AI Glossary
Agentic workflow
agentic workflow, agent workflow
An agentic workflow is a process in which an AI agent carries out a task in defined steps — planning, using tools, checking the result and adjusting its approach — instead of returning a single answer.
- The task is split into repeatable steps, not a single call.
- It combines planning, tool use and result-checking.
- It works better with clear boundaries and checkpoints.
A single question to a model is one call. An agentic workflow breaks the task into stages: the agent plans, performs a step, evaluates the result, and decides whether to continue or revise its approach. Some steps can be fixed in advance, others left to the agent's judgment.
This kind of flow shines on multi-step tasks — gathering data, processing it and preparing a report, for example. The clearer the boundaries of the tools and the criteria for success, the easier it is to steer the run and stop it when something goes wrong.
Related terms
In guides
Related articles