How DBS Bank uses a human-AI synergy approach to enhance customer experiences and improve efficiencies

DBS Bank, one of Asia’s leading financial institutions, has emerged as a pioneer in integrating artificial intelligence (AI) into banking operations. This case study examines DBS’s strategic approach to AI implementation, as detailed by Nimish Panchmatia, Chief Data & Transformation Officer at DBS Bank.

Under Panchmatia’s guidance, DBS has successfully transformed over 240 experimental AI initiatives into more than 20 practical use cases. The bank’s AI journey focuses on enhancing customer experiences, improving operational efficiency, and fostering innovation across its services. Panchmatia shares insights into DBS’s proprietary ‘Innovation Pyramid’ framework, their ‘Managing through Journeys’ approach, and how they’re leveraging AI to process both structured and unstructured data while maintaining customer trust and data security.

Can you elaborate on how AI is augmenting customer experiences rather than simply replacing human interactions, particularly in your monthly quarter-million customer touchpoints?

At DBS, we view AI not as a replacement for human interaction, but as a powerful co-pilot to reduce toil for our employees so that they can deliver differentiated outcomes for our customers. Our Gen AI-enabled CSO Assistant, which will be progressively rolled out to our 500-strong customer service workforce in Singaporeby end of the year, helps our customer service officers (CSOs) manage over 250,000 monthly customer queries more efficiently and effectively. The system transcribes, summarizes and recommends solutions for customer queries in real-time, empowering CSOs to quickly access relevant information and provide more accurate, timely responses.

What sets our approach apart is a laser focus on creating an omnichannel, multilingual AI solution that seamlessly integrates across various customer touchpoints — from phone calls to emails. This interoperability is pivotal in today’s digital banking landscape. Moreover, recognizing the linguistic diversity of our Asian markets, we’ve prioritized localization. This effort ensures that we continue to provide personalized, culturally nuanced services across our regional markets.

As a result, we are expecting to see a 20% decrease in average handling time per service request, freeing up our CSOs to focus on more complex, high-value work. Importantly, this efficiency gain hasn’t come at the expense of quality — our solution maintains near 100% accuracy, with no customer-facing elements to maintain authenticity and personalization. By automating routine tasks and providing real-time assistance, we’re not only improving our operational metrics but also enhancing our employees’ job satisfaction and capabilities. This human-centric approach to AI integration ensures that technology serves to amplify, rather than diminish, the human touch in our customer interactions.

You’ve translated over 240 experimental Gen AI initiatives into over 20+ practical use cases. What key factors enabled this transition from testbed to implementation, and what advice would you offer other organizations facing similar challenges?

Our transition from 240+ experimental ground-up Gen AI initiatives to over 20+ practical use cases start with inculcating an environment for innovation. We’re guided by our proprietary ‘Innovation Pyramid’ framework, which socializes all major innovations across the bank and imbues employees with the mindset to experiment. This framework helps clarify processes, adds quality controls, and ensures ambition in our initiatives.

Building on this foundation, we implemented a structured, customer-centric approach we call “Managing through Journeys” (MtJs). This framework bridges the gap between innovation and implementation through data-driven experimentation and rapid iteration. We’ve embedded AI/ML into our value map drivers, optimizing decision-making and improving outcomes. Our Control Towers, which provide real-time data on business drivers and customer indicators, enable timely interventions and guide our AI development priorities. This approach allows us to quickly identify high-potential use cases and scale them effectively.

Our strategy emphasizes the synergy between human expertise and AI capabilities, positioning Gen AI as a supportive tool for our workforce. We’ve established a cross-functional Responsible AI taskforce comprising senior leaders from multiple disciplines to evaluate and mitigate potential risks before use-cases are deployed to production. Our AI initiatives are guided by our PURE framework, which ensures our data usage is Purposeful, Unsurprising, Respectful, and Explainable to customers. Furthermore, we’ve developed a secure technological foundation that facilitates the safe integration of Large Language Models into our operations.

For organizations facing similar challenges, I’d advise prioritizing use cases that directly impact employee efficiency and customer experience. Start with internal applications to build confidence and expertise before expanding to customer-facing solutions. Importantly, establish a robust governance framework early on — emergent risks such as data confidentiality, Gen AI hallucination, bias and toxicity are important pitfalls to avoid from the get-go, particularly as it relates to customer trust and security. Remember, the goal isn’t just to implement AI, but to create tangible value for your stakeholders.

