Usage analytics

Reading the dashboard activity timeline, volume chart, and per-key cost breakdown to answer what happened and what it cost.

The Activity page in the dashboard is where usage data becomes an answer, not just a log. It combines a timeline, a volume chart, and a breakdown by key so you can go from "something changed" to "here's exactly what and where" without exporting a CSV first.

What's on the page

• Timeline — individual requests in order, each with model, cost, latency, and status. Use it to inspect a specific incident or spot-check a specific agent's recent calls.
• Volume chart — request count and cost over time, so a spike is visible as a shape, not just a number in a table.
• Key breakdown — usage grouped by API key, which is how you answer "what did this agent or this integration cost" without cross-referencing request ids by hand (see Agent fleet on Relixr for why per-key attribution is the unit that matters once you have more than one caller).
• Filters — narrow by date range, model, key, or status, so the timeline and chart both reflect only what you're actually investigating.

Reading it in practice

• A cost spike: check the volume chart for when it started, then filter the timeline to that window and group by key — a spike is almost always one key or one model, not a uniform increase.
• An agent behaving unexpectedly: filter to that agent's key and read the timeline directly. Combined with a per-key spend cap, this is usually enough to catch a runaway loop before it becomes a billing surprise.
• Reconciling with an invoice: the same cost figures shown here are what settle against your prepaid balance — see Billing & credits for how reservation and settlement work.

Programmatic access

Everything visible on the Activity page is also available through the management API for teams that want to pull it into their own dashboards or alerting — see Management API for the usage and keys endpoints.