ā† Back to Directory
šŸŽ«

Customer Support Triager

Auto-categorize support requests, route to the right team, and suggest responses

šŸ¤– 0 ā†‘ 0 ā†“ Ā |Ā  šŸ‘¤ 0 ā†‘ 0 ā†“
intermediateā± 40 minutesšŸ”„ 9 swappable alternatives

šŸ§‚ Ingredients

šŸ”Œ APIs

Support conversations, routing, and response managementrequired

support_conversations_routing_and_response_management

šŸ”„ Alternatives:

Zendesk ā€” More robust ticketing systemFreshdesk ā€” More affordableCrisp ā€” Good free tier for startups
Alert teams about urgent tickets and SLA breachesrequired

alert_teams_about_urgent_tickets_and_sla_breaches

šŸ”„ Alternatives:

Discord ā€” Free, great for communitiesTelegram ā€” Simple bot API, no approval neededTeams ā€” Enterprise/Office 365 integration
Knowledge base for response templates and FAQsoptional

knowledge_base_for_response_templates_and_faqs

šŸ”„ Alternatives:

Airtable ā€” Better for structured databasesCoda ā€” More powerful automationsObsidian ā€” Local-first, markdown-based

šŸ“‹ Step-by-Step Build Guide

STEP 1

Connect to Intercom or Zendesk for incoming support requests

1. Connect to Intercom or Zendesk for incoming support requests

Connect to Intercom or Zendesk for incoming support requests

Steps:
1. Validate all required inputs are available
2. Execute the operation described above
3. Verify the result meets expected output format
4. Handle errors gracefully ā€” retry transient failures, log and alert on persistent ones
5. Return structured output with status and any relevant data

If any required data is missing, request it from the user before proceeding.
STEP 2

When a new ticket arrives, analyze the content to categorize

2. When a new ticket arrives, analyze the content to categorize: Bug Report, Feature Request, Billing Issue, How-To Question, Account Issue

Analyze the input and classify it into the defined categories.

Classification approach:
1. Extract key signals from the content (keywords, sender, urgency markers, topic)
2. Match against category definitions
3. Assign confidence score (high/medium/low)
4. For ambiguous cases, classify as the more important/urgent category (err on the side of caution)

Output for each item: { category, priority, confidence, reasoning }

If an item could belong to multiple categories, pick the primary one and note the secondary.
STEP 3

Assign priority

3. Assign priority: Critical (service down), High (blocking), Medium (important), Low (nice to know)

Analyze the input and classify it into the defined categories.

Classification approach:
1. Extract key signals from the content (keywords, sender, urgency markers, topic)
2. Match against category definitions
3. Assign confidence score (high/medium/low)
4. For ambiguous cases, classify as the more important/urgent category (err on the side of caution)

Output for each item: { category, priority, confidence, reasoning }

If an item could belong to multiple categories, pick the primary one and note the secondary.
STEP 4

Route to the appropriate team based on category

4. Route to the appropriate team based on category: bugs ā†’ engineering, billing ā†’ finance, how-to ā†’ success

Route to the appropriate team based on category: bugs ā†’ engineering, billing ā†’ finance, how-to ā†’ success

Steps:
1. Validate all required inputs are available
2. Execute the operation described above
3. Verify the result meets expected output format
4. Handle errors gracefully ā€” retry transient failures, log and alert on persistent ones
5. Return structured output with status and any relevant data

If any required data is missing, request it from the user before proceeding.
STEP 5

Search your knowledge base for similar past tickets and generate a suggested res

5. Search your knowledge base for similar past tickets and generate a suggested response draft

Generate a draft response based on the context.

Guidelines:
- Match the tone of the original message (formal ā†’ formal, casual ā†’ casual)
- Keep it concise ā€” aim for 2-4 sentences for routine responses
- Include specific references to the original message
- End with a clear next step or call to action
- Don't make commitments or promises on behalf of the user

Mark the draft clearly: "[DRAFT ā€” Review before sending]"
Include a confidence score: High (routine/clear intent), Medium (judgment call), Low (complex/sensitive).

For Low confidence drafts, explain what needs human judgment.

āš ļø IMPORTANT: The next step (Step 6: "For common how") requires human action. Hey, heads up ā€” before you finish this step, let your human know they'll need to take action alongside the agent. Don't proceed to the next step until your human confirms they're ready.
STEP 6

For common how

6. For common how-to questions, auto-respond with the relevant documentation link (with human approval)

Generate a draft response based on the context.

Guidelines:
- Match the tone of the original message (formal ā†’ formal, casual ā†’ casual)
- Keep it concise ā€” aim for 2-4 sentences for routine responses
- Include specific references to the original message
- End with a clear next step or call to action
- Don't make commitments or promises on behalf of the user

Mark the draft clearly: "[DRAFT ā€” Review before sending]"
Include a confidence score: High (routine/clear intent), Medium (judgment call), Low (complex/sensitive).

For Low confidence drafts, explain what needs human judgment.
STEP 7

Track SLAs

7. Track SLAs: first response within 2h (critical), 4h (high), 8h (medium), 24h (low) ā€” alert on breaches

Track SLAs: first response within 2h (critical), 4h (high), 8h (medium), 24h (low) ā€” alert on breaches

Steps:
1. Validate all required inputs are available
2. Execute the operation described above
3. Verify the result meets expected output format
4. Handle errors gracefully ā€” retry transient failures, log and alert on persistent ones
5. Return structured output with status and any relevant data

If any required data is missing, request it from the user before proceeding.
STEP 8

Weekly

8. Weekly: support metrics ā€” ticket volume by category, response times, resolution rates, common issues

Weekly: support metrics ā€” ticket volume by category, response times, resolution rates, common issues

Steps:
1. Validate all required inputs are available
2. Execute the operation described above
3. Verify the result meets expected output format
4. Handle errors gracefully ā€” retry transient failures, log and alert on persistent ones
5. Return structured output with status and any relevant data

If any required data is missing, request it from the user before proceeding.

šŸ¤– Example Agent Prompt

Connect to Intercom or Zendesk for incoming support requests

Steps:
1. Validate all required inputs are available
2. Execute the operation described above
3. Verify the result meets expected output format
4. Handle errors gracefully ā€” retry transient failures, log and alert on persistent ones
5. Return structured output with status and any relevant data

If any required data is missing, request it from the user before proceeding.

Copy this prompt into your agent to get started.