Thoughtworks Technology Podcast Unlearning, experimentation and engineering rigor in an agentic world
88 snips
Feb 5, 2026 
Patrick Debois, DevRel and DevOps platform engineer focused on practical AI adoption and org change. Nathen Harvey, DORA lead at Google Cloud researching delivery performance and AI’s team impact. They discuss how AI amplifies team strengths, shifting roles from producer to agent manager, treating prompts and specs as versioned artifacts, and the rising importance of knowledge and context infrastructure.
AI Snips
Chapters
Transcript
Episode notes
AI Amplifies Existing Team Dynamics
- AI acts as an amplifier: it improves well-functioning teams and magnifies dysfunction in broken ones.
- Adding AI without fixing bottlenecks (e.g., slow reviews) increases pain, not productivity.
Progressive Rollouts And AI Tutoring
- Roll changes out progressively to reduce risk and gather feedback before broad release.
- Use AI as a tutor to accelerate learning about new languages, frameworks, or domains.
Context Over Code
- Specs and context become the locus of engineering as agents act on intent.
- Current model context limits make iterative, focused context disclosure necessary.