Review Queues
Review Queues provide structured human oversight for sessions that require intervention before or after action is taken. Sessions are routed into queues by policy rules, then managed through assignment, SLA tracking, escalation, and recorded review outcomes.
Use queues when policy requires a human checkpoint. They create a durable record of oversight decisions and feed directly into exported evidence.
Review Queues
Review Queues provide structured human review for AI sessions that need a person to confirm, reject, or escalate the outcome. Sessions enter a queue when an active policy rule routes them there, usually because they are high risk, regulated, or below an acceptable threshold.
Administrators, Governance Engineers, and Compliance Officers can access Review Queues. Governance Engineers can review sessions in the queue and record outcomes — but queue creation and SLA configuration belong to Admins and Compliance Officers. Developers, Business Owners, and Auditors do not have access to this workspace.
You can review sessions routed to your queue, open them in Time Machine, and record outcomes (Approved, Rejected, Escalated). You cannot create queues or configure SLAs — those actions belong to Admins and Compliance Officers.
You create queues, assign reviewers, set SLAs, and configure escalation policies. Review decisions you record appear in compliance evidence exports tied to the originating session.
You manage all queues, including archiving inactive ones and reassigning reviewers. You can also bulk-acknowledge items to clear backlogs. Review outcomes you record are attributed to your identity and included in evidence exports.
How Review Queues Work
- A policy rule with an Add to review queue action matches an incoming session.
- The session is added to the named queue and marked as Pending review.
- A reviewer assigned to that queue sees the session in their queue list and in their Focus Queue on Home.
- The reviewer opens the session in Time Machine, evaluates the decision, and records an outcome: Approved, Rejected, or Escalated.
- The review decision, reviewer identity, timestamp, and notes are recorded on the session as a permanent policy review record.
- The annotation is included in subsequent evidence exports.
Creating a Review Queue
- Navigate to Review Queues and click New Queue.
- Configure the queue:
- Name — a short label (e.g., “High-Risk Loan Decisions”)
- Description — the review purpose and escalation criteria used in exported evidence
- Reviewers — assign one or more users who can process items in this queue
- SLA — the target time from session ingestion to review completion (in hours)
- Escalation policy — what happens if the SLA is missed: notify a secondary reviewer, escalate to a named Compliance Officer, or both
- Click Create Queue.
To route sessions into this queue, create a rule in the Rules Builder with an Add to review queue action pointing to this queue.
Processing a Review
To review a session:
- Open Review Queues and select your queue.
- The queue list shows sessions ordered by ingestion time (oldest first). Sessions past their SLA are highlighted in amber.
- Click a session row to open it in Time Machine.
- Review the session: inspect the Decision Snapshot, step timeline, and policy context.
- Click Record Review in the Time Machine toolbar.
- Select Approved, Rejected, or Escalate and add a note.
- Click Submit review.
The session is removed from the queue and the review record is written to the audit log.
Use the Blockchain Verification tab before submitting an approval if you need to verify tampering-resistance for high-stakes compliance decisions. A verified blockchain anchor result can be included in the review note.
Queue Metrics
The queue list view shows per-queue operational metrics:
| Metric | Description |
|---|---|
| Pending | Sessions awaiting review |
| Average wait time | Mean time from ingestion to review completion over the last 7 days |
| SLA breaches | Count of sessions that exceeded the configured SLA in the last 30 days |
| Throughput | Reviews completed per day over the last 7 days |
These metrics appear in exported evidence when the selected date range includes sessions from the queue.
Escalation Handling
If a reviewer escalates a session, it is removed from the original queue and added to the relevant escalation flow in Compliance. An email notification is sent to the designated Compliance Officer.
Escalated sessions require Compliance Officer sign-off before they are considered resolved. The full escalation chain, including original routing, reviewer notes, and final decision, is preserved in the policy record and audit log.