CHAT_DASHBOARD_API_TOKEN.
Reload the page to re-enter it.
Overview
Volume & performance
AI quality
Historical Conversations
Derived from stored conversation records — not live instrumentation
These figures come from conversations / conversation_messages, which predate
the chat monitoring instrumentation above and survive in any historical snapshot. Response latency here
is an approximation — the time between a user's message and the assistant's next reply
in the same conversation — not the measured value chat_turn_metrics.latency_ms records for
live traffic. Tokens, cost, cache ratio, tool-call accuracy and every AI-quality score are not derivable
from this data at any effort level; see the Overview and Tool-Call Monitoring sections for those once
live instrumentation is running.
Conversations by category
| Category | Conversations | Share |
|---|
Analytics
Tool-Call Monitoring
Judge verdicts on tool selection, arguments and result use · live trafficRecent Tool-Call Judgments
| Source | Verdict | Judge model | Correct / evaluated | Findings | Time |
|---|
Alerts
No alerting rules configured
Alerts will appear here once thresholds are defined for latency, cost or
tool-call accuracy.