.NET Rocks! Making Reliable Software in 2026 with Damien Brady
Feb 18, 2026
Damien Brady, a developer and senior manager on GitHub's developer advocacy team with deep DevOps and .NET experience. He explores how generative AI is reshaping who writes code. Short iterative prompts, tests, and CI remain crucial. They discuss local LLMs for privacy, orchestrators and specialist models, and using AI as reviewers and automated PRs while balancing cost and trust.
AI Snips
Chapters
Transcript
Episode notes
Intent Becomes The New Source Code
- AI changes who 'writes' code by turning intent into generated implementations instead of fingers-on-keyboard typing.
- That raises a larger gap between developer intent and the actual code running in production.
JIT Bug That Removed A Catch Block
- Damien shared a story of a.NET JIT bug that optimized away a catch block in a specific runtime version.
- That illustrated how platform behavior can silently change and break assumptions in release builds.
Same Prompt, Different Outcomes
- AI agents are non-deterministic so identical prompts can return different results at different times.
- That unpredictability means a prompt's output can't be assumed stable without checks and versioning.