JetBrains research (AI Pulse, Jan 2026) reveals that 73% of organizations don't use AI in CI/CD pipelines at all, despite widespread AI adoption in development workflows. The core reason isn't technical: it's about trust, unclear value, and the higher cost of errors in validation systems. The post analyzes where AI does work today in CI/CD (failure diagnosis, security fixes, test optimization), introduces a four-stage maturity model from no AI to agentic workflows, and argues that as AI agents generate more code changes, CI/CD becomes the critical trust and governance layer that determines what reaches production.
Table of contents
AI is everywhere except in CI/CDHow AI is used in software development todayWhy AI adoption in CI/CD is still lowCI/CD is an evidence system, not just automationWhere AI actually works in CI/CD todayFrom copilots to agents: What’s changingThe maturity model of AI in CI/CDWhat this means for CI/CD systemsConclusion: AI in DevOps needs a trust layerSort: