The Overview
Every team's default report: runs, errors, cost, latency, usage and eval results, ready without building anything.
The Overview (Observe → Overview) is a report every team has from the start. It answers "how are my agents doing right now" without building anything.
Tabs
| Tab | Shows |
|---|---|
| Overview | Total runs, failure and error rates, volume and errors over time, latency (average and p95), satisfaction against rage, failing sub-agents |
| Cost | Total and average cost per run, cost over time by agent and by model, cost by span, top users by cost, sub-agent cost by root agent |
| Latency | Run latency percentiles, p95 by agent, model and span, model speed, max latency by user |
| Usage | Tokens, unique users and spans; tokens over time by model and agent; span volume; top users; sub-agents by caller |
| Evals | Pass rate, verdicts, failures and eval spend; pass rate by eval, agent and what is judged; why verdicts failed; time to verdict; slowest and erroring eval steps |
The last bucket of a trend is marked as unfinished, so a partial hour never reads as a drop.
Make it yours
The Overview is an ordinary report: change the time window and filters, turn on auto-refresh, and, as its owner or an editor, add, move and remove cards. Reset to default in its menu puts the original tabs and cards back.
It isn't listed with your other reports, and can't be deleted.
Next
- Reports for your own dashboards
- Observability for the traces behind every card