Use Case: Multi-Step Approval Workflow

The Problem

Your organization needs multiple levels of approval before content goes live — writer, editor, legal, and final sign-off. Managing this via email chains is chaotic.

The Solution

AI generates draft
  → Writer reviews (style, tone)
    → Editor reviews (accuracy, grammar)
      → Legal reviews (compliance, disclaimers)
        → Director gives final sign-off
          → Published

How to Build It

Step 1: Design the Pipeline

On the canvas, build a chain of Pause nodes:

Trigger → Chat (AI drafts) → Pause 1 (Writer) → Pause 2 (Editor) → Pause 3 (Legal) → Pause 4 (Director) → Publish

Each Pause node is configured with:

• Who should review (by role or user)
• Escalation timeout (e.g., 4 hours)
• What information the reviewer sees

Step 2: Configure Notifications

At /admin/notifications, set up routing:

• Writer reviews → notify via in-app
• Editor reviews → notify via email
• Legal reviews → notify via email + Slack
• Director → notify via all channels

Step 3: Use the Inbox

Each reviewer sees their pending items in the Inbox:

• What content is waiting
• Who created it and when
• How long it’s been in the queue
• Approve or Reject with one click

Step 4: Track and Audit

Every decision is logged:

• Who approved/rejected at each stage
• Timestamps for each approval
• Full content at each stage
• Available at /admin/audit

Variations

Variation	How
Conditional routing	If low risk → skip legal, go straight to director
Parallel review	Editor and Legal review simultaneously, both must approve
Auto-approve	If AI confidence is high + low risk → auto-approve first stage
Feedback loop	Rejected → back to AI with feedback → regenerate → re-review

Results

• Clear accountability — every stage has a named reviewer
• No bottlenecks — escalation ensures timely responses
• Full audit trail — compliance-ready logs
• Flexible — add, remove, or reorder review stages easily