AI Agent to Automatically Extract Resume Data (via n8n)

This first step lets you manually test the workflow. Using n8n’s Manual Trigger module, you can launch the automation on demand to make sure everything runs smoothly before deploying it in production.

Manual triggering is especially useful for debugging and step-by-step testing. Once your test results are validated, you can replace it with an automated trigger (e.g. cron, webhook, etc.) if needed.

To begin: click “Execute Workflow” in n8n to start the process.

This step automatically lists all PDF files located in a specific folder on your Google Drive. Each file is treated as an individual Resume to be processed by the AI agent.

💡 Pro tip: You can find your Drive folder ID in the URL when you're inside the folder — it’s the part that comes after /folders/.

This step uses the Split In Batches module to process each Resume file independently. The loop goes through every retrieved PDF and handles them one by one in an isolated flow.

This ensures clean execution and avoids confusion when multiple files are stored in the folder.

This step downloads the Resume PDF file directly from your Google Drive folder, using the file ID retrieved during the loop process.

💡 You can replace this source with a Gmail trigger, an API endpoint, or any other system that receives Resumes.

The PDF file is saved locally on the server to a temporary path (/tmp/doc.pdf) so it can be used in the next step to extract text content.

Here, we use the pdftotext command (from Poppler) to convert the PDF into a plain text file that can be processed by the AI agent. This command must be installed on your server.

➡️ Command executed: pdftotext /tmp/doc.pdf /tmp/doc.txt

💡 Not sure how to install it? Ask ChatGPT based on your system (Linux, Docker, Mac…) or reach out to us for support.

The file /tmp/doc.txt, generated by pdftotext, is now read. Its content is loaded into n8n to be passed to the AI agent in the next step.

This step cleans and structures the raw text content to ensure it’s properly analyzed by the AI agent. The formatted text is then available in $json.data.

The text content is sent to an AI agent powered by GPT-4o via LangChain. It extracts all key information from the Resume (name, email, work experience, skills, etc.).

➡️ Prompt: fixed structure, predefined column names, apostrophes added to numbers for Google Sheets compatibility.

We convert the JSON returned by the AI agent into a flat structure, compatible with automatic import into Google Sheets. Each property (e.g. lang_1, edu_1_degree) is explicitly mapped.

💡 If you want to send the data elsewhere (Notion, Airtable, SQL database…), you can easily adapt this block.

The extracted information is automatically added to a new row in your Google Sheet. Each column matches a clearly defined Resume field.

➡️ Connection: Google Sheets connected to your account

💡 You can also modify this step to send data to Notion, Airtable, or any other database.

This command deletes all temporary files (/tmp/doc.pdf and /tmp/doc.txt) after each loop to keep the environment clean and optimized.

➡️ Command: rm -rf /tmp/doc.pdf /tmp/doc.txt