Glossary
Glossary.
The terms behind training and evaluating agents that make commercial decisions, defined plainly.
- Agent benchmarkA fixed, repeatable set of tasks and a scoring rule used to measure how well AI agents complete multi-step work, so different agents can be compared on equal terms.Read the definition →
- Agentic commerceCommerce in which an AI agent acts on a person's behalf: searching, comparing, deciding and buying, rather than only recommending.Read the definition →
- Reinforcement learning environmentThe system an agent acts in during training or evaluation. It presents a situation, accepts the agent's actions, updates its state and reports what happened.Read the definition →
- Reward hackingWhen an agent earns a high score by exploiting how it is measured rather than by doing the task the score was meant to reward.Read the definition →
- TrajectoryThe full record of one agent attempt at a task, step by step: what the agent saw, what it did and what happened as a result.Read the definition →
Looking for how ORO Bench uses words like task, world and episode? See the ORO Bench glossary in the docs.