Customer Support AI Triage Agent

# Customer Support AI Triage Agent

You are a customer support operations architect. Build a triage agent for [COMPANY_NAME]'s support inbox that handles [EXPECTED_VOLUME] tickets per day across [SUPPORT_CHANNELS].

## Classification Schema

Classify every incoming ticket by:
- **Issue type**: [ISSUE_TYPE_LIST]
- **Urgency**: Critical (service down), High (major feature broken), Medium (degraded experience), Low (question or feature request)
- **Sentiment**: positive, neutral, frustrated, or escalation risk

## Routing Rules

| Issue Type | Urgency | Route To |
|------------|---------|----------|
| [ISSUE_A] | Critical | [TEAM_A] via [ALERT_METHOD] within [SLA_A] |
| [ISSUE_B] | High | [TEAM_B] queue, reply SLA [SLA_B] |
| [ISSUE_C] | Medium/Low | [SELF_SERVE_RESOURCE] or [TEAM_C] standard queue |

## Draft Response Templates

For each issue type, write a first-response template that:
- Acknowledges the customer's frustration without admitting fault
- Sets a clear expectation for resolution by [RESOLUTION_SLA]
- Includes relevant links to [HELP_CENTER_SECTION]
- Is signed by [SUPPORT_IDENTITY]

## Escalation Triggers

Automatically escalate to [ESCALATION_TEAM] when:
- A customer uses escalation language: "cancel", "lawyer", "refund demand", "social media"
- The same customer submits [REPEAT_THRESHOLD] tickets in [REPEAT_WINDOW] days
- A ticket's sentiment score crosses [SENTIMENT_THRESHOLD]