n8n QA/QC Compliance Classifier with Gemini, GitHub & Sheets

QA/QC compliance decisions—automated with Gemini, GitHub, and Google Sheets

This n8n workflow answers QA/QC compliance and non-conformance questions by retrieving evidence from a controlled GitHub knowledge base using Google Gemini, classifying results (Compliant/Observation/Minor/Major), logging an audit record to Google Sheets, and escalating to your QA team via email when manual review or CAPA is needed.

What this workflow does

• Receives a QA question through an n8n chat trigger and captures session metadata (query, session ID, timestamp, and source).
• Validates the input to ensure the query is present and long enough to process.
• Uses Google Gemini (RAG) with GitHub to:
  • List approved QA/QC documents in a GitHub repository,
  • Retrieve the most relevant document text,
  • Generate an evidence-grounded, structured assessment.
• Normalizes the structured output into fields such as classification, severity score, requirement violated, recommended action, escalation/CAPA flags, manual review flag, and source files.
• Escalates when evidence is insufficient: if the agent indicates insufficient evidence, it appends a “Pending Manual Review” record to Google Sheets and emails the QA team.
• Logs and routes outcomes: otherwise, it appends the decision to a Google Sheets audit log and emails the QA team when escalation is required—then marks escalation as sent in the sheet.

Use cases

• Auto-assess QA/QC compliance questions against approved documentation stored in GitHub.
• Classify non-conformance into Compliant/Observation/Minor/Major with an evidence trail.
• Trigger QA team review and CAPA/escalation workflows when Gemini lacks sufficient evidence.

Technical details

• Google AI (Gemini) with model models/gemini-3.5-flash-lite
• GitHub tool for retrieving approved QA/QC documents (public repo or configured access)
• Google Sheets for audit log entries and escalation status
• Routing logic and notifications using nodes such as if, set, switch, and gmail (email to QA team)
• Node includes sticky note for workflow clarity

Setup requirements

• Add credentials for Google Gemini and ensure the selected model is available.
• Configure GitHub access (or use a public repository).