Audit Scenarios

Audit trail for AI agents that approve, reject, or escalate invoices. Logs decision, input fingerprint, rationale, and outcome. Reliability measured by human reviewer agreement rate.

Learn more →

Audit trail for AI agents that approve, flag, or block journal entries. Nine flag subtypes including segregation of duties (auto-block). Reliability from false positive rate, SOD detection rate, and block accuracy.

Learn more →

Audit trail for AI agents that flag, clear, or escalate financial transaction anomalies. Thirteen anomaly subtypes and five reliability signals, with critical items automatically routed for immediate review.

Learn more →

Audit trail for AI agents that execute, reject, or flag trades. Nine flag subtypes including position limit breach and wash trade risk. Reliability from execution accuracy, rejection precision, false positive rate, and human override rate under MiFID II Article 25 and SEC Rule 17a-4 record-keeping requirements.

Learn more →

Audit trail for AI agents that generate revenue forecasts, budget projections, or earnings guidance inputs. Decisions cover submit, revise, and reject. Reliability from forecast accuracy against post-period actuals, calibration error, revision rate, materiality flag rate, and human override rate. Builds on Sigmodx's existing forecasting infrastructure.

Learn more →

Audit trail for AI agents that classify expenditures as CapEx or OpEx. Policy versioning ties each decision to the capitalization policy in effect. Material decisions require human approval. Reliability from reviewer agreement, reclassification rate, material accuracy, and policy compliance.

Learn more →

Audit trail for AI agents that assess vendor KYC, sanctions screening, and onboarding risk. Sanctions hits auto-reject regardless of agent state. Critical tier requires human approval. Reliability from sanctions detection rate, reviewer agreement, and false positive rate.

Learn more →