Back to work
n8nCase study

Conflict Taxonomy Engine

AI-powered conflict classification system that validates HR discovery submissions, applies structured taxonomy analysis with a confidence gate, and publishes approved friction reports to Google Sheets and Slack, giving HR and operations teams auditable, high-quality conflict insights.

n8n · workflow canvasproduction build
Conflict Taxonomy Engine workflow

Context

HR teams lacked a reliable way to transform discovery submissions into consistent conflict taxonomy reports, making it hard to separate high-confidence classifications from noisy AI output while keeping founders informed of validation issues and system failures.

Approach

Built two interconnected n8n sub-workflows that validate incoming HR submissions, analyze conflicts with an OpenAI agent using memory and a structured output parser, pass only high-confidence results through a confidence gate, then merge, flatten, log, and broadcast the final friction report. A global error handler sends validation and system error alerts to the Founder via Gmail.

Architecture

  1. Webhook POST (HR discovery intake)
  2. Extract + validate
  3. Validation failure: Gmail to Founder
Validation success
  1. OpenAI conflict agent (Memory + Structured Output Parser)
  2. Parse AI response
  3. Confidence Gate
  4. Low confidence: withheld
High confidence
  1. Prepare review flag
  2. Merge
  3. Flatten for Sheets
  4. Google Sheets friction report log
  5. Slack friction report
  6. Global Error Trigger
  7. Gmail to Founder

Results

~4 hrs/weekHR review time saved
Filtered before reaching HRLow-confidence AI output
100% logged automaticallyAudit-ready reporting