AI Run Audit Viewer

Audit every AI operation — policy-checked, evidence-grounded, fully traceable. 9 operation types tracked.

Filters

Operation Types

  • recommendation
  • classification
  • extraction
  • summarization
  • risk_scoring
  • anomaly_detection
  • generation
  • grounded_retrieval
  • reasoning

Run Statuses

pendingrunningcompletedfailedrejected_by_policy

Audit Record Fields

  • orgId, entityType, entityId
  • operationType, modelId, prompt
  • result, confidence, reasoning
  • evidence (citations with source type)
  • policyConstraints (pre/post checks)
  • tokenUsage (input/output/total)
  • latencyMs, startedAt, completedAt