Sony Bank and Fujitsu integrate generative AI into core banking development, achieving a 30% reduction in timelines and 40% fewer man-hours using AWS and Claude.

The modernization of financial infrastructure often moves at a glacial pace due to the extreme risk associated with core banking systems. However, a recent collaboration between Fujitsu and Sony Bank has demonstrated that generative AI can drastically accelerate this process without compromising the stability of live operations. By integrating AI into the very fabric of their development lifecycle, the two companies have transitioned from theoretical proofs-of-concept to a full-scale industrial application.

A 30% acceleration in system delivery

The results of the initiative, confirmed in July 2026, reveal a significant shift in productivity. The application of generative AI across the development pipeline led to a 30% reduction in the time required to move from basic design to integration testing. This is not merely a marginal gain but a structural shift in how banking software is built. Beyond the calendar timeline, the project reported a 40% reduction in total man-hours, suggesting that AI is absorbing the most labor-intensive aspects of the coding and documentation process.

This efficiency was achieved by creating a specialized mechanism that allows the AI to ingest and utilize existing development assets. Rather than relying on generic prompts, the system leverages design documents, source code, and test assets to ensure that the generated output is aligned with the specific architectural requirements of the bank.

The architecture behind the automation

The foundation of this success lies in the synergy between cloud-native infrastructure and advanced Large Language Models (LLMs). Sony Bank utilizes the Fujitsu Core Banking xBank solution, which operates on Amazon Web Services (AWS). This cloud-native approach provided the necessary agility to integrate AI services rapidly and conduct phased trials across different development stages.

At the heart of the intelligence layer is an AI agent centered on Anthropic PBC’s Claude, deployed via Amazon Bedrock. This choice of model allows for sophisticated reasoning and high-context window capabilities, which are essential when analyzing complex banking source code and dense technical specifications. The process is not fully autonomous; human oversight remains the final gate for judgment and quality assurance, ensuring that the AI-generated code meets the rigorous security and regulatory standards of the financial sector.

From cloud migration to AI integration

The move toward AI was the logical next step in a broader digital transformation strategy. In May 2025, Sony Bank completed a total migration of its systems to the cloud. This transition to a cloud-native core banking system provided the flexibility and scalability required to experiment with generative AI. Without the prior removal of legacy on-premise constraints, the rapid deployment of AI agents across the development lifecycle would have been technically prohibitive.

By first establishing a modern infrastructure and then layering AI on top of it, Sony Bank has created a blueprint for the evolution of financial institutions. The AI is now applied progressively to design, manufacturing, and testing, transforming the core banking system from a static piece of software into a dynamic ecosystem that can be updated and optimized with unprecedented speed.

Scaling the AI development ecosystem

This initiative represents the first phase of a larger ambition to build a comprehensive AI-driven development ecosystem. Sony Bank and Fujitsu intend to leverage the knowledge and assets gained from this project to expand AI usage into other internal processes and auxiliary systems. The goal is to move away from episodic AI use and toward a state where generative AI is a permanent, integrated component of the software development life cycle (SDLC).

The use of generative AI is not limited to proof-of-concept (PoC) but has been established as a development process for a core banking system in live operation.

Fujitsu, for its part, plans to export the lessons learned from the Sony Bank project to other financial institutions. By refining the Core Banking xBank framework, Fujitsu aims to standardize AI-assisted development for the global banking industry, potentially lowering the barrier for other banks to modernize their legacy cores.

Strategic implications for global enterprises

For entrepreneurs and tech leaders in the USA, UK, and other global markets, the Sony Bank-Fujitsu case study provides critical insights into the industrialization of AI. The project proves that generative AI can handle high-stakes, mission-critical systems if the right guardrails and infrastructure are in place. The transition from 'experimentation' to 'full-scale application' is the current frontier for enterprise AI.

In the US and UK markets, where financial regulations are stringent, the 'human-in-the-loop' model used here is particularly relevant. By using AI to handle the bulk of the production while reserving final validation for human experts, firms can achieve massive efficiency gains without bypassing compliance requirements. Furthermore, the reliance on a cloud-native stack (AWS) and a specialized LLM (Claude) highlights the importance of choosing a flexible ecosystem over a rigid, single-vendor proprietary solution.

The primary takeaway for global businesses is that the biggest gains from AI are not found in customer-facing chatbots, but in the 'invisible' back-end processes. Reducing development cycles by 30% in a sector as conservative as banking suggests that similar efficiencies are available in any industry burdened by complex legacy systems and heavy documentation requirements.

FAQ

---

Sources: Capitalradio, Laprensalibre, Frontier-enterprise ·