Agent
Agen
A model with tools, a system prompt, and a loop that lets it act and observe results.
An agent is a model wired to tools and given a loop: it decides what to do, the harness executes it, the result comes back, and it decides again. That loop is what turns text prediction into a job that can finish on its own.
The word is used loosely. What matters is the capability, not the label: a chatbot with no tools and no loop is a chat interface, while a program that calls a model, runs a search, reads the result, and tries again is an agent regardless of what the vendor calls it.
The loop is also where the failures come from. An agent that has decided badly will keep executing badly until something stops it, which is why guardrails, permission prompts, and stopping conditions are part of the definition rather than an add-on.
Related terms
Harness
Everything around the model that turns it into an agent: tools, system prompt, permissions, hooks.
Tool
A function the model is allowed to call, like reading a file or running a command.
Tool Call
The model's request naming a tool and its arguments. Structured output, not a side effect.
Agent Loop
The repeated cycle of decide, act, observe. Where autonomy actually comes from.
Human in the Loop
A person who reviews, approves, or corrects the agent's work at key points.
Want this working inside your team?
It starts with one conversation about your workflow.