Skip to content

Auto QA (Insight Recipe)

Auto QA evaluates conversations using your QA scorecard and produces: - an overall QA score (and optionally section/question scores) - evidence-based explanations for each evaluation

Where Auto QA appears

  • Conversation Details (QA results)
  • QA dashboards (score distribution and trends)
  • Search filters (e.g., QA score < threshold)

Configuration checklist

  1. Create or verify your QA Scorecard (sections/questions)
  2. Enable/configure the Auto QA AI Task
  3. Configure filters (which conversations to evaluate)
  4. Test in Playground and calibrate against human QA
  5. Roll out dashboards and workflows

See: - Auto QA (Scorecards and Tasks) (configuration) - Playground and Validation Checklist (testing)

Auto QA Dashboard

Auto QA Dashboard - High Level

Figure: Auto QA Dashboard showing overall score distribution and trends.

Auto QA Report View

Auto QA results are displayed in a dedicated QA tab on conversation details, showing:

  • Overall score
  • Section-by-section breakdown
  • Per-question results with explanations

Auto QA Report with Transcript

Figure: Auto QA report showing detailed results alongside the conversation transcript.

Users can provide feedback on Auto QA results if they disagree with the scoring:

Auto QA Report Feedback

Figure: Feedback option for flagging incorrect results.