Ted Hisokawa
Oct 14, 2025 17:06
GitHub Copilot and AI agents are revolutionizing the modernization of legacy systems, making obsolete COBOL systems accessible to today’s developers, according to GitHub’s insights.
In an era where legacy systems still underpin critical global infrastructures, GitHub Copilot and AI agents are stepping up to modernize these outdated frameworks. According to a report by GitHub, these technologies are particularly effective in making COBOL systems, which have been in use for over six decades, accessible to contemporary developers.
The Challenge of COBOL
COBOL, a programming language that predates the internet, remains integral to many banking, insurance, and government systems. Despite its importance, finding developers skilled in COBOL is increasingly difficult as the original experts retire. This has left a significant portion of the world’s 200 billion lines of COBOL code in need of modernization.
AI-Powered Solutions
Julia Kordick, a Microsoft Global Black Belt, has been at the forefront of using AI to modernize COBOL systems without needing to learn the language. By leveraging AI for reverse engineering, her team can extract and document business logic, identify dependencies, and provide a comprehensive understanding of legacy codebases.
A Systematic Approach to Modernization
Kordick’s team employs a three-step framework for modernization: code preparation, enrichment, and automation aids. GitHub Copilot assists in reverse engineering code to understand its functionality, while AI agents enrich this understanding by translating and analyzing the code’s structure.
Automation aids further scale this process, allowing for the generation of call chain maps and test-driven modernization plans. These steps not only streamline the modernization process but also ensure that human experts remain integral to validation and oversight.
Real-World Application and Limitations
While AI provides significant efficiencies, Kordick warns that it is not a panacea. Human expertise is essential for validating AI-generated analysis, and the unique complexities of each COBOL codebase mean that full automation is not yet feasible. Nevertheless, AI has already transformed legacy modernization from a daunting task into a manageable project.
Open-Source Framework for COBOL Modernization
GitHub and Microsoft have open-sourced a framework built with Microsoft Semantic Kernel, offering a practical starting point for organizations looking to modernize their legacy systems. This framework includes multiple specialized agents that handle various aspects of the modernization process, from dependency mapping to cost tracking.
The framework is designed to keep organizations in control of their intellectual property while allowing internal teams to learn and adapt to new business logic as modernization progresses.
For more detailed insights, visit the full article on [GitHub](https://github.blog/ai-and-ml/github-copilot/how-github-copilot-and-ai-agents-are-saving-legacy-systems/).
Image source: Shutterstock
