Overview

Mark relevance, identify unsupported output, and provide review feedback without changing the underlying evidence record.

Audience

• Lawyer
• Compliance
• Ops

Product Area

AI Governance

Prerequisites

• Search result or AI-assisted output available

Steps

1. Choose the feedback target

Select the specific result, citation, claim, classification, or workflow decision that requires correction.

1. Record a structured disposition

Mark relevant, irrelevant, supported, unsupported, incomplete, or unsafe and add a concise reason without copying privileged content.

1. Escalate material errors

Route errors that could affect a consequential decision to human review instead of relying on future model improvement.

Expected Result

Feedback is associated with the correct tenant, workflow, output version, reviewer, and audit record.

Security and Audit Notes

• Feedback is operational data and may be retained for quality and audit purposes.
• Do not include secrets or data outside the target matter scope.

Limits and Preconditions

• Available controls depend on the tenant plan, enabled modules, jurisdiction, and administrator policy.
• If a named control is not visible, confirm entitlement and role access before treating the behavior as a product failure.

Troubleshooting

• Feedback contains sensitive source text: Reference the controlled record and provide a redacted explanation instead.

Related Articles

• Review AI Citations
• Handle Low-Confidence Output
• Understand Confidence and Evidence Strength

Escalation

If the documented result cannot be reached after the checks above, capture the workspace identifier, affected product area, timestamp, and a redacted error message, then use Support. Never include secrets, privileged legal content, or customer documents in the initial report.