Automations
AI Classify Node
4 min readnode-ai-classify
Use AI to classify text into one of your custom labels — route leads, emails, and messages by intent without writing regex.
AI Classify Node
The Classify node sends a piece of text to an AI model and returns one of your predefined labels. It's a specialised wrapper around the LLM node optimised for classification tasks.
---
Fields
| Field | Required | Description |
|---|---|---|
input | Yes | Text to classify. Typically {{email.bodyText}} or {{triggerEvent.message}} |
labels | Yes | Comma-separated list of valid output labels (e.g. pricing, support, booking, spam) |
outputVar | No | Variable name for the result (default: label) |
---
Output
| Variable | Description |
|---|---|
{{variables.label}} | One of your configured labels |
---
Example: route inbound WhatsApp messages
json
{
"input": "{{triggerEvent.message}}",
"labels": "pricing, support, booking, complaint, other",
"outputVar": "intent"
}
Then connect to a logic.switch (or chain of logic.if) nodes to branch on {{variables.intent}}.
---
Tips
- Keep labels distinct and non-overlapping — ambiguous labels produce inconsistent results
- 3–6 labels works best; beyond 10 labels, accuracy drops
- For binary classification (e.g. spam / not spam), prefer
logic.ifwith a clear condition