Usage in the dashboard sidebar shows what the agents have actually been doing. Everything on the page respects the range picked at the top — 7 days, 30 days or 90 days — and the range is part of the page address, so a view can be bookmarked or shared.
The headline numbers
Four tiles summarise the period: Total calls, Talk-time (hrs), LLM tokens and Audio minutes, each with a percentage change against the previous period of the same length.
Calls and outcomes
The Call volume chart plots calls per day across the range. The Outcomes donut buckets every call into Completed, Transferred, No answer or Failed — a quick health check on how conversations are ending. When an outcome looks wrong, the next step is to find the calls behind it and inspect one.
Per-agent and per-provider breakdowns
Calls by agent ranks the busiest agents. By provider and the Usage by technology donut split the same period by the AI models and call technologies behind the agents — useful context when weighing choices in Models & platforms.
The metered detail
Below the charts, a raw metered-usage table lists every measured component —
technology, provider, detail, unit and quantity — which is exactly what
settlement is calculated from. Export downloads those rows for the
selected range as a CSV (for example usage-30d.csv).
The call-based charts are built from up to the most recent 1,000 calls in the period — a representative picture at very high volumes; the export gives exact counts.
Volumes here, money on Billing
Everything on the Usage page is measured volume — calls, minutes, tokens — never pounds. Real charges live on Billing: the Ledger holds the line-by-line account of what each day's usage actually drew from the balance.