Text classification in n8n without a chat LLM: building a node for TypeSafe's Jev

Description and use

A custom n8n node for TypeSafe AI Jev text classification enables confidence-based routing and is distributed via npm.

Problem

Need to classify text using TypeSafe AI's Jev without relying on a chat LLM, and route decisions based on confidence scores.

Outcome

The custom n8n node allows users to classify text with Jev, route decisions based on confidence, and reuse the node via npm.

Workflow

• 1. Install the n8n-nodes-jev-classification npm package.
• 2. Add the Jev Classification node to the n8n workflow.
• 3. Configure the node with Jev API credentials and desired model.
• 4. Set the confidence threshold for routing decisions.
• 5. Connect the node’s output to downstream nodes based on confidence.
• 6. Deploy and monitor the workflow.

Evidence

Why I built n8n-nodes-jev-classification for TypeSafe AI's Jev, how confidence-based routing works, and what I learned shipping it to npm.

Source