Bank of Singapore deploys agentic AI tool to automate writing of source of wealth reports

Bank of Singapore has rolled out an agentic AI-powered tool that automates a key part of the Know-Your Customer due diligence process to ensure the legitimacy of clients’ wealth and transactions. With the use of the tool, Source of Wealth Assistant (SOWA), relationship managers are able to deliver more consistent Source of Wealth (SoW) reports within a shortened timeframe, from 10 days to one hour, while ensuring alignment with regulatory standards. This in turn improves account opening efficiency, an area that an industry working group has been set up to look into.

Previously, relationship managers manually reviewed hundreds of pages of documents on the clients’ background, investment history, business activities, professional career, and corroborative evidence to write the SoW report. These documents typically include financial statements, tax notices, property valuation records, corporate filings, and payslips.

With SOWA, it automatically reviews and generates a comprehensive, well-structured, and standardised report after the documents have been gathered and uploaded into the tool by the relationship manager. The accuracy of the report is substantially improved. Human errors, such as inconsistent and missing information—often associated with the earlier manual process due to the varying experience among relationship managers—are minimised.

Leveraging the extensive database of Bank of Singapore and its parent company OCBC, SOWA is also able to validate the plausibility of the information provided by clients using benchmarks such as salary and company revenue.

Relationship managers maintain oversight of the SoW reports by reviewing and refining the AI-generated draft before submitting them for further assessment by the internal review teams as part of the Bank’s anti-money laundering and counter terrorism financing controls.

For security, information processed by SOWA remains hosted on the Bank’s private cloud – a computing environment dedicated to only Bank of Singapore’s use.

Mr Kam Chin Wong, Global Head of Financial Crime Compliance, Bank of Singapore, said: “In an increasingly complex risk landscape, artificial intelligence (AI) can play a pivotal role by automating repetitive tasks like report generation and data validation. Agentic AI pushes the envelope further by enhancing efficiency, accuracy and consistency in decision-making. With AI integrated into the source of wealth reporting process, relationship managers can shift their focus from manual documentation to meaningful client engagement and risk assessment. This not only strengthens client relationships but also maintains high standards of regulatory compliance while delivering greater value.”

Ms Ruth Yeo, Relationship Manager, Bank of Singapore, said: “SOWA has helped to make the process more intuitive. Instead of spending days sifting through emails and documents, looking for the relevant information before writing the source of wealth report, all I need to do now is verify the uploaded details and refine the writeup generated by SOWA. ”

Agentic AI – the next generation of AI

The use of agentic AI builds on the earlier investments in generative AI and traditional AI by OCBC that co-developed SOWA with Bank of Singapore.

Agentic AI marks a fundamental shift from passive assistance to proactive autonomy. Unlike generative AI, which reacts to prompts and does not act on its own, agentic AI systems pursue goals with intelligent initiative. They independently start tasks, coordinate tools across multiple platforms, adapt dynamically in real time, and continuously enhance their performance through memory and learning. By leveraging past interactions, agentic AI progressively refines its behaviour—capabilities that traditional large language models and machine learning models lack.

AI is embedded across OCBC’s operations, powering over six million decisions daily to drive revenue growth, mitigate risk and boost productivity. All in-house AI tools are built in accordance with the FEAT principles – Fairness, Ethics, Accountability, and Transparency – and undergo regular reviews to ensure accuracy and prevent bias across gender, nationality and other groups. 

Selected AI use cases in OCBC