Fujitsu Launches AI Service for COBOL Analysis and Modernization

Fujitsu has launched a generative AI service designed to automate the analysis of legacy source code and the generation of corresponding design documents. The service, known as Fujitsu Application Transform powered by Fujitsu Kozuchi, specifically targets the modernization of legacy systems that rely on COBOL and other older programming languages.

The tool is intended to reduce the significant human effort and time typically required to understand complex legacy systems. According to Fujitsu, the service can reduce the time spent on design document generation by approximately 97%, allowing organizations to produce documentation in minutes rather than hours.

Technical Implementation and RAG Integration

Fujitsu Application Transform leverages a combination of proprietary code analysis techniques and generative AI. To ensure the accuracy of the generated documentation and mitigate common AI issues, the service utilizes a specific retrieval-augmented generation (RAG) framework.

The system manages RAG through Fujitsu Knowledge Graph–Enhanced RAG for Software Engineering. By linking large volumes of source code via this knowledge graph, the service aims to prevent hallucinations and omissions, resulting in design documentation that is highly accurate and easy to read.

This approach allows the system to operate and generate documentation without requiring the user to possess expert knowledge of the specific programming languages involved, thereby reducing the industry’s reliance on specialized human programmers for legacy code interpretation.

Service Availability and Evolution

Fujitsu began offering the service as a Software as a Service (SaaS) in Japan on March 30, 2026. The launch follows the development of a predecessor software analysis and visualization service that was introduced in February 2025.

The new offering standardizes the analysis technology and the know-how required for generating design documents. It analyzes remaining design information and existing programs to create a clear understanding of current system specifications.

Impact on System Modernization

The primary goal of the service is to support the modernization strategies of companies and organizations burdened by legacy infrastructure. By automating the documentation process, Fujitsu aims to eliminate the need for extensive manual reviews of source code.

The service provides several key improvements over general-purpose generative AI tools:

• Enhanced quality of generated documentation through proprietary analysis techniques.
• Reduction of processing time for understanding legacy code by approximately 97%.
• Ability to generate design documents without deep programming expertise.
• Improved documentation completeness via Knowledge Graph retrieval.

By leveraging Fujitsu’s system development expertise and proprietary technology, the tool transforms decades of legacy code into accessible design documents, facilitating a more efficient transition to modern system architectures.