Fleet Operations

Fleet Operations gives owners and administrators one view across managed Nexus machines. Team managers get the same operational views, limited to the teams they manage.

The Fleet workspace separates day-to-day monitoring from machine configuration:

• Monitor shows live health, serving capacity, active work, and items that need attention.
• Insights summarizes trends, spend, savings, and recommended follow-up work.
• Machines manages fleet membership and machine details.
• Teams & policies controls who can use machines and what those machines should contain.
• Lobby walls creates read-only status displays.
• Audit records administrative changes.

Set up fleet reporting

Monitor and Insights need one reachable machine to collect fleet reports.

1. Open Fleet > Monitor.
2. Follow Set up reporting.
3. Choose a stable, always-on Nexus machine as the collector.
4. Review which machines will report to it.
5. Apply the setup and wait for the first check-in.

Keep the collector online and give it enough storage for the selected retention period. Reporting contains operational and usage aggregates, not prompts or model responses.

If you move the collector role, finish the guided move before removing the old collector. A machine still pointing at an unavailable collector cannot contribute current data.

Monitor the fleet

Monitor combines current data from the reporting collector with direct checks of reachable machines. Use it to answer:

• Which machines are online or need attention?
• What models, Smart Routes, Smart Balancers, and clusters are serving now?
• Are requests failing or blocked by policy?
• Which machines are out of team policy?
• Is the fleet short on capacity?

Live rollups can be incomplete while a machine is offline or updating. Open the affected machine from the rollup before changing configuration.

Use Insights

Insights turns retained usage into planning information, including request volume, model and provider mix, estimated spend, forecast, and Smart Route savings where enough comparable data exists.

Treat cost values as estimates unless the provider reports exact spend. Missing prices remain unpriced; Nexus does not treat them as zero-cost traffic.

Use the task list to focus on high-impact work such as unavailable capacity, missing policy items, or a route that frequently falls back. Confirm the underlying machine state before acting on a historical recommendation.

Create a lobby wall

Lobby walls are public, read-only dashboards for a TV, operations room, or event display.

1. Open Fleet > Lobby walls.
2. Create a wall and choose the whole fleet or selected machines.
3. Choose an optional expiration date and lock code.
4. Turn the wall on and copy its link.
5. Open the link on the display device.

The wall can show machine availability, requests, speed, success rate, estimated cost, active models, cluster state, and recent activity. It cannot change Nexus configuration.

Anyone with an unlocked wall link can view it. Use a lock code for shared or semi-public screens, set an expiration when appropriate, and turn the wall off when it is no longer needed.

Review the audit log

Open Fleet > Audit to review administrative events such as machine, team, policy, access, and configuration changes.

Owners and administrators can review organization-wide events. Team managers see only events scoped to teams they manage. Event scope is recorded when the change happens, so moving a machine later does not rewrite its older audit history.

Audit data explains who changed configuration and when. It does not contain provider credentials, client token values, prompts, responses, or tool payloads.

Team-manager scope

A team manager can operate the user and shared machines of the teams they manage, review those teams' audit events, and use scoped Monitor and Insights data. They cannot see another team's machines or organization-only settings such as billing, plan, SSO, or entitlements.

Shared machines borrowed from another team can be visible without granting full machine administration. Usage that cannot be safely separated by team is excluded from a team manager's scoped analytics.

See Teams for roles, machine access, and policy workflows.

When data is missing

Check these in order:

1. The reporting collector is online.
2. The affected machine is online and enrolled in the expected organization.
3. Fleet reporting is enabled on that machine.
4. The connector and Runtime are current enough for the requested report.
5. The selected time range includes completed traffic.
6. Your role includes the machine or team being viewed.

An update_required state means the machine or collector cannot enforce the current reporting contract. Update it before relying on scoped analytics.