The Ticket Router

Description

GOAL
Build a workflow that automatically classifies every newly incoming support ticket (category and priority), assigns it to the responsible team, updates the ticket system, and prepares a personalized first reply, with dispatch protection for critical cases.

Adapt the structure to the concepts of your automation tool (for example nodes, steps, zaps, or agent actions). While doing so, preserve the trigger logic, the sequence, the dispatch branch, and the data flow.

TRIGGER
Type: webhook or native integration with your ticket system (Zendesk, Freshdesk, or comparable) that fires on every new ticket and supplies subject line, message, and customer context.

STEPS
1. AI step: classification. Use exactly the Categorization and Prioritization working prompt (see the Working Prompts section). Output exclusively as JSON with the fields category, priority, team, and reasoning.
2. Routing: assign the ticket to the matching queue or team based on the category to team mapping. Team names must exactly match the identifiers in the ticket system.
3. Update ticket: set category, priority, team, and tags in the ticket system.
4. AI step: first reply. Use exactly the First Reply working prompt (see the Working Prompts section).
5. Dispatch branch: for low-risk cases (for example category Question with priority Low or Medium), send the first reply automatically. For Critical, High, or complaint, save the first reply only as a draft and request approval before it is sent.

PROMPTS
Take over both working prompts from the section of the same name unchanged. Do not alter the placeholders they contain, and map them as described in the data flow.

DATA FLOW
Trigger to step 1: ticket subject, ticket message, customer name.
Step 1 to steps 2 and 3: category, priority, team, reasoning as JSON.
Trigger and step 1 to step 4: customer name, ticket ID, category, priority, team.
Step 1 to step 5: category and priority drive the dispatch branch.

ERROR HANDLING
No valid JSON from step 1: keep the "JSON only" instruction, validate the output, and retry once. Then place the ticket unclassified in a triage queue.
Incorrect priority on vague tickets: add examples for each priority level to the prompt. When in doubt, choose the higher priority, the approval step will catch it.
Team names do not match: match the exact queue or team identifiers from the ticket system, otherwise routing fails.
Customer name not extractable: use a neutral greeting as a fallback instead of an incorrect name.
Ticket covering multiple issues: classify by the most urgent concern and note the rest in the reasoning.
Critical or complaint should not be sent automatically: enforce the dispatch branch from step 5 consistently. These cases always go through approval.

PLACEHOLDER CONVENTION
All placeholders in square brackets inside the two working prompts are mapped dynamically: [TICKET_SUBJECT], [TICKET_MESSAGE], [CUSTOMER_NAME], and [TICKET_ID] from the trigger, as well as [CATEGORY], [PRIORITY], and [TEAM] from the JSON of step 1. [SUPPORT_TEAM_NAME] is not dynamic: set this value before activation, along with the category to team mapping as workflow configuration. List all placeholders that require manual configuration separately at the end.

SELF CHECK BEFORE COMPLETION
Check, before you report the workflow as finished:
1. Are all steps connected correctly, including the dispatch branch (low-risk automatic, Critical/High/complaint only as a draft with approval)?
2. Are both working prompts inserted in full and unchanged, and are all placeholders mapped?
3. Is the JSON validated, and do the team names exactly match the queues in the ticket system?
4. Are all points listed that require manual configuration: ticket system access with write permissions, the category to team mapping, the low-risk rule, [SUPPORT_TEAM_NAME], and LLM access?

Working prompts

You are a support ticket analysis assistant. Analyze the following support ticket and categorize it precisely.

**Ticket details:**
- Subject: [TICKET_SUBJECT]
- Message: [TICKET_MESSAGE]
- Customer: [CUSTOMER_NAME]

**Your task:**
1. Assign the ticket to EXACTLY ONE of these categories:
   - Bug (technical error in the software)
   - Feature Request (request for a new function)
   - Question (general inquiry or how-to)
   - Complaint (dissatisfaction with service or product)
   - Technical Issue (access or configuration problem)
   - Billing (questions about invoices, payment, subscription)

2. Set the priority:
   - Critical: system down, no usage possible, data loss
   - High: important function impaired, multiple users affected
   - Medium: single issue, workaround possible
   - Low: cosmetic issue, feature request, general question

3. Determine the responsible team:
   - Development: bugs, technical issues, feature requests
   - Customer Success: questions, onboarding, best practices
   - Sales: upgrade requests, product information
   - Finance: billing, invoices, payment issues
   - Management: complaints, escalations

**Respond exclusively in this JSON format:**
{
  "category": "[Category]",
  "priority": "[Priority]",
  "team": "[Team]",
  "reasoning": "[Brief reasoning in 1-2 sentences]"
}

Create a professional, friendly first reply for a support ticket.

**Context:**
- Customer name: [CUSTOMER_NAME]
- Ticket ID: [TICKET_ID]
- Category: [CATEGORY]
- Priority: [PRIORITY]
- Responsible team: [TEAM]

**Requirements:**
- Personal address by name
- Confirmation of ticket receipt
- Brief summary that we have understood the request
- Realistic turnaround time based on priority:
  * Critical: within 2 hours
  * High: within 4 hours
  * Medium: within 24 hours
  * Low: within 48 hours
- Friendly, helpful tone
- Signature with "[SUPPORT_TEAM_NAME]"

**Respond ONLY with the email body, no subject line.**

JSON export