Monitoring is the fleet-wide dashboard across the agents you've registered under AI Governance, giving you one place to watch how your governed agents are behaving as a whole.
Overview
Monitoring shows activity across the agents you've registered, rather than one at a time. To open it, select AI Governance in the side-nav and choose Monitoring under AI Observability on the launch pad.
Note: It is agent-shaped: registered models don't have a Monitoring tab of their own, and their activity isn't part of these fleet metrics — check a model's Overview and Spans instead.
What Monitoring Shows
Monitoring presents fleet-wide agent activity across five families of metrics:
Family | What's covered |
|---|---|
Traces | Daily volume, successes vs. errors, error rate, latency percentiles. |
Run types | Runs, failures, and latency per run type. |
Cost & tokens | Input/output tokens, costs, and average cost per trace. |
LLM calls | Call volume and latency, with a per-model breakdown. |
Tool usage | Runs, failures, and latency per tool. |
The dashboard is a configurable page rather than a fixed screen, so the exact charts and filters you see depend on how your workspace's monitoring dashboard is set up.
Note: If that dashboard hasn't been published for your workspace, the page appears blank below the breadcrumb. Use it as your starting point, then drill into a specific agent for detail.
Fleet-wide vs. an Agent's Own Monitoring Tab
They use the same dashboard, pointed at different scopes:
Monitoring on the AI Governance launch pad shows every registered agent together — the whole fleet (models aren't part of it).
A single agent also has its own Monitoring tab on its detail page, which shows the same dashboard filtered to just that agent.
Start at fleet Monitoring for the big picture, and open an agent's Monitoring tab when you need to focus on one.
Note: The dashboard is a workspace-deployed page — if it's blank or you need it changed, that's a workspace-admin task, not a per-user setting.