Automated Invoice Data Extraction with LLamaParse, Gemini, and Google Sheets

This n8n workflow automates the process of extracting key data from invoice PDFs received via email, processing it with advanced AI models, and storing the structured information in a Google Sheet. It's designed to streamline invoice management, reduce manual data entry, and improve accuracy.

What it does

1. Monitors Gmail for New Invoices: The workflow is triggered when a new email arrives in a specified Gmail inbox.
2. Filters for Invoice Emails: It checks if the email subject contains "invoice" (case-insensitive) to ensure only relevant emails are processed.
3. Extracts PDF Attachments: For each qualifying email, it extracts any attached PDF files.
4. Extracts Data from PDFs using LLamaParse: It sends the PDF content to a custom HTTP endpoint (presumably a LLamaParse instance) to extract structured data.
5. Processes Extracted Data with Google Gemini: The extracted raw text from the PDF is then sent to the Google Gemini Chat Model, along with a structured output parser, to transform it into a clean JSON format containing specific invoice details (e.g., invoice number, vendor, total amount, line items).
6. Formats Data: The extracted and processed data is then formatted into a structured object suitable for Google Sheets.
7. Checks for Existing Invoices: It queries a Google Sheet to see if an invoice with the same invoice number already exists.
8. Conditionally Adds/Updates Google Sheet:
   • If the invoice number is new, it adds a new row with the extracted data to the Google Sheet.
   • If the invoice number already exists, it updates the existing row with the new data.
9. Sends Confirmation Email: After successfully processing and recording the invoice data, it sends a confirmation email.
10. Handles Errors: If an error occurs during the processing (e.g., no PDF found, data extraction failure), it sends an error notification email.

Prerequisites/Requirements

• n8n Instance: A running n8n instance.
• Gmail Account: Connected to n8n with appropriate credentials to read emails and send new ones.
• Google Sheets Account: Connected to n8n with appropriate credentials to read and write data to a specific spreadsheet.
• LLamaParse Instance/Endpoint: An accessible HTTP endpoint for LLamaParse (or a similar PDF parsing service) to extract raw text from PDFs. This is represented by the "HTTP Request" node.
• Google Gemini API Key: Configured within the Google Gemini Chat Model node for AI-powered data extraction and structuring.

Setup/Usage

1. Import the workflow: Download the provided JSON and import it into your n8n instance.
2. Configure Credentials:
   • Gmail Trigger & Gmail Node: Set up a Google OAuth2 credential for your Gmail account.
   • Google Sheets Node: Set up a Google OAuth2 credential for your Google Sheets account.
   • Google Gemini Chat Model: Configure your Google Gemini API key.
3. Configure Nodes:
   • Gmail Trigger: Specify the mailbox and any filters (e.g., subject:invoice has:attachment filename:pdf).
   • HTTP Request (LLamaParse): Update the URL to your LLamaParse (or similar service) endpoint. You may also need to configure any necessary authentication headers or body parameters.
   • Google Sheets:
     • Specify the Spreadsheet ID and Sheet Name where invoice data will be stored.
     • Ensure the column names in your Google Sheet match the keys expected by the workflow (e.g., Invoice Number, Vendor, Total Amount, Line Items).
   • Gmail (Confirmation/Error): Update the recipient email addresses for confirmation and error notifications.
4. Activate the Workflow: Once all credentials and configurations are set, activate the workflow.

Now, whenever an email containing "invoice" in the subject and a PDF attachment arrives in your configured Gmail inbox, the workflow will automatically extract the data, process it, record it in your Google Sheet, and send notifications.