How to Build an Automated Ticket Triage System with Make.com and Claude Sonnet 4.6

Support teams waste up to 30% of their day just reading and assigning Zendesk tickets. Stop doing this. In this guide, we wire up a Make.com webhook to intercept tickets the second they open, feed them to Claude Sonnet 4.6, and instantly route the parsed data to the right Slack channel.

Step 1: Capture incoming Zendesk tickets

Polling Zendesk every 15 minutes is a waste of API calls. You need real-time triggers. Webhooks push the payload to Make.com the exact second a customer submits a ticket.

1. Create a Make scenario: Log into Make.com, create a new scenario, and add a Webhooks module. Select Custom webhook.
2. Generate a payload URL: Click Add, name the webhook 'Zendesk Intake', click Save, and copy the generated URL to your clipboard.
3. Configure Zendesk triggers: In your Zendesk Admin Center, navigate to Apps and integrations > Webhooks. Create a new webhook pointing to your Make URL.
4. Set the firing condition: Under the webhook settings, map the trigger to fire on Ticket Created, passing the ticket description in the JSON body.

Step 2: Connect Claude Sonnet 4.6

Claude 4.6 kills the old token-guessing game with 'adaptive thinking'. You dump the raw, unedited ticket text into the prompt. The model decides how much compute it needs to figure out if the user is asking for a refund or reporting a server outage.

1. Add an HTTP module: Connect an HTTP module to your Webhook and select Make a request.
2. Define the endpoint: Set the URL to https://api.anthropic.com/v1/messages and the Method to POST.
3. Set authentication headers: Under Headers, add x-api-key (paste your Anthropic key), anthropic-version (set to 2023-06-01), and content-type (set to application/json).
4. Inject the payload: Paste the JSON configuration block below into the Request content field. Map your Zendesk description variable directly into the content value.

{
  "model": "claude-sonnet-4-6",
  "max_tokens": 64000,
  "thinking": {
    "type": "adaptive"
  },
  "output_config": {
    "effort": "high"
  },
  "system": "You are a ticket routing engine. Analyze the user request. Respond ONLY with a JSON object containing a 'category' key. Allowed categories: 'billing', 'technical', 'sales', 'general'.",
  "messages": [
    {
      "role": "user",
      "content": "{{1.ticket_description}}"
    }
  ]
}

Step 3: Route categorized tickets to Slack

Anthropic spits back a clean JSON string. We need to parse that string into usable Make variables, then set up a router to ping the exact Slack channel—whether that's #billing or #dev-urgent.

1. Parse the AI response: Add a JSON module after the HTTP request and select Parse JSON. Map the Data field to the text output from Anthropic.
2. Configure the router: Connect a Router module immediately after the JSON parser.
3. Set up a routing path: Add a Slack module to the top path and select Create a Message. Apply a filter on the path connection so it only triggers when the parsed 'category' variable equals 'billing'.
4. Format the alert: Select the #support-billing channel. Map the Zendesk ticket ID and Description into the Text field.
5. Duplicate for other teams: Clone this path for 'technical' and 'sales', mapping them to their respective Slack channels.