Extract Invoice Data from PDFs with Claude + Google Sheets

Extract Invoice Fields from PDFs in n8n using Claude + Google Sheets

Upload an invoice PDF, and this n8n workflow extracts key invoice data with the Anthropic Claude API—then automatically logs high-confidence results to Google Sheets while sending uncertain items to a separate review queue.

What this workflow does

• Receives invoice PDFs via an n8n Form trigger (file upload from your form).
• Converts the PDF to raw text so the invoice content can be analyzed.
• Sends the extracted text to Claude (Anthropic Messages API) and requests strict JSON containing key invoice fields plus an overall confidence score.
• Parses and validates Claude’s JSON output, checking confidence against a threshold and flagging unknown or missing fields.
• Routes results by confidence:
  • Appends trusted extractions to the “Extracted” tab in Google Sheets.
  • Appends low-confidence or incomplete extractions to the “Review Queue” tab for manual verification.

Use cases

• Invoice processing for SaaS operators who need structured data from PDF invoices without building custom parsers.
• Automation engineers creating an extraction pipeline with human-in-the-loop review for edge cases and low-quality PDFs.
• Operations teams centralizing invoice fields in Google Sheets while keeping uncertain records out of the “Extracted” sheet.

Technical details

• n8n nodes: form trigger, HTTP request, code, if, and sticky note.
• Anthropic (Claude) integration: HTTP request with x-api-key using Anthropic’s required headers; prompts Claude to return strict JSON with a confidence score.
• Google Sheets: two append steps targeting the spreadsheet tabs “Extracted” and “Review Queue”.
• Setup requirements: add your Anthropic API key to HTTP header auth, replace REPLACE_WITH_YOUR_SPREADSHEET_ID, create the two tabs, and test with both clear and low-quality invoices.