Organization dashboard

The default landing view in Insights: KPIs and a ranked list of every agent in your organization.

The Organization dashboard is the Overview entry in the sidebar at the organization level. It rolls every agent in the org into a single fleet-level read, the top-of-funnel view of how your whole foundation is performing, from adoption and cost to quality. Where this page shows how adoption is going, Engage is how you drive it.

At a glance

The page has four blocks, top to bottom:

1. Header with the shared date filter and a three-dot menu.
2. Primary KPIs: Active Users, Interactions, Average Rating, Evaluated Agents.
3. Secondary KPIs: Total Cost, Tokens Used, Carbon Footprint.
4. Top Agents by Usage ranked list and the All Agents grid.

On your first visit, a guided tour starts automatically and walks you through the KPIs, the cost and usage block, and the agents grid. You can skip it at any point, and it won’t reappear once completed.

Reading history across the counting change. Days recorded before this change still carry the old LLM-request counter and are not recomputed. On those days, Messages per day and Top Agents stay higher than the message total shown elsewhere, so a period spanning the change mixes the two units. Compare like-for-like periods after the change rather than across it.

The date filter

Every view reads the same window. The date filter sits in the header and offers three presets, 7 days, 30 days and 90 days (30 days by default), plus a custom range you pick day by day in the calendar.

The selection is shared by the organization Overview, the agent Overview and the Conversations list, so moving from the fleet view to one agent and then to its conversations keeps the same period instead of resetting it.

The calendar only offers the last 90 days: the analytics endpoints read hourly pre-aggregates and stop at roughly that depth, so a wider range was already being truncated silently.

The window is not applied to the same timestamp everywhere. Conversations filters on last_message_at (when the conversation was last active, which is also its sort key and the date shown in the table), while the agent Overview and the insights analytics filter on analyzed_at (when the analysis ran). An old conversation analyzed yesterday therefore counts towards “Last 7 days” on the agent Overview, yet is absent from the Conversations list for that same window. Each screen stays consistent with the date it displays.

The KPIs explained

Active Users

Distinct users who interacted with at least one agent in the period. The card also shows the new users added in the period and the daily active count, plus a trend badge against the previous period of the same length.

Interactions

Total user-agent interactions in the period. The card breaks it down to per-day and per-user averages, with a trend badge.

Interactions, Messages per day and Top Agents count distinct messages. They previously counted LLM requests, which is a larger number: a single message can trigger several model calls (tool calls, title generation, summaries).

Average Rating

The average user rating across all rated conversations, plus the total count of ratings. A dash (-) means no ratings have been collected yet.

Evaluated Agents

Number of agents that have at least one analyzed conversation, over the total number of published agents. A low ratio means most of your fleet is not being evaluated, likely because the agents are too new or because analysis hasn’t run yet.

Total Cost

Aggregate model spend in the period (USD). Trends against the previous period.

Tokens Used

Total input + output tokens across all agents in the period.

Carbon Footprint

Estimated CO₂-equivalent emissions for the period in kg, derived from the model usage.

Top Agents by Usage

The five agents with the highest interaction count, with a horizontal bar showing each agent’s share of the leader’s volume. Click a row to jump straight into that agent’s per-agent overview.

All Agents

A grid of every published agent your account can see. Each card surfaces the agent’s name, category, and a quick read on its analyzed volume and average score. Click any card to enter the agent view.

Recalculate metrics

The KPIs on this page are computed from a daily aggregation. After a large bulk analysis run, the dashboard may show stale numbers until the next aggregation cycle.

The three-dot menu in the top-right exposes Recalculate metrics; use it to force the rollup immediately. The action requires platform-admin authentication (verified through Governe). Without that role the request is rejected at the API layer.

What this dashboard doesn’t show

• Per-agent quality detail. Use Agent overview instead.
• User feedback. Likes, dislikes, and dislike categories live on Feedback.
• Adoption segmentation. Power / Regular / Basic users live on Adoption.
• Memory and topic trends. See Memories and Topics.

Where to go next

Drill into one agent

Click any agent card to enter its scoped view.

Visualize the fleet

See agents, tools, and categories as an interactive network.