DBS

DBS, Southeast Asia's largest bank, faced a dual challenge as it scaled its retail and wealth businesses: how to drive new customer acquisition and deepen engagement at volume without proportionally expanding operational headcount. The bank had invested heavily in AI and machine learning over several years, but as these tools became more deeply embedded in daily workflows, isolating their precise economic contribution grew increasingly difficult. Meanwhile, the burden of technical debt — maintaining and fixing outdated systems — was consuming engineering capacity that could otherwise fuel growth, making the status quo costly in both direct spend and opportunity cost.

DBS deployed machine learning-powered contextual nudges and automated customer engagement systems to attract new-to-bank customers and drive volume growth across retail and wealth channels. These tools operate by surfacing personalized financial insights to customers at the right moment, with economic impact measured through rigorous A/B testing — comparing outcomes between customers offered AI-driven solutions and a matched control group. In parallel, the bank rolled out DBS GPT, an in-house generative AI platform now used for translation, internal policy queries, and employee guidance. AI was also applied directly to technical debt reduction, compressing remediation work that previously spanned months or years into weeks, freeing engineering capacity for growth initiatives.

DBS recorded S$1 billion in AI-derived economic value in 2025, up one-third from S$750 million in 2024 — a 33% year-on-year increase. CEO Tan Su Shan attributed record deposit inflows, wealth growth, and fee income in part to the bank's sustained AI investment. Key outcomes include:

• S$1B total AI economic value in 2025 (vs S$750M in 2024)
• 33% YoY growth in measured AI value
• 60%+ of staff now actively using DBS GPT
• Technical debt work compressed from months or years to weeks
• Record deposit inflows and wealth growth attributed to AI-driven customer engagement

• Measure AI impact through A/B testing early in deployment; as adoption deepens, economic value increasingly shows up as capacity freed for growth rather than discrete revenue lines.
• Generative AI tools improve significantly with iteration — DBS GPT was described as poor at launch but substantially better within months, so early underperformance should not trigger premature abandonment.
• Applying AI to technical debt reduction delivers compounding returns: faster remediation frees engineering resources to build growth-oriented capabilities.
• Framing AI ROI as capacity redeployment — not headcount reduction — makes the business case more durable and aligns leadership incentives with growth.