Using classifaily in Make (Integromat)

What you need

• A classifaily API key from your dashboard
• A Make account (any plan - HTTP module is available on all plans)

Add the HTTP module

In your scenario, add a new module. Search for HTTP and select Make a request. Configure it with the following settings:

• URL: https://api.classifaily.com/v1/classify
• Method: POST
• Headers:
  • Name: Authorization, Value: Bearer cai_live_your_key_here
  • Name: Content-Type, Value: application/json
• Body type: Raw
• Content type: JSON (application/json)
• Request content:

{
  "input": "{{1.message}}",
  "categories": ["sales_inquiry", "support_request", "spam", "partnership", "other"]
}

Replace {{1.message}} with the mapped field from your trigger module. Make uses {{moduleNumber.fieldName}} syntax for field mapping.

Enable Parse response so Make automatically parses the JSON and gives you named fields in subsequent modules.

Run the module once to verify. You should see a response like:

{
  "result": {
    "label": "support_request",
    "confidence": 0.91
  }
}

Branch with the Router module

Add a Router module after the HTTP module. Each route gets a filter. Set the filter condition for each branch:

• Route 1: result.label equals sales_inquiry - connect to HubSpot or your CRM
• Route 2: result.label equals support_request - create a Freshdesk or Zendesk ticket
• Route 3: result.label equals spam AND result.confidence greater than 0.80 - no action
• Route 4: fallback - post to Slack for manual review

Extract fields with the Set Variable module

If you need to reference the label or confidence in multiple places later in your scenario, add a Set Multiple Variables module right after the HTTP module:

• Variable name: label, Value: {{http_module.result.label}}
• Variable name: confidence, Value: {{http_module.result.confidence}}

Both values are then available throughout the rest of the scenario by name.

Tips for Make

• Use the Data Store to log every classification result - label, confidence, and the original input - for auditing and tuning.
• Aggregate before classifying. If your trigger fires on individual items in a list, use an Iterator and then classify each item separately rather than trying to classify a whole array at once.
• Error handling. Add an error handler route on the HTTP module. If classifaily returns an error (network timeout, rate limit), route to a Slack alert rather than silently dropping the item.