Invoice Processing Workflow: Gmail PDFs to Sheets & Slack

Turn Gmail PDF invoices into organized Sheets records—and Slack alerts—in minutes

This n8n Invoice Processing Workflow: Gmail PDFs to Sheets & Slack automatically captures PDF invoices from Gmail or an n8n upload form, extracts invoice fields with OpenAI (GPT-4o-mini), de-duplicates entries, files PDFs in Google Drive, logs them in Google Sheets, and notifies your Slack channel—with an optional finance review email for anything that needs attention.

What this workflow does

• Triggers on new PDFs when a Gmail email arrives with a PDF attachment or when you upload one or more PDF invoices via an n8n form.
• Splits and filters binaries so each PDF becomes a single item, while non-PDF files are ignored.
• Extracts invoice data using OpenAI: text is sent to GPT-4o-mini, which returns strict JSON invoice fields.
• Validates and normalizes dates and number formats, checks totals (net + tax vs total), and flags inconsistent or missing data as “Needs review.”
• De-duplicates with Google Sheets by looking up the invoice number in the “Invoices” tracker and marking matches as duplicates.
• Files and records: duplicates are posted to Slack as “skipped”; non-duplicates upload the PDF to Google Drive, append a new row (including the Drive link) to Google Sheets, and notify Slack.
• Routes finance review: if the appended invoice is “Needs review,” it sends a Gmail message to finance with validation notes and the Drive link.
• Weekly digest: every Monday at 08:00, it reads the tracker, builds a summary (booked last 7 days, due in 7 days, review backlog), and posts it to Slack.

Use cases

• SaaS operators who receive vendor invoices via email and want a clean, auditable invoice ledger.
• Accounting teams who need automatic “Needs review” flags and a weekly backlog digest in Slack.
• Automation engineers building invoice processing pipelines with de-duplication and Drive/Sheets logging.

Technical details

• Integrations/Nodes: Gmail, Google Sheets, Google Drive, Slack, OpenAI (GPT-4o-mini)
• Core logic: splitting/filtering PDF binaries, OpenAI JSON extraction, normalization + totals checks, de-duplication, conditional routing (includes “Needs review”).
• Utility nodes mentioned: if, code, filter, sticky note.