TLDR:<p>This intriguing paper delves into the creation of autonomous agents through the emulation of human brain functions, perception, and actions. It introduces a conceptual framework for LLM-based agents consisting of three core components:<p>1. The “brain” module, acting as the central controller, handles fundamental tasks such as memory retention, cognitive processes, and decision-making.<p>2. The “perception” module is responsible for sensing and processing diverse information from the external environment, encompassing various sensory inputs.<p>3. The “action” module executes tasks using tools and exerts influence over the agent’s surroundings, enabling it to interact effectively with its environment.