Unlocking CI Performance in the Age of Agentic Development
summary
Agentic commits multiply PRs and expose CI waste: GitHub reports 46% AI-generated code and 1.2M Copilot PRs/month, pipeline overhead often hits 58% of run time. Concrete wins include SeatGeek cutting queue time 3 minutes to 4 seconds by auto‑scaling CI in Kubernetes, and an infra fix dropping startup time 30–45s to 5s. A CI Time Budget Analyzer that polls the GitHub API surfaces P95 spikes (example: 7m21s average run, Next.js spikes to 37m with 50% success) and shows caching and test‑sharding pay back immediately.
AI-written summary · about
chapters (10)
- 0:00CI bottleneck
- 1:55AI agents in CI pipelines
- 3:19CI performance killers
- 5:27Queue time and startup tax
- 8:59CI Time Budget analysis
- 13:17Baselines and prioritizing fixes
- 15:45Fixes for CI slowdowns
- 18:54Auto-scaling runners and testing
- 21:43Optimizing CI feedback loops
- 22:45Actionable steps and closing
related talks
21mAI-driven optimization of CI/CD pipelines
12mThe Synergistic Impact of Artificial Intelligence on DevOps
14m50 shades of PR dark matter
50mFast Doesn’t Mean Fragile: Delivering AI-Powered Software at Scale
06mWith great AI power comes great responsibility
25mMaximize CI/CD efficiency using AI and a vertically integrated stack
44mSpeed Is Easy, Trust Is Hard: AI's Real Impact on Developer Productivity
50mHow GitHub operationalizes AI for team collaboration and productivity