n8n Invoice PDF Extract & Validate via Webhook + DeepSeek JSON

Extract, validate, and return invoice JSON automatically—via a secured webhook + DeepSeek

This n8n workflow accepts invoice PDF uploads through a protected webhook, extracts the text, converts the invoice into a single validated JSON object using DeepSeek (OpenAI-compatible), and responds with a clear valid vs needs_review result.

What this workflow does

• Receives invoice PDFs via webhook: A POST webhook endpoint accepts an uploaded file and requires header authentication.
• Validates file type early: If the upload is not a PDF, the workflow immediately flags the result as needs_review and includes a reason.
• Detects missing/usable text: It extracts the PDF text layer and flags scanned or image-only PDFs (where text is not usable) as needs_review.
• Extracts invoice fields using DeepSeek JSON: The cleaned invoice text is sent to an OpenAI-compatible /chat/completions endpoint (DeepSeek by default) to produce a single JSON object of invoice fields.
• Normalizes and validates: Dates and numbers are normalized, required fields are checked, totals and line-item math are validated, currencies are validated, and date plausibility is applied.
• Returns one JSON response row: The webhook response includes status (valid or needs_review), issues, and the extracted invoice fields.

Use cases

• Automatic invoice ingestion for SaaS back offices where PDFs arrive from customers.
• Pre-validation before pushing invoice data into accounting or fulfillment systems.
• Reducing OCR/scan failures by detecting image-only PDFs and routing them for manual review.

Technical details (n8n + integrations)

• n8n nodes: webhook, http request, set, code, switch, plus a sticky note.
• Webhook security: Header authentication (e.g., X-Api-Key).
• LLM extraction: OpenAI-compatible /chat/completions using DeepSeek (Authorization: Bearer API key).