What you need
- A classifaily API key from your dashboard
- n8n - self-hosted or cloud
Store your API key
In n8n, go to Settings > Credentials and create a new Header Auth credential. Set the name to Authorization and the value to Bearer cai_live_your_key_here. This keeps your key out of individual nodes and reusable across workflows.
Configure the HTTP Request node
Add an HTTP Request node to your workflow and configure it as follows:
- Method: POST
- URL:
https://api.classifaily.com/v1/classify - Authentication: Select the Header Auth credential you created above
- Body Content Type: JSON
- Body:
{
"input": "{{ $json.message }}",
"categories": ["sales_inquiry", "support_request", "spam", "partnership", "other"]
}
Replace {{ $json.message }} with the expression that maps to the text field from your trigger node.
After configuring, click Execute Node to test. The response will appear in the output panel:
{
"result": {
"label": "support_request",
"confidence": 0.91
}
}
Branch with the Switch node
Add a Switch node after the HTTP Request node. Set the value to {{ $json.result.label }} and add a rule for each label:
- Value equals
sales_inquiry- connect to CRM node - Value equals
support_request- connect to helpdesk node - Value equals
spam- connect to a No-op or end the branch - Fallback - connect to a Slack notification for manual review
Add a confidence check
Insert an IF node between the HTTP Request and Switch nodes to filter low-confidence results:
- Condition:
{{ $json.result.confidence }}greater than0.75 - True branch - continue to Switch node
- False branch - route to manual review Slack channel
Full workflow example: contact form routing
- Webhook node - receives form submissions
- HTTP Request node - POST to classifaily with the form message and categories
["sales", "support", "partnership", "spam"] - IF node - confidence greater than 0.75
- Switch node - branches on
result.label - Per-branch actions: HubSpot, Zendesk, Gmail forward, Slack
This workflow replaces manual triage and runs 24/7 for fractions of a cent per submission.
The spam detection blog post walks through a complete n8n implementation.
Read the guide