Glossary

What is agentic AI?

Agentic AI is AI that carries out a task in several steps on its own, instead of only answering a question. It decides what to do next, uses other software such as search, email or a company database, checks the result, and keeps going until the task is done. A system built this way is called an AI agent. Vendors also use the word loosely, so a product sold as agentic may do a lot on its own or very little.

Why it matters for a business

The shift is from AI that writes to AI that acts. A model that only writes text is cheap to be wrong, because a person reads the text and discards what is off. A system that sends, books and updates can do real damage when it is wrong, and small error rates multiply across steps: a step that goes right 95% of the time goes right about 60% of the time over 10 steps in a row. The risk is set by what the system is connected to and where a person approves the step that matters, far more than by how capable the model is.