Agents & Automation
Agent Loop
Definition
The repeated cycle in which an agent receives context, chooses a next action, uses a tool or produces output, observes the result, and decides whether to continue.
Why it matters for your business
Explains why one agent task can involve many model calls and why stopping conditions matter for cost and reliability.
In practice
An agent searches for a policy, reads a passage, checks whether it answers the question, then either searches again or prepares a reply.
How it differs from related terms
A fixed workflow follows predefined steps. An agent loop can choose its next step based on intermediate results, within the limits set by its harness.