n8n Trust & Safety Triage: Jev Classification, Sheets & Telegram

Automatically triage Trust & Safety reports—classify, queue with SLA, log to Sheets, and page humans on Telegram before it’s too late

If you run trust & safety moderation, this n8n workflow routes every incoming report from your intake form webhook into the right policy category, logs the decision, and alerts the correct human reviewer via Telegram (or email) with an enforced SLA. It’s designed to keep the reports that matter from being lost in the queue.

What this workflow does

• Receives reports via webhook: Accepts a POST request to jev-c1-report-intake with report_id, content_id, content_text, and reporter_note.
• Normalizes payload for decision-making: Uses Normalize Report to preserve policy-relevant fields, including report_id, content_id, reporter_note, and report_text.
• Guards against unusable submissions: Runs Report Text Usable?. If report_text is empty, it marks the report as intake_rejected, sets required_action to ask the reporter to resend, and notifies ops via Email Intake Problem.
• Classifies into policy categories: For usable text, Jev Classify Policy Category assigns one of five policy categories using a 0.6 confidence threshold. Uncertain cases are sent to human review.
• Routes by category with real SLA: Each of the six outputs goes to its own route/queue with defined SLA hours, priority, and required action.
• Logs decisions and escalates: Appends a row to the Report Triage tab in Google Sheets, then pages the matching human on Telegram (or sends an email) before the SLA clock runs out.

Use cases

• Moderating user-generated content when reports arrive continuously and must be classified consistently.
• SaaS teams needing audit-friendly triage records in Google Sheets while escalating high-risk items faster.
• Handling incomplete/blank reports by automatically rejecting and requesting resubmission.

Technical details

• n8n nodes/tools: if, set, gmail, webhook, telegram, summarize
• Core logic: Jev classification to policy categories at a 0.6 confidence threshold, plus SLA-based queue routing
• Documentation: Last updated October 2026