July 31, 2026, 1:12 p.m.
Mitsubishi Research Institute (MRI) began providing its “Credit Review AI Service” to Shizuoka Bank on May 18, 2026, to support the bank’s unsecured loan review processes. By automating credit review tasks with AI, the service aims to speed up the review process, improve operational efficiency, and lead to the provision of higher-quality financial services.
Amid an increasingly challenging business environment and intensifying market competition in the financial industry, business reform through DX (Digital Transformation) has become an urgent priority for financial institutions.In the retail lending sector in particular, there is a growing need to further improve “response speed,” “efficiency,” and “objectivity” in loan screening, and the effective utilization of accumulated data has become a critical challenge from the perspective of preserving screening expertise.
To address these challenges, MRI and Shizuoka Bank have been working since August 2023 to implement MRI’s “Credit Review AI Service,” and began its practical application on May 18, 2026.The service applies to the underwriting of unsecured loans at Shizuoka Bank that are handled by Shizugin Credit Guarantee. Through this initiative, the banks aim to improve operational efficiency and speed up underwriting decisions.
The “Credit Review AI Service” trains an AI model to make approval/rejection decisions—previously performed by humans (credit analysts)—and integrates this model into MRI’s “Credit Review AI System” to provide it as a service. By linking this service to financial institutions’ loan review systems, it promotes the automation of the credit review process.
Specifically, for each application received, the AI model calculates the probability of approval. Based on this probability, approximately 50% to 80% of all applications can be approved automatically and in real time without human review. The company also expects that these rapid responses will help reduce customer churn.
Furthermore, the AI model utilizes the information necessary for lending decisions to replicate each financial institution’s review criteria and decision-making logic (credit policy). Additionally, by consolidating review expertise within the AI, the system reduces variations in judgment caused by differences in experience or the specific reviewer, thereby improving review accuracy and lowering default rates.
Furthermore, APIs will be used to integrate the “Credit Review AI System” with the financial institutions’ existing loan review systems. Since development on the loan review system side will require only minor modifications, such as to the interface, implementation can be achieved at a low cost.
Furthermore, since an independent “Credit Assessment AI System” is built for each financial institution, monitoring and maintaining the AI models is made easier, resulting in high maintainability.
この記事の著者
ペイメントナビ編集部
カード決済、PCI DSS、ICカード・ポイントカードの啓蒙ポータルサイト
