Recent research has shed light on the capabilities of Large Language Model (LLM)-brained GUI agents, which leverage advanced natural language processing to enhance software interaction. These agents combine large language models with graphical user interface (GUI) automation tools, enabling users to control software through conversational commands. The study outlines the evolution of these agents, their key components, and identifies gaps that need to be addressed for future advancements. The findings suggest that LLM-brained GUI agents could fundamentally transform the way users interact with software, moving away from traditional manual operations towards more intuitive, conversation-driven interfaces.