An integrated Artificial Intelligence and optimization model for operational efficiency and risk reduction in Letter of Credit examination process

Digital transformation in banking has significantly improved efficiency, including in the critical Letter of Credit (LC) examination area. Despite these advancements, LC examination remains complex, labor-intensive, and error-prone, leading to operational risks and inefficiencies. Integrating Artificial Intelligence (AI) offers a promising solution but requires human checkers to verify AI-generated decisions, ensuring accuracy and compliance. Assigning these verification tasks is essential to fully capitalize on AI’s potential, balancing time savings with risk reduction. This paper explores the underexamined challenge of optimizing the hybrid process of AI-assisted LC examination to enhance trade finance. The research aims to minimize examination risk and maximize checker capacity utilization by offering practical strategies for improvement. Through data-driven research collaboration with international banks and FinTech companies and benchmarking relevant literature, an Integer Linear Programming model was developed to assign review tasks for LC documents indexed by AI based on their criticality and discrepancies to human checkers. The model also considers monetary value, checker expertise, and availability factors. Real case studies evaluated improvements over baseline practices, objectives prioritization, trade-off analysis, and varying supply–demand scenarios. The model achieved a 68.3% reduction in operational risks, improving compliance and trustworthiness while minimizing errors and financial losses. Utilization rates ranged from 34% for risk-focused strategies to 73% for efficiency-driven approaches, providing flexibility for resource allocation aligned with organizational priorities. Using the proposed AI-optimization framework and results, the study offers actionable insights for managers and guidance for researchers. • Propose an integrated artificial intelligence and optimization framework for letter of credit examination. • Develop an integer linear programming model to enhance efficiency and reduce risks through dual objectives. • Use real-world data to evaluate improvements, objectives trade-offs, and supply–demand sensitivity. • Achieve 68.3% risk reduction while ensuring high utilization and zero delays compared to baseline. • Provide research avenues on artificial intelligence integration, compliance, and sustainability.