Autonomous Infrastructure: Building Safe AI Agents for DevOps and MLOps Operations

DevOps and MLOps agents operate where mistakes directly impact production systems, outages, and cloud costs—fundamentally different from AI applications in research or product development. This chapter bridges the gap between automation potential and the mandatory safety guardrails required to prevent costly failures. Understanding layered autonomy, policy enforcement, and signal correlation is critical for organizations planning to delegate real infrastructure decisions to autonomous systems.