Skip to main content

The Dashboard & Cost Analytics

Two pages give you the bird's-eye view: the Dashboard (operational health) and Cost and Analytics (spend). They answer different questions — "is everything running?" versus "what are we spending?" — and it's worth knowing which is which.

The Dashboard​

The dashboard is your system-health landing page. At the top sit KPI cards: total spend to date, workflow health (success/failure behaviour), money spent on failed workflows, and one completion card per plugin — completed entities out of total (each plugin defines what "completed" means for its own entities, so the core can't compute it centrally). Below the KPIs:

  • an in-progress / queued table showing entity, workflow, and status with elapsed runtime inline — click a row to open its execution log;
  • a recent errors list — the last ten failed workflows, with entity, workflow, error, and cost;
  • a global execution list of all runs in chronological order, filterable by status.

It's the page you glance at to know whether the pipeline is healthy right now, and to jump straight into anything that needs attention.

Cost and Analytics​

Cost and Analytics is the business-facing spend report. You filter it by timeframe (at least all-time, current month, and last 30 days) and it breaks expenditure down several ways: by model, by agent, by workflow, internal operations, and infrastructure/service usage. It shows lifetime total and, separately, current-month spend.

Two honest caveats worth internalising: figures are in USD only, and the report tracks in-app costs — the model and service usage Circus knows about. Some external infrastructure (for example, cloud storage such as S3) is not tracked here, so this is a clear and useful view of your AI spend, not a complete ledger of every infrastructure dollar. For safety, it's still wise to set spend limits on the provider side too.