Agentic AI
Agentic AI describes systems where a language model runs in a loop — choosing actions, calling tools, and observing results — to pursue a goal, rather than answering in a single pass.
The distinction from a chatbot is the loop. A chat model responds once; an agent takes an action, sees what happened, and decides what to do next, repeating until it reaches an answer or a stopping condition. That ability to act, observe, and re-plan is what the word 'agentic' points to.
Agentic systems are defined by four parts: the model that reasons, the tools it can call, the context it carries between steps, and the stopping rule that ends the loop. Their power and their cost both come from the same place — each step re-sends the accumulated context, so agent cost grows faster than the number of steps.
Learn this stuff by email
New definitions, tools, and explainers when I publish them. No spam, unsubscribe anytime.