The AI engineering world is using “loop” to describe several different agent architectures. This post maps execution loops, task loops, product loops, system loops, and the human oversight loop that controls them. The post What is a loop in AI engineering, anyway?