Fujitsu's AI Aims to Decode Decades-Old COBOL Systems for Modern Teams

Fujitsu has introduced an artificial intelligence system designed to analyze aging COBOL code and automatically generate technical design documents. This move positions the company as a potential rival to IBM in the complex, high-stakes market for modernizing legacy enterprise systems.

The tool addresses a chronic problem: an enormous volume of critical business software—powering everything from bank transactions to government services—still runs on COBOL, a programming language from the late 1950s. Maintaining these systems is difficult because the programmers who understand them are retiring, and creating the documentation needed for updates is a slow, manual process requiring rare expertise.

Fujitsu's system uses machine learning to parse COBOL programs, map their logic and data flows, and produce structured design blueprints in minutes—work that typically consumes hours or days. The objective is to allow a broader range of IT professionals to understand and work on these old systems without first becoming COBOL specialists.

This development arrives as industries face increasing pressure to update foundational software. The effort highlights a strategic shift: instead of solely focusing on migrating code off mainframes, Fujitsu is automating the initial, labor-intensive analysis phase. If successful, this could offer organizations a faster, less expensive starting point for modernization projects, potentially altering competitive dynamics in a field long dominated by IBM and its deep mainframe ecosystem.

The tool's practical value will be determined by its accuracy when confronted with real-world COBOL code, which is often heavily customized and poorly documented. For business leaders overseeing legacy IT estates, however, it represents a promising attempt to apply AI to a concrete and expensive operational challenge.

Source: Webpronews