Glossary term
Glossary term
Agentic Systems
An agent that assesses another agent's results before those results are finalized. You can imagine one agent as manufacturing a product and a separate agent—the evaluator agent—testing that product before it is released.
Critic is a synonym for evaluator agent.
Created for this library
An LLM product team deploys an evaluator agent that scores each generated response against a rubric so low-quality answers are escalated.
A code-review pipeline uses an evaluator agent to grade pull-request suggestions before they are shown to the developer.
A research lab uses an evaluator agent in its multi-agent setup to filter outputs from a generator agent before they pass to downstream steps.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License