AI Agent n8n Auto Receipt Scanner to Google Sheets Sheets

The Daily Trigger node serves as the entry point for your entire receipt scanning pipeline. Instead of manually running the workflow whenever you have new receipts, this scheduler ensures the automation executes consistently every single day without any intervention.

By setting the trigger to fire at midnight, you create a “set and forget” system. Toss your receipt photos into the designated Google Drive folder throughout the day, and the workflow handles everything while you sleep.

This Google Drive node scans your designated “Receipts” folder to identify all files waiting to be processed. It’s the reconnaissance step — finding out what’s new since the last run so the workflow knows exactly which images to analyze.

The node returns a list of all files in the folder, including their unique IDs, names, and metadata. These file IDs become crucial in subsequent steps for downloading, analyzing, and moving each receipt.

The Loop Over Items node (technically a “Split in Batches” node) ensures each receipt image is processed as an individual unit through the entire pipeline. While n8n nodes typically handle multiple items automatically, this explicit loop gives you precise control over the flow.

Setting the batch size to 1 means each receipt goes through download → AI analysis → Google Sheets append → file move as a complete sequence before the next receipt begins. This prevents race conditions and keeps your data organized.

This Google Drive node downloads the actual receipt image file so it can be sent to Gemini AI for analysis. The file ID is pulled dynamically from the current loop item, meaning each iteration downloads the specific receipt currently being processed.

The downloaded image is stored in n8n’s binary data format, making it ready to pass directly to the Gemini vision AI node. This approach works with any image format — JPEG, PNG, HEIC — that Google Drive can store and Gemini can analyze.

This is the intelligence core of your workflow. The Google Gemini node sends your receipt image to Google’s Gemini 2.5 Flash vision model along with detailed instructions for extracting structured data. Gemini “reads” the receipt and returns comprehensive JSON containing every detail it can identify.

The prompt is critical — it instructs Gemini exactly what to look for: company information, transaction timestamps, individual line items with quantities and prices, automatic category assignments, tax breakdowns, totals, and payment methods.

The Code node processes Gemini’s raw text response and converts it into a proper JavaScript object that n8n can work with. Even though we asked Gemini for pure JSON, AI models sometimes wrap their responses in markdown code blocks. This node strips that formatting and parses the clean JSON.

This transformation is essential because the Google Sheets node needs structured JSON data with named fields to map to spreadsheet columns.

The Google Sheets node takes all the structured receipt data extracted by Gemini and appends it as a new row in your expense tracking spreadsheet. Using automatic column mapping, the node matches incoming JSON field names directly to your spreadsheet column headers.

This means your Google Sheet needs column headers that exactly match the field names returned by Gemini: company_name, transaction_date, total, payment_method, etc.