Classify Support Tickets & Build Review Queue with Apixly API

Instantly classify support tickets with the Apixly API—and build a review queue you can trust

This n8n manual workflow validates a batch of support tickets, sends each one to Apixly’s OpenAI-compatible chat/completions API for classification and draft reply drafting, then produces a single human review queue that includes both successful results and any failures—so your team can triage fast and safely.

What this workflow does

• Manual execution: Runs when you manually execute the workflow in n8n.
• Ticket validation: Loads a configured model, endpoint, and a batch of up to 10 tickets, then validates required fields, size limits, and unique ticket_id values.
• Per-ticket classification + reply draft: Processes tickets one by one and calls Apixly’s chat/completions endpoint, requesting strict JSON containing: category, priority, summary, and a draft reply.
• Response validation: Checks HTTP status, verifies JSON format, confirms the schema matches expected fields, ensures ticket_id is preserved, and flags incomplete outputs.
• Error + safety handling: Flags transport/API errors, non-2xx responses, invalid JSON, schema mismatches, and unsafe refund claims for manual triage.
• Consolidated review report: Produces a single human review output containing every ticket, available classifications, and summary counts by category and priority.

Use cases

• SaaS support teams who want consistent triage across high-volume ticket batches.
• Automation engineers building n8n automation workflow patterns around OpenAI-compatible APIs.
• Operators who need a review queue that includes successes and failures for accountability.

Technical details

• Nodes used: manual trigger, set, code, split out, sticky note, and HTTP request.
• Apixly integration: Use an n8n HTTP Header Auth credential with header name Authorization and value Bearer <your Apixly API key>, then configure the Apixly endpoint and model in the ticket configuration step.

Replace the synthetic demo tickets with your own authorized ticket data and keep batches to the configured limit for predictable performance.