DBS Deploys AI Agents to Handle Credit Assessment Tasks Across 1,500 Staff

Automating the Credit Memo Workflow

DBS has introduced an agentic AI system designed to handle credit assessment tasks across its institutional banking division, serving approximately 1,500 employees globally. The tool automates more than 70 discrete tasks involved in preparing credit memos, documents that evaluate a company's financial health and risk profile before the bank extends financing.

The bank disclosed that relationship managers currently spend up to 40% of their time on credit memo preparation and related administrative work. DBS expects the new system to reduce that burden by at least 30%, according to the bank.

The AI tool pulls data from annual reports, industry research publications and internal bank records to generate a first draft of a credit memo. Employees then review, refine and approve the document. Relationship managers and credit risk managers retain final responsibility for the assessment and must layer in their own knowledge of the client, sector dynamics and macroeconomic context.

Knowledge Capture at Scale

Han Kwee Juan, Group Head of Institutional Banking at DBS, described the deployment as an effort to codify expertise from top-performing staff. "We believe that agentic AI can help to reimagine corporate banking," Han said. "Through this capability, we have been able to capture the knowledge and insight of our best relationship managers and credit risk managers, turning these into a solution which enables us to level up the quality of our credit analysis at scale."

The system uses specialised AI agents, each trained to perform specific sub-tasks within the credit assessment process. Employees can prompt the agents to conduct additional research or revise sections of the draft memo, creating an iterative workflow that blends automation with human judgment.

Freeing Time for Strategic Work

DBS anticipates that the time saved will allow relationship managers to focus on higher-value activities such as strategic client conversations, deal structuring and portfolio planning. Credit risk managers, meanwhile, could devote more attention to portfolio strategy, risk calibration and monitoring emerging threats across sectors and geographies.

The move reflects broader momentum in Asia's banking sector toward deploying AI in middle- and back-office functions. Singapore, Hong Kong and Tokyo have seen financial institutions experiment with generative AI for compliance, risk management and client onboarding, driven by regulatory openness and competitive pressure to improve efficiency.

Wider Agentic AI Push

The credit tool is part of a larger initiative at DBS to embed agentic AI across both customer-facing and internal operations. In July, the bank announced upgrades to its DBS Joy and DBS digibot virtual assistants, which will serve roughly 10 million customers across Singapore, Hong Kong and Taiwan. Those assistants now use agentic AI to handle multi-step queries, moving beyond scripted responses to perform tasks such as transaction research and account adjustments.

Agentic AI differs from earlier chatbot models by enabling systems to plan, execute and adapt sequences of actions with minimal human intervention. In the context of credit assessment, that means the AI can identify gaps in a draft memo, pull additional data from multiple sources and revise its output without requiring step-by-step instructions from the user.

Risk and Oversight Considerations

While the automation promises efficiency gains, it also raises questions about oversight and accountability. Credit decisions carry significant financial and reputational risk, and regulators across Asia have emphasised that banks remain fully accountable for AI-driven processes. DBS has structured the tool so that human approval remains mandatory, but the quality of that oversight will depend on how thoroughly staff review AI-generated drafts under time pressure.

The bank has not disclosed the specific AI models or training data used, nor whether the system has been tested for bias in credit assessments across different industries or geographies. Transparency around model performance and error rates will be critical as DBS scales the tool across its institutional banking operations.

For now, the deployment signals DBS's confidence that agentic AI can handle complex, judgment-intensive workflows at scale. Whether other regional banks follow with similar systems will depend on regulatory clarity, vendor maturity and internal risk appetite. The next twelve months will likely show whether the efficiency gains materialise without compromising credit quality or staff oversight capacity.