Governance Panel

The governance panel shows what happened “under the hood” when SUPERWISE Chat generated a response. Use it to understand how the AI arrived at its answer, what rules were checked, and what sources were used.

Opening the Panel

Click the shield icon to the right of the conversation. The panel opens as a side drawer.

Note: The governance panel requires at least one completed response in the conversation before data is available.

Steps Tab

Shows the execution timeline for the most recent response:

• Runtime trace: A step-by-step view of each processing phase, what ran, and how long it took
• Thinking steps: If the AI used multi-step reasoning, you’ll see each step with confidence scores and latency
• Token spend: A breakdown of how many tokens were used per phase

Rules Tab

Shows the guardrails and configuration that applied to this conversation:

• Configuration: Which assistant, persona, and features (web search, thinking) are active
• Guardian rules: The guardrail rules that were evaluated, with their severity (enforced, warning, or info)
• Safety policies: Policy enforcement status (on/off) and values
• Capabilities: What the AI agent is allowed to do (knowledge search, export, etc.)
• Limitations: Active limits (max tokens, budget caps, etc.)

Sources Tab

Shows what sources the AI retrieved and how they were filtered:

• Search transparency: How many results were fetched, how many survived deduplication and relevance filtering, and the score threshold used
• Knowledge retrievals: Each source retrieved, with its relevance score as a percentage
• External research: Whether web or external sources were consulted

RBAC Audit Trail

All role assignments, permission changes, and capability grants are recorded in an immutable, append-only audit trail. Administrators can review who made RBAC changes and when from the admin governance tools — see the Admin Guide for details.