What is an AI agent?

An AI agent is a program that can plan a sequence of actions toward a goal and carry them out with some autonomy, deciding what to do next based on what happens, rather than following a fixed script. That's the difference from a plain chatbot: an agent doesn't just answer, it acts, reading files, calling APIs, running code, and adjusting its plan based on the results.

Agent vs. assistant. An assistant (like a chat-based helper) mostly responds to what you ask. An agent pursues a goal across multiple steps, often without a human approving each one. Claude Code is a good example: given a bug report, it inspects the repository, forms a hypothesis, edits the relevant files, runs the tests, and iterates, not a single question-and-answer exchange.

Agent vs. automation. A traditional automation (a Zapier zap, a cron job) follows a fixed set of steps every time. An agent makes judgment calls: which file to edit, which search result is relevant, when to ask for help instead of guessing. That judgment is also what makes agents harder to fully trust with high-stakes actions unsupervised.

What agents are actually used for today. The clearest wins are in domains with a well-defined goal and a way to check the work: writing and testing code, drafting and sending support responses, researching a prospect before outreach, triaging a support queue. Browse the agent directory to see agents organized by what they're built to do, or read agents vs. workflows vs. skills for how agents relate to the rest of the ecosystem.