To move beyond fragmented data estates and bring AI into day-to-day banking operations, United Overseas Bank (UOB) and Standard Chartered have each had to reconcile innovation with regulatory constraint. Both banks are pursuing AI to reduce inefficiencies and extract value from data, yet neither could do so without first addressing governance, scalability, and compliance requirements. In working with Cloudera, they have focused on building data foundations intended to support AI use cases across their organisations.
Dealing with data silos
UOB’s data transformation journey began in 2015 with the launch of its Enterprise Data Architecture and Governance (EDAG) program, built on Cloudera. At the time, the bank faced several challenges, including fragmented data sources, the absence of a coherent data strategy, and low data literacy across the organisation.
“As a bank, UOB accumulates large volumes of data across different business units through our daily course of work. Previously, much of the data had remained within individual legacy systems, making cross-system analysis difficult and slowing decision-making,” noted Alvin Eng, Head of Enterprise AI, United Overseas Bank.
Scalability was also a challenge for the multinational bank. With exponential growth in data from digital transactions such as online banking and ATM usage, legacy platforms became slow, costly, and difficult to scale.
“Rigid data structures further slowed innovation, and continued investment in such infrastructure was also not sustainable. These challenges therefore hindered the bank’s ability to adopt advanced technologies such as AI,” Eng explained.
Meanwhile, Standard Chartered needed a platform that could centralise governance without compromising on-premises regulatory and security requirements. This required a platform with governance controls such as native data lineage, cataloguing, policy enforcement, and the ability to operate within on-premises environments.
“These qualities enable us to adhere to high levels of data governance and AI safety, which we see as an essential cornerstone of our AI strategy,” said David R. Hardoon, Global Head of AI Enablement, Standard Chartered.
Preparing to deploy
From the outset, Standard Chartered needed to overcome two challenges before it could deploy its generative AI use cases: The first was ensuring that AI-ready data was available and could be ingested reliably. The second was maintaining access to sufficient compute.
“We worked with Cloudera to address these challenges so our internal AI teams could operate more effectively. This allows us to route requests to selected generative AI models and to test scenarios in controlled environments for experimentation, prototyping, and validation,” Hardoon said.
UOB, for its part, needed to address data accessibility, quality, and scalability issues. These included navigating fragmented data storage across multiple systems, inconsistent data quality, high infrastructure costs, and limited scalability.
“Using Cloudera as our data platform, we were able to store, clean, and retrieve data more efficiently. This helped us address issues such as metadata consistency, data quality remediation, and natural language query capabilities. As a result, we established a data foundation that supports our AI initiatives, including more advanced use cases such as generative AI,” Eng explained.
UOB also introduced “playpens,” or sandbox environments, where teams could prototype solutions in a controlled setting. This approach, Eng said, encouraged experimentation among employees while maintaining governance standards.
“We also organised a series of training programs and workshops for our employees to equip them with the right skills to use new technologies effectively,” he added.
According to Remus Lim, Senior Vice President, Asia Pacific & Japan, Cloudera, banks’ most critical requirements centred on data unification, governance, and the responsible deployment of AI at scale.
“The goal is to derive real-time insights to improve decision-making and customer experiences,” he said.
Business transformation
Following its partnership with Cloudera, UOB has scaled several AI initiatives into production over the years. These include AI models focused on portfolio optimisation and customer engagement, which are now embedded within the bank’s operations and used to generate real-time insights with measurable outcomes.
UOB also reported improvements in operational efficiency, particularly in cash-in-transit optimisation.
Previously, ATM replenishment relied on manual forecasting, which often resulted in inefficient replenishment schedules and higher operational costs. UOB’s data scientists analysed ATM transaction data and built a predictive model to forecast cash requirements at each ATM.
“This optimisation ensured ATMs are stocked when needed, reducing unnecessary cash movements. Since deploying this solution into production in 2022, we have achieved up to 33% reduction in replenishment trips and saved SG$1.3 million annually,” Eng said.
In risk management, UOB data scientists developed behavioural indicators and machine learning models to identify unusual activities and transactions. This approach improved efficiency and accuracy in risk monitoring while reinforcing the bank’s compliance and governance requirements.
Meanwhile, Standard Chartered reported tighter controls over personalisation, improved precision in interactions, and more flexible approaches to data quality and compliance.
“We are able to limit training datasets for model training to certain document types, such as economic reports. These granular controls allow us to strengthen risk oversight and explainability, reinforcing AI safety practices across the bank,” Hardoon noted.
Following the adoption of Cloudera’s platform, both banks were able to manage, secure, and analyse data across on-premises and cloud environments through a single governed architecture. This foundation, Lim said, allowed UOB and Standard Chartered to build and operationalise AI use cases while maintaining compliance with strict regulatory standards.
Banking on reliability
Looking ahead, Standard Chartered intends to continue with a targeted and structured approach to AI.
“Our focus is on establishing an operating model built around reusability, efficiency, and scalability. To do that, we need strong foundations across strategic processes, technical infrastructure, talent, and especially AI governance and safety. This work illustrates how we are building a data ecosystem to support responsible and sustainable use of AI,” Hardoon said.
For UOB, the stated objective is to build an AI-enabled bank anchored in trust and operational resilience.
“To this end, we are accelerating AI adoption by advancing the use of generative AI and agentic AI across the bank. These technologies will support efforts to streamline operations, strengthen compliance and control frameworks, and deliver more personalised customer experiences,” Eng said.
As banks move away from siloed systems towards unified data architectures, this reflects a broader shift in how financial institutions approach AI adoption, Lim observed.
“Banks are embedding governance, lineage, and access controls into every stage of the data lifecycle so that AI outputs remain explainable, fair, and auditable. By treating data as a strategic asset rather than a technical by-product, banks are ensuring that insights are built on a foundation of integrity and transparency,” he said.
