The Bike Shed 469: How are we using AI? with Jimmy Thigpen
4 snips
Aug 12, 2025 Join Jimmy Thigpen, a developer at Thoughtbot, as he dives into the world of AI workflows. He and fellow developer Joël Quenneville discuss the real impact of LLMs on programming. They explore the balance between rapid AI assistance and the need for careful planning in software development. The duo also examines AI's limitations in creativity and flexibility, sharing insights on how to improve interactions with AI tools and maximize coding efficiency. A thought-provoking conversation for anyone navigating the evolving landscape of AI in tech!
AI Snips
Chapters
Transcript
Episode notes
Water Skiing Transfers Skills
- Jimmy describes learning water slalom skiing and compares it to snow skiing to explain transferable skills.
- He uses this story to illustrate how prior experience speeds learning in a new but related domain.
Mixed Reactions To LLMs In Teams
- Teams split between early adopters and skeptics about LLMs, creating varied experiences with features like automatic PR reviews.
- Writing tests with LLMs shows clear positive outcomes compared to noisy automated reviews.
Give Clear Prompts And Conventions
- Give the model explicit prompts and codebase conventions to improve generated tests.
- Always review generated tests because models often produce only the happy path.