TutorialsVideo tutorials

Introducing Judges: Enhancing AI Response Quality Monitoring

Published March 2, 2026

Portrait of Paul Loeb.

by Paul Loeb

Overview

Judges are a LaunchDarkly capability that automatically evaluates the responses generated by AI models in your applications. Judges score outputs based on metrics like relevance, accuracy, and toxicity, allowing you to monitor quality over time.

AI Configs are called AgentControl configs

LaunchDarkly AI Configs have been rebranded as AgentControl and AgentControl configs. Some references in this topic may be outdated.

In this video, you will learn how to:

  • Attach multiple judges to a config
  • Customize judges to fit your business needs
  • Control costs by adjusting sampling percentages in different environments
  • Track metrics to detect regressions and compare variations

Introducing Judges

Learn more