Refactoring Podcast Navigating AI Development Workflows 🛠— with Birgitta Böckeler
24 snips
Sep 26, 2025 In this engaging discussion, Birgitta Böckeler, a distinguished engineer and global lead for AI-assisted software delivery at ThoughtWorks, delves into optimizing AI in development workflows. She shares insights on preferred AI workflows and the pros of using diverse tools. Birgitta highlights how successful teams combine critical thinking and experimentation. The impact of AI on both junior and senior engineers is explored, with a focus on mentorship and adapting learning strategies. Additionally, she discusses risk management and the role of documentation in AI integration.
AI Snips
Chapters
Books
Transcript
Episode notes
Non-Determinism Is A Core Property
- Large language models are inherently non-deterministic and will sometimes 'hallucinate' by design.
- This non-zero probability of unexpected outputs is a core property teams must accept and manage.
A Week Of Experiments And Briefings
- Birgitta balances hands-on experiments, internal sharing, and executive briefings in her role.
- This week she ran Cloud Code in GitHub Actions and explored reviewing large generated change sets.
Choose IDE Or Terminal By Need
- Prefer IDE-integrated assistants when you need context, visual tools, and linting to stay in control.
- Use terminal or Cloud Code when you value keyboard speed and agentic workflows, but weigh trade-offs carefully.
