Episode Summary
Martin and Adam explore how AI tools revolutionize legacy code analysis and modernization. What once took weeks of manual code archaeology now happens in hours as AI analyzes codebases, identifies patterns, and extracts the conceptual structure. They discuss using event modeling to capture legacy system behavior before rewriting, ensuring nothing gets lost in translation.
Main Discussion Points
- AI-Powered Analysis: Using tools like Claude and GPT to rapidly analyze large legacy codebases
- Conceptual Structure Extraction: How AI identifies business logic patterns buried in technical implementation
- Event Modeling Legacy Systems: Reverse-engineering event models from existing code to understand current behavior
- Modernization Strategy: Using event models as the blueprint for rewriting legacy systems
- Time Compression: Reducing legacy analysis from weeks to hours through AI assistance
- Pattern Recognition: How AI identifies recurring patterns that manual analysis might miss
Key Takeaways
AI dramatically accelerates legacy code understanding by rapidly analyzing large codebases and extracting conceptual structures. Event modeling provides an excellent target format for documenting legacy system behavior - AI can help identify the implicit workflow, business rules, and data flows buried in code. This combination enables modernization projects to quickly understand current system behavior before rewriting. The time savings are substantial: tasks requiring weeks of manual code reading now complete in hours of AI-assisted analysis, though human expertise remains essential for validating AI findings and making architectural decisions.
Memorable Quotes
- “I will never ever work in a different way. If they understood how that works they will never ever work in a different way.” - Martin (quoting workshop participants)
- “We have a statistical network to draw conclusions from previous uh findings, and you’re saying that you’re going to ignore it at the expense of someone dying.” - Adam (about AI in medical research)
- “Just focusing on code is is is a losing game. You will lose. Maybe not in the next 3 months, maybe not in the next six months, but pretty sure in the next 12 months if you focus on code, you might lose the game.” - Martin
- “You cannot out code an an encoding agent, you cannot out code an an an AI agent. They they will be faster than you no matter how how fast you are.” - Martin
- “The only thing that uh is keeping our fast our fastest developers is is the rate at which code generation is happening” - Adam
Key Learnings
- Event modeling combined with AI allows analyzing legacy systems quickly - start with high-level flows, pick one to zoom into, and go deeper until you understand how it works
- Focusing only on the technical parts of DDD (aggregates, entities, repositories) misses the point - DDD is about understanding business processes and shared language
- AI is taking away repetitive coding jobs, but developers who focus on understanding business problems and information flow will remain valuable
- The AI “bubble” criticism is premature - we’re still in the early stages, similar to criticizing websites in 1996 or calling the internet a fad
- Event modeling works as a programming language for AI because it provides structured, timeline-based specifications that AI can understand and execute
Ready to Learn More?
Explore Event Modeling and Event Sourcing in depth with our tutorials and book.
Join our Event Modeling Workshop to get hands-on experience.
Want to learn how to apply Event Modeling and Event Sourcing in practice?
Follow the Online Course “Implementing Eventsourcing” - comes with a Lifetime Event Modeling Toolkit License.