Trust is crucial in financial services. How is DBS approaching the use of AI with unstructured data to ensure reliability and maintain customer confidence?

The emergence of Gen AI has opened up exciting possibilities for processing vast amounts of unstructured data. Combined with our existing capabilities in structured data, this allows us to sharpen our current AI use cases and explore entirely new data-driven applications. A great example is our Enterprise Knowledge Base that we are currently building. which consolidates unstructured data from across the bank, making it easily searchable while maintaining robust access controls.

When it comes to data use, we always ask ourselves three key questions: Can we use it legally? Should we use it ethically? And how can we use it responsibly? This approach helps us navigate the complex ethical landscape of data use, especially given the diverse markets we operate in. What might seem acceptable in one location could be considered intrusive in another, so we’re constantly balancing these considerations.

To ensure reliability and maintain customer confidence, we’ve embraced the Monetary Authority of Singapore’s FEAT guidelines — that’s Fairness, Ethics, Accountability, and Transparency. In practice, this means we’re meticulous about using representative data sets, justifying any use of sensitive information, and setting clear fairness objectives that we can measure. Importantly, we always maintain human oversight of our AI models. It’s about striking the right balance between innovation and responsibility, ensuring our customers can trust in the AI-driven services we provide.

As one of Asia’s largest banks, DBS has significant resources at its disposal. How do you balance leveraging this scale with the need for agility and innovation in AI implementation?

Simply put, we’ve embraced the challenge of becoming more like a technology company than a traditional bank. This mindset shift has been crucial in balancing our scale with the need for agility and innovation in AI implementation. To do this, we’ve drawn inspiration from tech giants to drive our transformation initiatives, keeping our focus firmly on the customer. 

Honing into AI specifically, our decade-long journey to become an AI-fuelled bank has been built on three pillars: process, technology, and people. On the process front, we’ve created a standardized, repeatable approach called the DBS AI Protocol (ALAN, named after Alan Turing), which has enabled us to scale to over 800 models and 350 use cases across the bank. Our technology infrastructure, centered around the bank’s internal data-as-a-service ADA Platform (Advancing DBS with AI), houses over 5.3 petabytes of data and allows us to hyper-personalize customer interactions. On the people side, we’ve built a Data Chapter with 700 professionals, fostering a culture of continuous improvement and innovation.

This approach has allowed us to drive agility at scale. Today, over 60% of our bank’s revenue is delivered through our Agile at scale program – Managing through Journeys (MtJ), where AI plays a pivotal role. Resources alone are not enough, we will also need to create a culture that embraces innovation as a collective force, while delivering tangible results for our customers and shareholders.

Looking ahead, what emerging trends do you see at the intersection of AI and financial intelligence, and how is DBS positioning itself to capitalize on these opportunities?

At DBS, we’re seeing AI and financial intelligence converge in exciting ways, particularly in hyper-personalization and predictive analytics. We’ve already created over 100 AI and machine learning algorithms that analyze 15,000 customer data points, generating personalized nudges that guide customers in their financial decisions. We’re now exploring how to incorporate behavioral and location data to serve specific customer segments even more effectively, such as parents, and avid shoppers.

One emerging trend we’re particularly excited about is the use of AI in risk management and customer protection. We’ve developed capabilities to alert customers to unusual transactions in real-time, especially if the transaction location doesn’t match the customer’s. For SMEs, we’re leveraging AI-powered algorithmic models to identify potential credit risks before they become issues, helping these businesses manage their finances more effectively.

We firmly believe predictive and generative AI will fundamentally reshape banking, which is why we’ve scaled our AI capabilities across all operations over the past five years, yielding tangible benefits of SGD 370 million for DBS in 2023 – more than double that of the previous year’s results. As we continue to expand our AI initiatives, we anticipate not only growing economic impact, but also strengthened operational resilience and enhanced adaptability to market changes and customer needs. This affords us greater agility to navigate evolving financial cycles and serving our customers effectively in an increasingly dynamic environment.