Skip to main content
The Endpoint Explorer is Origin’s central observability dashboard. It gives you a fleet-wide summary of AI activity across every reporting endpoint, organized by agent and model over a time window you choose. From a single screen you can gauge overall AI usage volume, drill into individual agents, follow request traces, and spot unusual activity patterns — without leaving the console.

​
Left Panel

The left panel scopes the entire view and lets you navigate between agents.
  • Date range picker — sets the time window for all data displayed on the page. Click to select a preset window or type a custom start and end date.
  • Overview — returns you to the fleet-wide summary from any agent-specific view.
  • Agents — shows every AI agent detected during the selected period. The section header displays a total agent count. Each entry includes the agent name, a sparkline summarizing its activity across the selected range, and a total request count for the period. Use Filter agents to search the list by name, and use Sort to reorder entries — for example, by Most active — so the highest-volume agents surface first.
Agents that Origin cannot identify are grouped together under Unknown Agent. If you see a large Unknown Agent volume, it may indicate new tooling in your fleet that has not yet been classified.

​
Right Panel — Overview

When you select Overview in the left panel, the right panel shows three headline metrics for the chosen time range. These three numbers give you an at-a-glance pulse check on fleet-wide AI consumption before you dive into charts or traces.

​
Activity Surface Chart

Below the headline metrics, the Overview activity chart plots fleet activity over time. Use the time-range buttons in the top-right corner of the chart to change the visible window: Switch between the five tabs to view different dimensions of the same period:
Use the Token volume tab with the hover tooltip to identify which specific model versions are consuming the most tokens. This is especially useful when you are evaluating model consolidation or tracking spend by provider.

​
Bottom Panel

The bottom of the Endpoint Explorer surfaces two log views that let you move from aggregate trends to individual events.
  • Trace Log — a chronological record of individual AI requests captured by the fleet. Each entry represents a single request and shows the endpoint, agent, model, and prompt metadata, giving you the granular detail you need to investigate a specific interaction.
  • Execution Log — a record of agent-level execution events. Use this to understand the sequence of actions an agent took during a session, beyond individual prompt-response pairs.
The Trace Log and Execution Log both respect the date range and agent filters set in the left panel, so what you see is always scoped to the context you have selected.