Screen German Job Ads for AGG Risk with Groq & Data Tables

Screen German Job Ads for AGG Risk—using Groq, AGG Rule Tables, and Clear Verdicts

This n8n workflow automatically reviews a German job title and job ad text for potential AGG discrimination wording. It combines deterministic checks from an AGG terms data table with Groq-based language flagging, then outputs a clear “LIKELY CLEAR” result or a “REVIEW” verdict with a 0–100 risk score—plus a detailed screening log in an n8n Data Table.

What this workflow does

• Collect inputs via an n8n Form: receives a job title and the full job ad text.
• Applies your screening policy: uses a configurable review threshold and a maximum points cap for model contribution.
• Loads AGG wording rules from an n8n Data Table: deterministically scans the ad for matching phrases and flags relevant risks.
• Checks gender marker presence: adds a warning if a gender marker like “(m/w/d)” is missing.
• Uses Groq to identify risky phrasing: sends the ad text to a Groq Chat Model to find potentially discriminatory wording verbatim, the protected ground, a brief reason, and a neutral rewrite suggestion—along with a confidence score.
• Generates a combined verdict: merges rule-based flags with a capped, confidence-weighted model contribution to produce LIKELY CLEAR, REVIEW, or VIOLATION RISK, plus a 0–100 risk score.
• Logs results to an n8n Data Table: stores verdict, score, matched phrases, model confidence, and the threshold used.
• Shows an outcome page: a detailed review page with flagged passages when needed, or a short confirmation page when likely clear.

Use cases

• HR operations and SaaS teams screening job ads before publication to reduce AGG wording risk.
• Automation engineers building compliant content review pipelines in n8n.
• Recruiting platforms generating postings from templates and running AGG risk checks automatically.

Technical details

• n8n nodes: if, set, code, form, data table, and sticky note.
• Integrations: Groq API via a Groq Chat Model node (requires Groq API credential selection).
• Data: an n8n Data Table for “AGG term rules” (used by the “Load AGG Term Rules” step) and an additional screening log Data Table.