Employee Attrition Risk Detection & HR Alerts using Azure OpenAI GPT-4o-mini & Gmail

Description  Automatically ingests new employee data, extracts relevant signals, scores attrition risk, and notifies HR/managers with structured insights and recommended actions. Built on Azure OpenAI Chat with Structured Output Parser and true/false routing for escalation.

What This Template Does  Trigger for new data: Starts when a new profile, survey, or report file is added. Download & extract: Retrieves the file and converts PDFs/text into analyzable content. Analyze signals: Uses Azure OpenAI Chat to interpret sentiment, workload, performance notes, feedback, and changes (role, compensation, manager, location). Structured parsing: Maps to fields like risk_score, risk_level, key_drivers, recommended_interventions, escalation_required. Logic routing: Applies thresholds (e.g., risk_score ≥ 0.7) and flags for urgent follow-up. Email alerts: Drafts and sends tailored notifications to HR/manager with action steps.

Key Signals Considered  Sentiment & language: Negative tone, burnout cues, disengagement in feedback. Activity trends: Drop in participation, delayed responses, meeting absenteeism. Performance & goals: Recent rating changes, missed OKRs, quality issues. Role & compensation: Lateral moves, pay gaps vs. market, stalled progression. Managerial context: Team churn, conflict mentions, low recognition frequency.

Features  Azure OpenAI Chat: Interprets unstructured text into consistent risk fields. Structured Output Parser: Guarantees schema for downstream decisions. Conditional Logic (true/false): Threshold checks for escalation. Memory: Maintains context across multiple files per employee for trend-aware scoring. Calculate avg span: Computes tenure or recency metrics used in risk scoring. Email Composer & Sender: Generates and dispatches HR-ready alerts.

Requirements  n8n instance with access to employee data sources (Drive, Inbox, HR folder). Extract From PDF configured for clean text output. Azure OpenAI credentials (e.g., GPT‑4o‑mini) connected to Chat Model. Email service (Gmail/SMTP) set in n8n Credentials. Parser schema aligned to your People Analytics fields (risk_score, drivers, actions).