In the field of artificial intelligence, prompt engineering was long considered the core skill for interacting with models. However, a new wave is sweeping the industry—loop engineering is emerging as the next big focus. Boris Cherny, the creator of Claude Code, recently revealed that he has stopped writing prompts altogether. In an interview with CNBC, Cherny stated: “This is an agent that can prompt Claude by itself—I no longer write prompts.” He further noted that loops and similar features may be among the achievements he is most proud of over the next decade.
Cherny is not alone. Peter Steinberger, an OpenAI engineer and creator of the viral project OpenClaw, reminded users on X: “This is your monthly reminder: you should no longer be writing prompts for coding agents.” Claire Vo, founder of ChatPRD and host of the “How I AI” podcast, summed it up more succinctly: “It’s really a reminder that you don’t need to use human fingers to type a prompt—your agent can do the work for you.” The core idea behind loop engineering is to build repeatable structures that allow AI agents to interact autonomously across multi-step tasks, significantly reducing the need for human intervention.
