Claude Code creator Boris Cherny pioneered the concept of "loops" in Agentic AI, where agents will continuously prompt other agents to get the job done. Cherny notes this as a major evolutionary step comparable to the shift from manual coding to AI agents. These are recursive systems that call each other until they're finished with their tasks. An example is the "Ralph Loop", which allows models to correct themselves and keep redoing tasks like code architecture until the goal is met. This effectively automates complex workflows in the background, that wouldn't be plausible with a single AI agent. While it promises staggering efficiency through relentless compute usage, it also comes with skyrocketing costs because how fast tokens are consumed without any hard cap on spending. The push for agentic loops is already reshaping the workforce: tech giants such as Oracle, Meta, and Microsoft say that AI efficiency as a primary driver for layoffs. Ultimately, the industry is betting that swarms of AI agents will outweigh upcoming financial risks, and it is fundamentally changing the workforce.