I recently had the pleasure to talk with Siew Choo Soh, group head of consumer banking technology and big data/AI at DBS, on how the bank has set up its enterprise data platform to enable a data-driven organization. That session was part of Forrester’s APAC Financial Services Webcast Week 2020, and you can find the full session for replay (as well as all other session replays) at https://forr.com/apacfsweek (free registration required).
In the session, Siew Choo and I talked about the situation at the starting point of the bank’s journey and how the team set the objectives and design principles for a single end-to-end platform. That platform, named “ADA — Advancing DBS with AI,” was conceptualized to provide data ingestion, data security, data storage, data governance, data visualization, and analytics model management capabilities.
“We saw the need to democratize the access to data and reduce the time from data to insights to become a truly data-driven organization.” Siew Choo Soh, group head of consumer banking technology and big data/AI, DBS Bank
I realized that there were three aspects of DBS’s journey that would be particularly helpful for others starting out on a similar journey. This blog post aims to summarize the first aspect: the architecture principles that guided the design of the ADA platform. I will cover the other aspects in a later blog post.
Four Architecture Design Principles Emphasize Continuous Evolution, Scale, and Security
The first aspect is about the design principles for ADA. Having been an architect myself in the past, I was very excited by the design principles that the ADA team had laid out at the very beginning:
Constant evolution of the platform and components. This allowed the bank to keep up with the rapidly changing landscape by continuously implementing, updating, and expanding individual frameworks and capabilities without breaking the architecture.
Multi-cloud infrastructure with open source priority. Virtualizing and decoupling the data and compute layers enables scale and frictionless use of private cloud and any public cloud. Further, DBS can benefit from the speed of development across a global open source community.
Autonomous self-service data and machine-learning platform. The ADA team set out to create a data and machine-learning platform-as-a-service, where users can use the platform via training by and pairing with the ADA team.
Security, privacy, and quality by design. The ADA team took great care to align closely with the bank’s strict security and data governance guidelines and integrated the capabilities into the design from the outset.
Setting up these four design principles at the very beginning of their journey has allowed Siew Choo and her team to establish the ADA enterprise data platform as an integral part of DBS’s enterprise architecture and make its capabilities available to all lines of business within the bank.
Operating Model, Processes, and Governance Change the Way the Bank Works
The second aspect may well be the one that organizations struggle with the most. In my experience, technology mostly delivers and is not an inhibitor. But changing the way an organization works to leverage those new technology capabilities can be daunting. Changing the organizational structure and processes — the way people work and interact — is very much a cultural change as it is a structural change.
DBS decided on a platform-as-a-service approach for data and ML to facilitate scaling the platform rapidly throughout the bank. The ADA team then set out to:
Define a new operating model and processes. The ADA team architects, develops, and supports an integrated data and ML platform. Business-aligned tech teams bring the data and build the necessary business logic. Each line of business (LOB) that uses the platform brings its resources and developers to build business functionality.
Establish governance and clear responsibilities across stakeholders. Governance is managed by an enterprise data council. The ADA team is responsible for the technical engineering, while the Data First team (lead by the bank’s chief analytics officer) owns the governance and processes. The bank also established an analytics center of excellence.
An Expanded Skills Matrix and People Enablement Drive ADA’s Success
The final aspect I am keen to share is about skills and talent. This is about assembling the right team and the right set of skills at the outset of the journey. It is also about enabling the entire bank to make use of the technology and data for insights. The two key takeaways from ADA for me are:
Equip your team with a diverse set of skills and backgrounds. To infuse the ADA team with new skills and ways of thinking, DBS added data and analytics technologists from outside the banking industry. Recruits from data- and AI-based startups and tech companies brought technical expertise in state-of-the-art technologies and architectures to the team.
Scale the use of the platform with continuous training and user enablement. DBS designed a comprehensive training program for ADA to enable its use quickly and at scale. In addition, customer success managers work with each business. The ADA team also provides a high-touch enablement approach by pairing with LOBs for more complex use cases.
“The assimilation of nonbanking technologists into the bank has been key to our success. It has not been an easy exercise, but it is a rewarding journey, as it resulted in creativity and innovation implemented at scale with security and controls embedded at its core.” Siew Choo Soh, group head of consumer banking and big data/AI technology, DBS Bank
Conclusion: ADA is a Major Step on the Way to Data-Enabling DBS
Multiple LOBs have implemented strategic AI/data initiatives on the new platform to move DBS closer to being a truly data-driven bank. Currently, ADA supports initiatives including customer science, intelligent banking, financial planning, recommendations, fraud surveillance, and sales surveillance across DBS.
First published on Forrester Blog
