Boris Cherny, creator of Claude Code, says loops are a crucial step in agentic AI development. Two years ago, agents started writing code, and now they’re prompting each other to create code. This loop is key to agentic AI, allowing agents to continuously work in the background. While not entirely new, recursive loops are a natural progression in AI development, enabling models to handle real work. With models improving fast, this could be the next step in AI’s capabilities. Why it matters: The emergence of agentic loops marks a significant shift towards AI autonomy, raising questions about the limits of trust in AI systems.