Nimish Panchmatia, chief data & transforma-tion officer at DBS, elaborates on how the Singaporean bank effectively leverages AI, with a focus on purpose and empathy.
Global Finance: How does DBS’ AI-first strategy differ from other banks’ digital transformations?
Nimish Panchmatia: Our early commitment to AI has resulted in deep expertise, enabling sophisticated implementations across all banking operations. Recognized as the World’s Best AI Bank 2025 and Best Bank in the World in 2018, 2020, and 2022 by Global Finance, our success stems from a strong culture of experimentation, advanced digital capabilities, and a pervasive data-centric mindset. This iterative development, coupled with a willingness to embrace new technologies and bring our people along on the journey, ensures our competitive advantage and continued leadership in a rapidly evolving financial landscape.
GF: What is DBS’ approach to measuring the value and ROI for AI initiatives?
Panchmatia: DBS’ philosophy is “AI for everywhere,” emphasizing pervasive integration across all business units to maximize value. We are deliberate in terms of where we apply it; we have a very good custom control mechanism, which gives us the impact in a very transparent and auditable way.
We are one of the only banks in the world that publish in our annual report a number, which is the value we get out of AI. Last year it was $750 million, and in 2025, we expect to cross over $1 billion Singapore dollars.
You need to be careful to avoid the risk of creating a micro-industry by meticulously trying to quantify every penny of value you hope to gain. If a solution offers improvements, what is the tangible value of that “better”? If you cannot clearly define it, then that value does not exist.
GF: How does DBS balance AI-driven automation with the need for human interaction and oversight in customer-facing roles?
Panchmatia: Our vision is that of an AI-enabled bank with a heart: for simple, high-volume servicing tasks, we believe AI can operate effectively on its own. However, for more complex areas like advisory sales, conflict resolution, and complaints, human involvement is essential. In these processes, humans must always be “in the loop” and be the final decision-makers.
We must be very careful to tailor AI interactions without being intrusive, especially in lifestyle banking where personal data is involved. There’s a fine line between providing tailored suggestions and overstepping privacy boundaries.
While we are comfortable with AI handling reactive servicing independently, anything beyond that requires human oversight. The technology is evolving rapidly; a year ago, we were hesitant to deploy any AI directly to customers, but now it’s available to both our corporate and retail customers. We anticipate further advancements, but our fundamental principles, values, and ethics will always guide our adoption of new technology—with purpose and empathy—rather than solely focusing on financial gain.
GF: What are the biggest challenges in scaling AI?
Panchmatia: Challenge number one is navigating the pervasive hype and noise surrounding AI to identify truly applicable use cases. Many organizations, including our peers and non-competitors, get stuck on narrow applications.
Instead of piecemeal automation, AI should be applied with an end-to-end process perspective. For customer-facing applications, this means considering the entire customer journey. Internally, focusing on end-to-end processes rather than departmental or personal-level applications yields significantly higher value.
Banks struggle with the rapid pace of AI development. Their long investment cycles (months to years) are too slow for AI, which evolves in days or weeks, hindering timely ROI. They need agile architecture and fast decision-making for low-investment, high-return, short-cycle projects to adapt to new innovations quickly.
At the same time, it is important to bring your employees along on this journey – to equip them with the knowledge and skills to use AI effectively and efficiently. In DBS’ example, we have doubled down on our upskilling efforts to ensure employees remain relevant even as AI reshapes operating models. Since the beginning of 2025, we have identified more than 12,000 employees for upskilling or reskilling. To date, nearly all have commenced their respective learning roadmaps, including skills such as AI and data.
