Large-scale US regional bank: Transforming the enhanced due diligence process with AI

1 min read

AI & Machine Learning Digital Strategy Generative AI

A large-scale US regional bank’s risk analysis and enhanced due diligence (EDD) process required colleagues to manually review documents, analyse calculated risks and craft narratives, resulting in significant time consumption. They engaged us to explore the use of generative AI technology to reduce review times and enhance productivity without compromising accuracy.

Outcome 1: Achieved an 88% reduction in time to due diligence report generation, significantly accelerating workflows and improving service delivery
Outcome 2: Delivered 94% model accuracy, enabling trust in automated recommendations, with 100% accuracy when combined with a human
Outcome 3: Fitted the solution seamlessly into existing processes with compliance guardrails remaining intact
Outcome 4: Provided a strong foundation for future automation across adjacent workflows

The challenge

Our client wanted to streamline risk analysis and report generation in their EDD customer reviews while maintaining control over sensitive and obfuscated data.

The approach

We partnered with stakeholders to evaluate the existing enhanced due diligence (EDD) review process, identifying inefficiencies in both document analysis and narrative generation. This assessment provided a clear view of where automation could deliver value without disrupting established workflows.

Our approach began with a detailed mapping of the end-to-end process, highlighting opportunities for generative AI to augment risk analysis and reporting. Building on this, we developed a solution powered by large language models to automate critical steps in review and report creation.

To ensure safety and compliance, we ingested obfuscated sample customer data for testing, validating the model’s performance against organizational guardrails. The pilot confirmed the solution’s ability to handle sensitive data responsibly while maintaining accuracy and consistency.

Finally, we designed a scalable integration framework, enabling the solution to be embedded into daily workflows and extended to other due diligence processes across the organization.

The Impact

The new generative AI-powered solution sped up the EDD process by 88%, improving organizational efficiency and freeing up capacity for colleagues to take on higher-impact tasks. Model accuracy reached 94%, enabling trust in automated recommendations, and achieved 100% accuracy when combined with human review. The bank also gained a firm foundation for extending automation capabilities across other, similar processes.