n8n PDF Q&A Workflow: Google Gemini with Gmail & Sheets

Ask questions about your PDFs—get Gemini answers with citations, then email + log everything automatically

This n8n PDF Q&A Workflow turns form submissions into a complete document Q&A pipeline: users upload a PDF and enter a question, Google Gemini answers using only the document (with quoted citations), the result is sent via Gmail, and the Q&A is saved to Google Sheets.

What this workflow does

• Receives form submissions containing the user’s email address, a question, and an uploaded PDF.
• Uses Google Gemini document understanding to generate an answer based exclusively on the PDF content, quoting the exact clause/sentence or stating when the document lacks the answer.
• Handles unreadable PDFs by sending a failure notification to the user via Gmail.
• Formats the output with the user’s email, the original question, and the cited answer.
• Generates 3 follow-up questions using a Google Gemini chat model (as three unnumbered suggestions based on the Q&A).
• Sends an email via Gmail including the cited answer and the suggested follow-ups.
• Appends the Q&A to Google Sheets for record-keeping (e.g., a tab named “Document QA”).

Use cases

• Support teams answering customer questions from policy PDFs or contract documents with verifiable quotes.
• SaaS operators performing faster internal reviews by turning document lookups into an email + spreadsheet workflow.
• Legal/compliance teams needing document-grounded Q&A and audit-friendly logging in Google Sheets.

Technical details

• n8n Form Trigger for collecting email, question, and PDF uploads.
• Google Gemini (PaLM) document model for cited answers + Gemini chat model for follow-ups.
• Gmail OAuth2 connection to send success or failure notifications.
• Google Sheets OAuth2 integration to append Q&A to a selected spreadsheet and sheet/tab.
• Nodes mentioned: code, gmail, sticky note, form trigger, google sheets, and n8nn8n-nodes-langchainchain llm.

Setup note: Use text-based PDFs (not scanned images) and ensure your n8n instance supports the upload size.