DBS Bank, Southeast Asia's largest bank by assets, embarked on a comprehensive AI transformation to maintain competitive advantage in an industry undergoing rapid digital disruption. Despite being recognized as the world's best bank by multiple industry publications, DBS leadership recognized that their traditional technology operating model — characterized by long development cycles, siloed data infrastructure, and manual processes — would not sustain their leadership position.
The bank's AI initiatives were fragmented across business units, with over 40 separate AI projects running independently without unified governance, shared platforms, or consistent deployment practices. This created inefficiencies, duplicated effort, and slow time-to-market. Model deployment timelines averaged 18 months from concept to production, during which business requirements often shifted, rendering the deployed solution partially obsolete.
Data governance presented another fundamental challenge. DBS held petabytes of customer and transaction data across legacy core banking systems, but the data was inconsistent, poorly cataloged, and difficult to access for analytics and model training. Credit risk models, fraud detection systems, and personalization engines operated on different data definitions, creating reconciliation challenges and limiting the bank's ability to build integrated AI solutions.
DBS implemented a comprehensive AI transformation program centered on building an enterprise-wide AI platform with standardized data infrastructure, model governance, and deployment automation. The bank established a central AI & Data Office that set standards, provided shared services, and accelerated AI adoption across all business units.
The transformation included building a unified data lake with harmonized customer, transaction, and product data accessible to all AI development teams. DBS invested heavily in cloud infrastructure, migrating AI workloads to scalable platforms that reduced infrastructure deployment time from months to days. The bank adopted DevOps and MLOps practices, automating model deployment pipelines and reducing the time from model training to production deployment from 18 months to under 5 months.
DBS also reimagined its organizational structure, embedding data scientists and AI engineers directly into business units rather than maintaining them in a separate technology function. This brought technical expertise closer to business context, enabling faster iteration and more relevant AI solutions. The bank established an AI ethics framework and model risk management protocols that met regulatory requirements while enabling innovation at scale.
“AI is not a project or an initiative — it is a fundamental reimagining of how we serve customers, manage risk, and operate the bank. Our AI transformation has become a competitive moat.”— Piyush Gupta, CEO, DBS Bank
This case study is based on publicly available information about DBS Bank.
Learn more about DBS BankEvery transformation starts with a conversation. Let's discuss your challenges and opportunities.
Discuss Your Challenge