n8n Workflow: Clocked Extract & Decide Ticket Routing

Route every support ticket automatically—or flag it for human review—using Clocked.ai inside n8n

Want faster ITSM processing without sacrificing accuracy? This n8n Workflow: Clocked Extract & Decide Ticket Routing takes raw support ticket text, extracts structured fields with Clocked.ai Extract, classifies urgency and category with Clocked.ai Decide, then applies validation and confidence rules to either send an automated update path or route to a human review queue.

What this workflow does

• Runs on manual execution in n8n (you control when it processes).
• Loads a sample support ticket plus a JSON extraction schema, then normalizes ticket text while preserving the original source and timestamps.
• Calls Clocked.ai Extract via HTTP, sending the normalized ticket text and schema, then merges the structured extraction results back into the ticket context.
• Validates extracted fields (including list fields and email format) and flags cases that require human review.
• Calls Clocked.ai Decide via HTTP to classify category and urgency and indicate whether human review is required.
• Applies confidence thresholds and rule checks to approve automated processing or route to human review, generating an audit record either way.
• Saves outputs to local JSON files for auditability; optionally it can update your ITSM ticket via an HTTP API request (disabled by default).

Use cases

• Triaging IT support requests by category/urgency from messy email or chat ticket text.
• Reducing manual workload by auto-preparing ITSM updates only when extraction confidence and field validation pass.
• Sending low-confidence or malformed tickets (e.g., invalid email) to a human review queue with full audit output.

Technical details

• Clocked.ai integrations: Extract and Decide endpoints via HTTP with Bearer Auth.
• n8n nodes used: if, set, code, merge, function, sticky note.
• Core outputs: automated-path or review-path payloads written to local JSON files; optional ITSM HTTP update (off by default).