Glossary entry

Reinforcement learning

Learning behavior through actions, feedback, and rewards in an environment.

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.