What is an AI agent?
An AI agent is software that uses a language model to choose its own next steps, call tools like APIs and databases, and work toward a goal with limited supervision.
In more detail
A normal script follows a fixed path. An agent reads the situation, picks an action, checks the result and decides what to do next. The model supplies the judgment. The code around it supplies the tools it may use, the limits on what it may change and the point where a person must approve. Without those limits you have a demo, not a business system. Most useful agents are narrow: one job, a short tool list, a clear definition of done. The wide, do-anything agent is still unreliable on long multi-step work, and we rarely recommend one for production.
Example
A collections agent at a distributor reads the overdue-invoice list, looks up each customer's payment history in the ERP, drafts a reminder in the tone that fits that account and queues it for the credit controller. Anything above a set balance goes to a person first. The agent never edits the ledger itself.