The AI Native Dev - from Copilot today to AI Native Software Development tomorrow Is Your Team Ready for AI-Driven Modernization? | Birgitta Böckeler
14 snips
Sep 2, 2025 
Birgitta Böckeler, a Distinguished Engineer at ThoughtWorks and expert in AI-driven software development, shares her insights on legacy modernization. She discusses the limitations of relying solely on AI for code migration, emphasizing the need for proven engineering practices. The conversation addresses the evolving roles of developers as managers of AI workflows and the need for a disciplined approach to refactor instead of rewrite. Birgitta highlights the importance of balancing AI creativity with deterministic tools for successful legacy transitions.
AI Snips
Chapters
Transcript
Episode notes
Client With Missing Backend Source
- ThoughtWorks worked with a client who had lost backend source code and couldn't upgrade their database.
- The team used AI and forensic techniques to understand and plan a migration without original source files.
Start With Lightweight MCP Servers
- You don't need a complex agent platform to start forensic modernization; combine coding assistants with small custom MCP servers.
- Build lightweight MCP tools to query databases, timestamps, or DOM to feed agents accurate context.
Create Parity Tests For Safe Rollouts
- Build end-to-end tests or fitness functions that run against old and new apps to validate parity.
- Use AI-generated specs to accelerate writing tolerant browser tests, but validate their fidelity manually.