AI agent glossary
Evaluator–optimizer loop
A pattern where one agent produces work and a second agent critiques it against explicit criteria and sends it back until it passes or stops improving.
The generator writes a draft. The evaluator, sometimes called the critic, checks it against a clear standard and returns specific, actionable feedback. The generator revises, and the loop repeats.
It works best when you can state what good looks like and the first draft is usually close but not quite right: writing against a style guide, code that must pass checks, translations with terminology rules. It works poorly when the criteria are vague, because the critic either approves everything or nitpicks forever.
Always cap the number of rounds, stop when the score stops improving, and measure whether the loop beats a single, better-prompted attempt, since each round costs another full generation.
A writer agent drafts a product description; a critic checks length, banned phrases, and that every claim appears in the spec, and sends back a list of fixes. Most drafts pass by the second round.
Learn more in the course
Related terms
Go from definitions to doing
Learn to use and build AI agents in 130 practical lessons.