Large Language Models (LLMs) are fundamentally stateless—they cannot proactively fetch real-time information or directly execute external operations.