Reports
Please select both dates.
Steps
Log Output
Usage — Total Chats
Feature Breakdown
Quality Scores
Helpfulness P50 · Relevance P50 · Helpfulness P10
Safety & Reliability
Hallucination P90 · Toxicity P90 · Blocked sessions
User Feedback
Session Types
Feature Usage (Real Users)
System Events (Real Users)
Session Types (Real)
User Feedback (Real)
LLM Quality Scores — P50 (Real)
Median score across all evaluated traces
LLM Quality Scores — Worst-case (Real)
P10 for quality metrics · P90 for safety metrics
| Metric | Real Users | Test Users |
|---|