Thoughtworks Technology Podcast Accelerating mainframe modernization using generative AI
12 snips
May 15, 2025 Rob Mee, CEO of Mechanical Orchard, and Rachel Laycock, CTO of Thoughtworks, discuss the challenges of modernizing mainframe systems using generative AI. They explore how their collaboration has led to innovative tools that simplify the complexity of legacy codebases. The conversation highlights the gradual approach required for effective modernization and the importance of skilled developers. They also emphasize the role of generative AI in identifying crucial data flows and creating reliable new code, prioritizing quality and maintainability throughout the process.
AI Snips
Chapters
Transcript
Episode notes
Essential vs Incidental Complexity
- Mainframe modernization has essential complexity in adapting old code and platforms and incidental complexity in gaining system access due to security and organizational barriers.
- Addressing both complexities is crucial for successful and scalable mainframe modernization projects.
Value-Driven Incremental Modernization
- Incremental approaches focusing on porting only valuable, used portions of code reduce risk and workload in mainframe modernization.
- Using generative AI to analyze code and data flows identifies important components and improves modernization efficiency.
Data-Driven Verification Approach
- Treat the legacy system as a specification and capture input-output data flows to verify new code behavior.
- Use generative AI iteratively with unit tests to ensure migrated code matches legacy behavior before production deployment.