Wealth Management
APAC wealth managers must free relationship managers (RMs) from low-value administration so that they can become more trusted decision partners and wealth orchestrators. Agentic AI tools offer a way to accomplish this while helping RMs to gather compliance information as part of a strengthened client conversation, rather than as a checklist exercise.
In this Q&A, Kelvin Chiang, Head of Compliance Transformation at OCBC (formerly Head of Platform & Analytics at Bank of Singapore), explains how he used an agentic-AI powered Source of Wealth Assistant (SOWA) to help RMs along this path, the lessons learned, and what the innovation signals for the future of compliance and wealth management.
Why did you want to develop SOWA?
In my time at UBS I learnt the importance of using a strong platform to drive human behavior. I wanted to build a tool to enable RMs to apply the practices we see in our most experienced RMs. SOWA was designed to improve risk awareness and help RMs prepare stronger and more consistent source-of-wealth narratives.
SOWA helps RMs to draft source-of-wealth narratives, read financial statements, extract relevant information, perform calculations, and identify gaps between documented wealth and supporting evidence. The AI highlights these gaps for the RM to review and supplement, or apply judgement as appropriate.
SOWA is an assistive tool. It does not make client acceptance or compliance decisions. Accountability for assessing and corroborating a client’s source of wealth remains with the RM, supported by the Bank’s established review and approval processes.
What benefits have you experienced since implementing SOWA?
The key benefits have been greater consistency in source-of-wealth narratives and better alignment with the Bank’s SOW requirements. RM and reviewer discussions have shifted from correcting administrative details to focusing on material risk and compliance considerations.
We have also seen productivity benefits. Straightforward drafts can be prepared significantly faster, while SOWA can help reduce the amount of rework required for more complex cases.
Why is it better for RMs to use SOWA rather than asking a general-purpose copilot?
We wanted to drive consistency. SOWA embeds the most appropriate prompts, standards and review logic within a controlled workflow and can leverage internal reference datasets to support industry and financial benchmarking subject to applicable data governance and access controls.
SOWA also helps RMs understand the importance of client documentation when corroborating the source of wealth. With SOWA, the more client corroborative evidence that is collected, the better the draft, resulting in less rework by the RM.
What ‘lessons learned’ can you share with others looking to build a SOWA or similar AI solutions?
A key success factor was having a strong product manager who understands the business, technology and compliance requirements. Bringing together Business, Technology and Compliance expertise from the outset also helped us translate regulatory requirements into the solution design.
Our user focus group was also an important source of knowledge. Compliance and business-aligned risk and control colleagues provided subject-matter expertise and reviewed AI outputs, helping the team understand where the technology worked well and where further refinement was required. Finally, engaging the appropriate governance and assurance stakeholders early was important. We set up cross-functional governance and provided regular updates to the steering committee so that issues could be identified, challenged and resolved as the solution developed.
Early engagement with relevant risk, governance and assurance stakeholders also helped us identify potential control considerations before implementation, while maintaining the respective responsibilities of the different control functions.
What would you say to those currently building AI solutions about setting expectations and timelines?
It is easier to get out a PoC and MVP, however, after the initial couple of iterations, stakeholder expectations tend to significantly increase regarding the final, integrated product. In our case, the pilot took one month, while full end-to-end implementation took around 12 months including embedding it within our IT environment.
What can be transferred or leveraged for other AI use cases?
I work closely with our Group Data Office and agentic AI solutions are shared in the OCBC AI marketplace as resuable AI where appropriate. We have already been able to leverage some of SOWA’s capabilities for other use cases. For example, we have built an agent that recognizes client documents for KYC remediation.
What do you see as the wider implications of AI from an operating model perspective?
Client expectations are changing. Clients increasingly have access to information and may conduct their own research before speaking to an RM. I believe private banking RMs should be like medical specialists. As a client, you may do your own research, but ultimately you will want to discuss this with an expert and then make decisions together after fully understanding the pros and cons of the various options.
I’m passionate about helping RMs across the wealth spectrum to have better conversations with their clients by bridging compliance requirements with business understanding. A client will have a notably different experience if the RM says “I need to go through this compliance check list”, compared to an experienced RM who naturally weaves the questions into a conversation about how a founder built their business.
AI can now help in crafting that conversation so that less experienced RMs can adopt the practices of more experienced RMs, enabling better client experiences while allowing RMs to spend more time on higher-value client engagement.
The future of Compliance will also change. I believe that in the future Compliance will be augmented with AI agents. However, the impact will be greatest where Compliance understands the business challenges in addition to the regulatory requirements – and moves upstream to guide better outcomes from the start.
Capco viewpoint: SOWA was created to improve source-of-wealth narratives by increasing consistency, strengthen risk awareness by shaping behavior, and help RMs approach compliance as part of a better client conversation.
The broader lesson is that compliance does not need to sit only at the end of the process, as a reviewer or gatekeeper. With AI, Compliance can move upstream. It can help design and shape the tools, prompts, controls and behaviors that guide better outcomes from the start.
This Q&A is part of a Capco thought leadership series on the transformation of wealth management in the APAC region. Explore further by reading our in-depth research paper, Asia’s Wealth Boom is Breaking the RM Model.
Related Insights
Get in touch
To find out more about working with Capco and how we can help you overcome any potential challenges, contact our experts or subscribe for the latest insights below.