have now moved beyond generative AI models, such as OpenAI’s ChatGPT and Google’s Gemini, and are now looking at agentic AI as the next stage of development.

Agentic AI models are a major development in the space, being capable of independent reasoning, decision-making and executing multistep tasks with minimum human oversight.

The efficiency engine

Even as some question the extent of the gains from using AI, banks at least are reporting tangible improvement. For instance, DBS’ S$1 billion figure was derived by comparing outcomes between customers who were offered AI-driven solutions and a control group.

“Agentic AI presents significant opportunities for the bank,” said Panchmatia. “We are piloting use cases such as a generative AI and agentic AI-enabled coding assistant for data scientists.”

The implementation of coding assistants has reduced coding time by over 20 per cent.

The other two local banks have also noted similar efficiency gains, notably in the wealth management sector.

OCBC’s private banking arm, Bank of Singapore, deployed its own AI tool, The Source of Wealth Assistant, which automates the drafting of source-of-wealth reports for relationship managers.

“This deployment reduced preparation time from 10 days to just one hour,” Melvyn Low, group chief of strategy and transformation officer at OCBC, told BT.

The accuracy and consistency of reports has improved, while remaining aligned with regulatory requirements, he said.

“AI has enabled OCBC to catalyse and accelerate our technology overhaul by reducing the time needed to modernise legacy systems,” he added.

In one major project involving the upgrade of technology stacks, AI shortened the timeline by up to three years. Low noted that this allowed engineers to focus on higher-value testing and problem-solving.

DBS, OCBC and UOB have employed large language models – DBS-GPT, OCBC GPT and Microsoft Copilot Chat, respectively – to support employees with their functions.

Nitin Datta, partner of financial services technology consulting at EY, noted that in 2025, the banking sector “saw a decisive shift from experimentation to substantive AI integration across front, middle and back-office functions”.

AI is also used to boost front-facing customer work.

For instance, UOB piloted a bot to help retrieve information quickly and assist with customer queries at counters.

“This shortens counter turnaround time and enables us to serve more customers efficiently”, said Lee Zhu Kuang, head of innovation at UOB.

The bot has since been rolled out to all UOB branches in Singapore. Lee said the bank has also launched a predictive queue wait-time model that provides customers with real-time visibility of expected wait times.

2026: The age of agentic

The next wave of AI integration, said Datta, is expected to be focused on agentic AI.

This sentiment was echoed by OCBC’s Low. “We will adopt AI even more extensively in the years ahead,” he said, adding that agentic AI models will “play a larger role”.

Similarly, Panchmatia noted that there is an increased reliance on agents by customers, for tasks such as finding solutions and procuring products.

The potential economic benefits of AI remain attractive for the banks.

UOB expects its collaboration with Accenture – which began in 2025, to co-develop and deploy advanced AI solutions across domains such as anti-money laundering – to generate about S$900 million in economic value over the next five years.

OCBC anticipates that, by 2027, 75 per cent of customer service requests will be AI-assisted; it also expects a 20 per cent uplift in overall staff productivity.

DBS, meanwhile, will continue using AI to strengthen relationships with its customers.

Yet, the value proposition of agentic AI – which requires less human involvement and oversight – can pose a threat to the human banking employee.

It was announced in 2025 that DBS’ contract and temporary staff workforce would be reduced by around 4,000 over the next three years, as AI increasingly takes on roles carried out by humans.

UOB and OCBC have not disclosed staff numbers affected by AI, if any.

However, the local banks told BT that they remain committed to upskilling and retraining their local talents. In 2025, DBS said that it would also be retraining its permanent staff.

In October 2025, Minister for National Development Chee Hong Tat – who is also deputy chairman of the Monetary Authority of Singapore – said OCBC, UOB and DBS are retraining all 35,000 of their Singapore staff over the next one to two years to be more adept with AI skills.

“Upskilling remains a priority,” said UOB’s Lee, noting that more than 20,000 employees have been trained in foundational generative AI through a series of webinars and workshops.

DBS does not see AI literacy as a technical skill, but a “life skill” instead. Panchmatia said the bank will continue to educate all employees on how to use AI to stay “future-ready”.