What is AI agent?

An AI system that can plan and take multi-step actions toward a goal — using tools, calling APIs and making decisions, not just answering a single prompt.

An AI agent is a system where a model does more than answer one prompt: it works toward a goal by planning, taking actions, observing results, and deciding what to do next. The typical loop is: the model receives a task, chooses a tool (search the web, run code, read a file, call an API), gets the tool's result back, and repeats until the task is done. Coding assistants that edit files and run tests, research agents that browse and compile reports, and customer service bots that look up orders and issue refunds are all agents. Practical nuance: agents multiply both the power and the failure modes of a model. A wrong step early can compound, costs grow with every loop iteration, and giving an agent access to real systems (email, payments, production code) needs guardrails and human review points.

Example

You ask a coding agent to "fix the failing checkout test." It reads the test output, opens the relevant files, edits the bug, reruns the tests, sees them pass, and reports back, all without step-by-step instructions from you.

Why it matters

Agent capability is now a key differentiator between AI tools. When comparing products, check what actions the agent can actually take, what permissions it needs, and how much it costs per completed task rather than per message. Browse the AI tools directory or the model leaderboard to put it into practice.