Parse PDF Invoices with LlamaParse & Save to Airtable

Overview

This workflow automatically monitors a Google Drive folder for new invoice PDFs, sends them to LlamaParse for intelligent document parsing, extracts line items using OpenAI's structured output, and stores both the invoice header and individual line items in Airtable. It eliminates manual data entry by transforming unstructured invoice PDFs into structured database records.

The workflow is particularly useful for accounting teams, freelancers, and small businesses who receive invoices as PDFs and need to track expenses, line items, and vendor details in a structured format. By combining Google Drive's file monitoring, LlamaParse's document parsing, OpenAI's structured extraction, and Airtable's database capabilities, it creates an end-to-end invoice processing pipeline.

Node-by-Node Breakdown

1. Google Drive Trigger (OAuth2) — Monitors a specific Google Drive folder for newly created files. Configured to poll every minute and watch the "Invoices" folder (folder ID: 1IC39VXU8rewBU85offxYlBd9QlYzf8S7). When a new file is detected, it triggers the workflow.

2. Google Drive (OAuth2) — Downloads the newly detected file from Google Drive using the file ID passed from the trigger node. The operation is set to "download" and retrieves the binary file data.

3. Upload File (Header Auth) — Sends the downloaded PDF to LlamaParse's API endpoint (https://api.cloud.llamaindex.ai/api/parsing/upload) using a multipart form-data POST request. Includes:

   • The file as binary data
   • A webhook URL for callback (https://n8n.lowcoding.dev/webhook/0f7f5ebb-8b66-453b-a818-20cc3647c783)
   • Disables OCR and image extraction
   • Parsing instructions to extract invoice line items (Name, Quantity, Unit Price, Amount)
   • Authorization header (Bearer token — you must add your LlamaParse API key)

4. Webhook (No auth) — Receives the parsed result from LlamaParse via HTTP POST callback. This is the webhook URL specified in the upload step. It captures the structured JSON output containing parsed invoice data.

5. Set Fields (No auth) — Prepares the prompt and JSON schema for OpenAI. Sets two fields:

   • prompt: "Please, process parsed data and return only needed."
   • schema: A strict JSON schema that defines the expected output structure (array of items with description, qty, unit_price, amount fields)

6. OpenAI - Extract Line Items (API Key auth via OpenAI credential) — Calls OpenAI's chat completions API (https://api.openai.com/v1/chat/completions) with:

   • Model: gpt-4o-mini
   • System message containing the prompt from Set Fields
   • User message containing the parsed invoice data from the webhook
   • Response format set to JSON schema for structured output

7. Create Invoice (API Key auth via Airtable Personal Access Token) — Creates a new record in the Airtable "Invoices" table (base: appndgSF4faN4jPXi, table: tbloPc7Eay4Cvwysq). The invoice record is created first to get its ID for linking line items.

8. Process Line Items (No auth) — A JavaScript code node that:

   • Extracts the content field from OpenAI's response
   • Parses the JSON string to get the items array
   • Splits each item into a separate output item for individual processing

9. Create Line Item (API Key auth via Airtable Personal Access Token) — For each line item extracted, creates a record in the Airtable "Line Items" table (table: tblIuVR9ocAomznzK) with:

   • Description, Qty, Unit price, Amount (parsed and cleaned)
   • A link back to the parent invoice record using the invoice ID

Setup Instructions

Prerequisites

• Google Account with Google Drive access
• LlamaParse API Key from LlamaIndex Cloud
• OpenAI API Key from OpenAI Platform
• Airtable Account with a base containing two tables: "Invoices" and "Line Items"

Step-by-Step Setup

1. Google Drive Credentials: Create an OAuth2 credential in n8n for Google Drive. Grant access to the folder you want to monitor.

2. LlamaParse API Key: In the "Upload File" node, add your Bearer token in the Authorization header. Get your API key from LlamaIndex Cloud.

3. OpenAI Credential: Create an OpenAI API credential in n8n with your API key from OpenAI.

4. Airtable Credential: Create an Airtable Personal Access Token credential in n8n. Generate a token from Airtable's developer hub with scopes for data.records:write and schema.bases:read.

5. Airtable Tables: Create two tables:

   • Invoices: Contains invoice-level data (name, date, total, etc.)
   • Line Items: Contains individual line items with fields: Description (text), Qty (number), Unit price (number), Amount (number), Invoices (linked record to Invoices table)

6. Google Drive Folder: Note the folder ID of your invoices folder and update it in the Google Drive Trigger node.

7. Webhook URL: The webhook URL in the LlamaParse upload must match the production URL of your n8n instance. Update it if needed.

Use Cases & Variations

Primary Use Case

• Automated Invoice Processing: Accounting teams can drop invoice PDFs into a Google Drive folder and have them automatically parsed and stored in Airtable without manual data entry.

Variations & Adaptations

• Different Document Types: Modify the parsing instructions to extract data from receipts, purchase orders, or contracts instead of invoices.
• Alternative Storage: Replace Airtable with Google Sheets, Notion, or a database like PostgreSQL for storing parsed data.
• Email Integration: Add an email trigger (Gmail, Outlook) to process invoices sent as email attachments.
• Approval Workflow: Add a Slack or email notification step to review parsed invoices before storing them.
• Multi-Format Support: Extend the workflow to handle images (JPG, PNG) by enabling OCR in the LlamaParse upload.
• Error Handling: Add error handling nodes to catch parsing failures and notify the user via email or Slack.
• Batch Processing: Modify the trigger to process multiple files at once or on a schedule rather than real-time.

{
"nodes": [
{
"id": "9df72ef9-3b9d-40e4-9cb5-a5ada153c0bb",
"name": "Google Drive",
"type": "n8n-nodes-base.googleDrive",
"position": [
120,
-180
],
"parameters": {
"fileId": {
"__rl": true,
"mode": "id",
"value": "={{ $json.id }}"
},
"options": {},
"operation": "download"
},
"credentials": {
"googleDriveOAuth2Api": {
"id": "wpiZXesxk9S8fkVG",
"name": "Google Drive account 2"
}
},
"typeVersion": 3
},
{
"id": "e21bb906-658c-4a52-9c7b-b77d6e0e7ea5",
"name": "Upload File",
"type": "n8n-nodes-base.httpRequest",
"position": [
360,
-180
],
"parameters": {
"url": "https://api.cloud.llamaindex.ai/api/parsing/upload",
"method": "POST",
"options": {},
"sendBody": true,
"contentType": "multipart-form-data",
"sendHeaders": true,
"bodyParameters": {
"parameters": [
{
"name": "webhook_url",
"value": "https://n8n.lowcoding.dev/webhook/0f7f5ebb-8b66-453b-a818-20cc3647c783"
},
{
"name": "file",
// ... truncated (copy to see full JSON)