An Agentic Loop is the repeated cycle through which an AI agent observes a situation, reasons about what to do, acts, and uses feedback to continue or stop. The loop continues until a goal is achieved, a failure condition occurs, or another termination rule is met.