Summarize and categorize PII-masked support tickets with Nexis and OpenAI

Quick Overview

This workflow manually runs to anonymize support ticket text with Nexis, then sends only masked content to an OpenAI chat model to generate a one-sentence summary and category, while routing any anonymization failures to a separate review output.

How it works

1. Runs when you click Execute workflow to generate example support tickets.
2. Sends each ticket’s text to Nexis to mask detected personal data (with an allow list to keep specific terms readable).
3. Routes tickets Nexis can’t process to a review output with the error details.
4. Sends the masked ticket text to an OpenAI chat model to produce a one-sentence summary and assign a category (Billing, Technical, Delivery, or Other).
5. Outputs the ticket ID, masked ticket text, count of masked entities, and the AI summary/category.

Setup

1. Install the community node n8n-nodes-nexis and create a Nexis API credential.
2. Select the Country/Regulation option in the Nexis anonymization step and adjust the allow list if needed.
3. Add an OpenAI credential for the chat model and choose the model you want to use.
4. Replace the sample ticket generator with your real help desk source (or keep it for testing) and connect the Result output to where you want to store or send the summaries.