AI Performance

Monitor how well your AI agent handles conversations.

Metrics

  • Confidence Score: Average confidence across all responses.
  • Escalation Rate: Percentage of conversations handed to humans.
  • Response Time: Average time to generate a response.
  • Token Usage: AI model tokens consumed this billing period.

Improving Performance

  • Add more knowledge sources to cover gaps.
  • Review low-confidence responses and create corrections.
  • Refine the persona instructions for edge cases.
  • Monitor the Knowledge Gaps section for unanswered topics.