Proof & Evidence
BeaconGuard evidence records are designed to show what was requested, which deterministic policy inputs were evaluated, what decision was returned, and why the request was allowed, denied, or routed for review. Existing systems remain the systems of record. BeaconGuard emits decision evidence at the AI request boundary.
Healthcare operational AI request
{
"correlation_id": "req_7f21_healthops",
"workflow_type": "care_management_summary",
"boundary_enforcement": {
"decision_class": "NEEDS_REVIEW",
"enforced": true,
"policy_id": "pol_healthcare_ai_context_v1.2",
"trigger_condition": "approved_pathway_required"
},
"request_context": {
"source_system": "care_management_workflow",
"user_role": "care_coordinator",
"context_tags": [
"patient_related",
"restricted_operational_context"
],
"approved_model_pathway": false,
"request_freshness_window_sec": 300
},
"decision_reason": [
"Requested model pathway is not approved for patient-related context.",
"Workflow requires privacy or architecture review before model execution."
],
"evidence": {
"request_id": "req_7f21_healthops",
"policy_version": "v1.2",
"decision_timestamp": "2026-06-22T15:42:11Z",
"review_surface": "privacy_security_audit"
}
}Financial workflow request
{
"correlation_id": "req_8a44_fincrime",
"workflow_type": "aml_fraud_review",
"boundary_enforcement": {
"decision_class": "NEEDS_REVIEW",
"enforced": true,
"policy_id": "pol_financial_analyst_ai_v1.1",
"trigger_condition": "customer_impacting_recommendation"
},
"request_context": {
"source_system": "aml_case_workflow",
"user_role": "financial_crime_analyst",
"case_metadata_present": true,
"approved_model_pathway": true,
"request_freshness_window_sec": 300
},
"decision_reason": [
"AML, fraud, or dispute recommendation requires analyst approval.",
"Decision evidence is required for risk, compliance, and audit review."
],
"evidence": {
"request_id": "req_8a44_fincrime",
"policy_version": "v1.1",
"decision_timestamp": "2026-06-22T15:45:18Z",
"review_surface": "risk_compliance_audit"
}
}Regulated internal automation request
{
"correlation_id": "req_4c19_internalops",
"workflow_type": "production_automation_request",
"boundary_enforcement": {
"decision_class": "DENY",
"enforced": true,
"policy_id": "pol_internal_ai_automation_v1.0",
"trigger_condition": "unauthorized_action_scope"
},
"request_context": {
"source_system": "internal_ops_console",
"user_role": "support_analyst",
"requested_action": "modify_production_record",
"change_window_active": false,
"approved_automation_pathway": false
},
"decision_reason": [
"Requested action is outside approved role scope.",
"Production modification requires approved automation pathway and active change window."
]
}What Reviewers Inspect
- Which existing workflow submitted the AI-bound request
- Which signed request metadata, source-system attributes, context tags, and freshness inputs were evaluated
- Which policy returned allow, deny, or needs-review
- Why model execution was allowed, blocked, or routed for review
- What audit-ready evidence was preserved for later review