Definition
What it means
In reinforcement learning, an agent takes actions in an environment and updates its behavior based on reward signals. For AI agents, environments may simulate software, professional workflows, or the physical world.
Why it matters
Role in the system
It allows models to improve multi-step behavior where a single static answer is not enough.
Example
In practice
An agent practices resolving a support case across email, a CRM, and a billing tool.