AI that pursues a goal on its own, planning steps, using tools, and taking actions with little or no human instruction along the way.
A plain chatbot answers a question. An agent books the flight, files the ticket, or moves the money. That autonomy is the point, and it is also the risk: an agent that acts on a wrong conclusion causes real-world consequences, not just a bad reply. The more an agent can do without a human in the loop, the more it needs oversight of what it is actually doing.
Related: AI Agent Monitoring, Guardian Agent, Human-in-the-loop