AI Agent n8n Scan Business Cards to Google Sheets Sheets

This node kicks off your entire workflow automatically—no manual clicking required. It fires once every day at midnight, scanning your business card folder while you sleep.

The Schedule Trigger is n8n’s way of running workflows on autopilot. You configure when it should fire, and n8n handles the rest. For business card processing, daily makes sense—it batches your new cards from the day before without overwhelming your Google API quotas.

Now we need to find the actual business card images in your Google Drive. This node scans a specific folder and returns all files it contains—these are your unprocessed cards waiting to be digitized.

The Search operation retrieves file metadata (including the crucial file ID) without downloading the actual content yet. This keeps things fast and lets us loop through potentially dozens of cards efficiently.

With potentially many business card images to process, this node ensures we handle them one at a time. This is crucial for reliable AI processing—Gemini analyzes one image per request, and we want each card’s data to flow cleanly into its own spreadsheet row.

The Loop Over Items node (also called SplitInBatches) takes your array of files and processes them sequentially. After each iteration completes, it loops back to handle the next card until all are done.

Now we actually download the business card image from Google Drive. The previous Search node gave us file metadata—this node grabs the actual image content so we can send it to Gemini for analysis.

The node uses the file ID from the current loop iteration to know exactly which file to download. The image gets stored as binary data in n8n’s memory, ready to pass to the AI.

This is where the magic happens. The Google Gemini node sends your business card image to Google’s vision AI along with a specialized prompt that tells it exactly what data to extract and how to format the response.

Gemini 2.5 Flash is perfect for this task—it’s fast, accurate with text recognition, and handles varied business card layouts well. The key is the prompt: we tell it to act as a “business card data extraction specialist” and return a clean JSON object with specific fields.

Gemini returns its response in a nested structure, and sometimes wraps the JSON in markdown code blocks. This Code node extracts the text, cleans off any formatting, and parses it into a proper JavaScript object that downstream nodes can work with.

This is a simple but essential transformation step. Without it, you’d be trying to map nested text strings into spreadsheet columns—not fun.

Now we write the extracted contact information to your Google Sheets database. Each processed business card becomes one new row with all the structured data neatly organized in columns.

The node uses automatic column mapping, which means your spreadsheet headers need to match the JSON field names from the AI extraction (first_name, last_name, email, etc.). Set up your headers once, and every future card automatically populates the right columns.