An autonomous agent is an AI agent that operates with minimal human intervention over extended tasks: it sets sub-goals, chooses tools, recovers from errors, and decides when the job is done. Full autonomy is powerful but risky, so in practice businesses set boundaries, budgets, and approval checkpoints. Think of an agent monitoring competitor prices for an online electronics store around the clock, adjusting listings within rules the owner defined, and only escalating unusual situations to a human.