Beyond this walkthrough

This 5-minute demo shows the core oversight loop. The production platform also provides:

• Coverage telemetry -- Answers 'did every AI decision actually pass through oversight?' by reconciling observed AI calls against scored decisions.
• Decision replay -- Deterministic re-execution of any recorded decision against its original policy version, for audits.
• Judgment decay measurement -- Longitudinal JIS tracking that detects skill decay in human reviewers before it becomes an incident.
• Escalation chains -- Multi-tier routing with policy-defined escalation targets, beyond the single escalation shown here.
• Webhooks -- Event notifications (for example must-review routing) into your existing incident and workflow tooling.
• Audit-log export -- Tamper-evident NDJSON export of the full decision and review history for regulators and internal audit.
• Reviewer console with SSO -- SAML and OIDC sign-in for reviewer teams, alongside magic-link access.
• Compliance reporting -- Org-level compliance reports mapped to US/EU AI oversight law oversight requirements, plus vendor due-diligence (DDQ) answer packs from measured behavior.
• Certification tiers -- A published certification ladder (L1 to L3) with a normative public specification. Read the specification
• Mandatory reasoning capture -- Reviewers commit their own reasoning before the AI recommendation is revealed, so the record cannot be back-fit.
• Signed evidence receipts -- Exportable per-decision receipts that anyone can verify through the public receipt checker.
• Judgment velocity early warning -- A predictive trend over each reviewer's independence scores that flags rubber-stamping before it becomes an incident.
• Organization dashboard and independence trend -- Org-level oversight metrics with cross-reviewer independence-collapse monitoring.
• Typed SDKs and evidence writeback -- TypeScript and Python SDKs, plus evidence sync into GRC, SIEM, and ticketing systems.