Approvals

Approval gates are the human-in-the-loop checkpoints that keep your team in control of AI workflows. When a workflow reaches a step that requires human judgement, it pauses and waits for someone on your team to review and approve.

How approvals work

1. A workflow reaches an approval point. For example, the AI has prepared code changes and wants to create a pull request.
2. You are notified. The portal shows the approval request with details about what the AI is proposing.
3. You review and decide. You can approve the request (the workflow continues) or reject it (the workflow stops and your feedback is recorded).
4. The workflow proceeds or stops. Approved workflows continue to the next step. Rejected workflows stop and your feedback is posted to the ticket.

Where approvals appear

You can see pending approval requests:

• On the ticket detail page, where the workflow is running
• In the approvals queue, accessible from the admin dashboard

Each approval request shows what the AI is proposing, including a summary of the changes and any relevant details.

Approving or rejecting

• Approve -- Click the approve button to let the workflow continue. The AI proceeds with the next step (such as creating a pull request).
• Reject -- Click the reject button and optionally provide feedback explaining why. The workflow stops and your feedback is posted to the ticket.

Timeouts

Approval requests expire after 24 hours by default. If no one responds in time, the workflow fails and a notification is sent.

Escalation

If an approval request is not acted on in time, it can escalate to additional team members. Escalation ensures that urgent approvals are not missed when the primary reviewer is unavailable.

Bulk approvals

For portals with many tickets, you can approve or reject multiple pending requests at once from the approvals queue. This helps teams manage high-volume workflows efficiently.