Reviewing quality and safety scores

  • Release version: Zurich
  • Updated April 3, 2026
  • 1 minute to read
  • Determine whether your AI systems are meeting quality and safety targets by reviewing scores and evaluation trends.

    Reviewing scores workflow

    When evaluations are active and sessions have been scored, start with the monitoring overview to assess performance across your AI portfolio. The overview provides score cards, AI-generated insights, trend data, and ranked lists of AI systems so you can identify which systems need attention.

    When you identify a system that needs closer review, select the Monitor tab on the AI system's asset page to see scores, metric breakdowns, trends, and evaluated sessions scoped to that system alone. See Monitoring an AI system.