An AI agent is defined as a system that allows a language model to autonomously decide how many times to run processes until it finds a resolution, rather than following a predefined path.

As AI models improve, agents are becoming more prevalent and capable, prompting the need for formal definitions to clarify their use.

The concept of agents includes both those that replace human tasks and those that operate as low-level system processes, complicating the definition further.

The distinction between agents and functions is blurred, as both can involve multiple functions with language models in the middle, making them appear similar from an external perspective.

The definition of an agent may evolve as we explore the capabilities of AI, particularly in how they interact with users and perform tasks.

AI agents are expected to handle most mundane tasks, significantly lowering barriers for developers and making complex tasks like coding more accessible to everyone.

AI agents can navigate multiple systems, understand context, and execute tasks that previously required human intervention, marking a significant shift in back-office operations.

All agents share a common element: reasoning and decision-making, which can be likened to a multi-step language model chain with a dynamic decision tree.

AI agents can perform tasks autonomously, working indefinitely until they achieve a goal, which raises concerns about their potential for abuse.