AI DevOps Podcast Damian Brady: GitHub AI - Episode 389
6 snips
Feb 16, 2026 Damian Brady, Staff Developer Advocate at GitHub with decades in software, AI, DevOps and community leadership. He discusses expanding Copilot beyond autocompletion to reviews, fixes, and lifecycle help. He talks TDD with AI, assigning work to Copilot, Copilot SDK/CLI and SpecKit for structured AI-driven specs. He also covers documentation living with code and integrating lifecycle tools.
AI Snips
Chapters
Transcript
Episode notes
Copilot Expands Across The Development Lifecycle
- GitHub is expanding Copilot beyond code completion into reviews, vulnerability fixes, and lifecycle tasks.
- They focus on tuned processes and evaluation to make AI outputs trustworthy before surfacing them to users.
Use TDD For AI-Assisted Functions
- Use TDD when asking AI to implement specific functions to keep behavior well-defined.
- Ask Copilot to write tests first, review edge cases, then generate code and iterate until tests pass.
Low Barrier Raises Risk And Opportunity
- AI lowers the barrier so many non-experts can produce working software quickly, creating both opportunity and risk.
- For small, low-risk projects that's fine, but critical systems still need rigorous engineering and safety standards.