Dashboards you shape
Build rooms from big numbers, trends, distributions, breakdowns, tables, and funnels—then save the layout for the team.
Analytics
Know what every agent is doing—and whether it works.
Build the operating view your team needs from trace, score, cost, latency, audience, and environment data. Drill from a metric into the runs behind it.
One canonical source
Analytics and trace inspection read the same observability store, so a dashboard number and the evidence behind it stay aligned.
From executive pulse to run-level evidence.
Quality and reliability
Follow success, warning, and error rates alongside evaluation scores and guardrail outcomes without leaving the operating console.
Cost and token truth
Inspect spend, tokens, latency, models, bots, and environments from the same canonical traces used by the trace explorer.
Test and Production separation
Filter dashboards by environment so experiments do not silently distort production operating numbers.
Threshold alerts
Notify the team when error rate, p95 latency, cost per turn, or spend crosses a limit you define.
Shareable evidence
Create revocable read-only links and scheduled exports for stakeholders who do not need workspace access.