A new framework defines five levels of agentic autonomy, ranging from basic tool-use to full independence. This taxonomy provides a concrete benchmark for developers to measure how much human oversight an AI requires. TLDR AI highlights this as a critical step for standardizing agentic workflows. Practitioners can now categorize their systems by specific operational autonomy.