AI Assistant dashboard

The AI Assistant Dashboard is the operational view of how RAUM is being used across Zendesk. It combines cost, usage, quality, autonomy, activity, and reliability rather than reducing the system to one vanity metric.

​

Dashboard widgets

The current Assistant widget catalog includes: Some widgets depend on Zendesk-shaped data. For example, Usage by agent is Assistant/Zendesk-specific.

​

Change the reporting period

Use the dashboard range selector to change the period you are analyzing. The default saved timeframe is 7 days, and administrators/users with the appropriate controls can customize the saved dashboard preference. A custom ad-hoc date range is not stored as the persistent default.

​

Customize the dashboard

The Assistant dashboard supports saved preferences such as:

• grid or list layout;
• visible widgets;
• widget order;
• default timeframe;
• chart presentation; and
• whether cost information is shown.

Depending on permissions, you may be editing your own dashboard preferences or organization defaults.

​

How to read the dashboard

​

Cost without quality is incomplete

A cheaper ticket is not an improvement if it creates more incorrect answers or human recovery work. Compare Cost per ticket with Answer quality, Autonomy, and Latency & reliability.

​

Autonomy needs outcome context

A high autonomous share tells you how often RAUM handled work without an agent. It does not prove that those tickets were handled correctly. Review CSAT & Outcomes, negative feedback, and Knowledge Gaps alongside autonomy.

​

Confidence is operational

Assistant confidence affects more than a chart. It can influence Zendesk tags and Agent Mode public/internal delivery behavior. A shift in confidence distribution after changing a model, procedure, or retrieval configuration deserves investigation.

​

After a configuration change

Use the dashboard to compare before and after when you change:

• AI models;
• procedures or guidelines;
• Agent Mode scope;
• tool connections;
• caching behavior; or
• enrichment configuration.

For ticket-level evidence, switch from aggregate metrics to Activity → Ticket Activity.