For those working in DevOps or platform engineering, where are you using AI in your everyday workflows? I'm especially interested in practical, real-time examples involving CI/CD, infrastructure as code, pull request reviews, testing, monitoring, incident response, or other platform tasks. What has actually been useful in production, and how do you keep AI from making critical automation unpredictable?
2 Answers
I wouldn’t put AI in charge of deterministic deployment steps or other pipelines where the same input should always produce the same result. It’s more useful for helping build and maintain those pipelines, explaining failures, suggesting configuration changes, and reviewing pull requests. That keeps AI in an assistive role while the actual deployment logic remains predictable.
We’ve added several review-focused AI agents to the CI process, and they’ve been useful for spotting bugs that human reviewers sometimes miss. They analyze the proposed changes and report potential problems alongside the normal checks.
That sounds useful. Are the agents allowed to fail the build when they identify a serious issue, or do they only publish findings for a human to review?