Sony Bank, Fujitsu deploy generative AI in live core banking build, cut time 30%

The two companies built the effort together with Amazon Web Services Japan, using Fujitsu's cloud-native, microservice-based Fujitsu Core Banking xBank on AWS as the foundation for applying AI. According to the body, that microservice structure mattered: it allowed quick AI integration and step-by-step application by function, so effects could be checked continuously across design, manufacturing, and testing. Rather than applying generative AI to a single step, they applied it across the whole span from design to testing, reusing development assets across stages, and folded it into the standard development process with Fujitsu specialists evaluating and improving it to keep safety and quality intact.

Beyond the headline figures, the body reports specific results: up to 90% fewer hours for impact studies in basic design and for integration test execution, up to 40% fewer hours for detailed design document creation, and a 99% code generation rate in manufacturing. Sony Bank says it aims to apply generative AI across the entire development process, including requirements definition, and to extend it gradually into security and operations and maintenance. Fujitsu says it will use what it learned as a development model for other financial institutions and in a dedicated AI platform supporting financial institutions' AI use.

What this ultimately amounts to may depend on whether the efficiency gains hold as the scope widens to requirements definition and to security and operations work, areas the body describes as still ahead. For Sony Bank, the stated goal is an "AI-native bank"; for Fujitsu, the value of the effort appears to rest on whether the approach travels to other financial institutions' development, which the body presents as the planned next step rather than a completed result.