AI-Powered CV Scanner with Google Sheets & Gemini

Overview

This workflow automates the initial screening of job applicants by combining a web form, AI-powered CV analysis, and Google Sheets. When a candidate submits their application (name, email, and CV PDF), the workflow instantly extracts the text from the PDF using Mistral OCR, then sends it to Google Gemini for evaluation against a predefined job description. The AI returns a qualification score (0.0 to 1.0) and a detailed explanation, which are then logged alongside the candidate's details in a Google Sheet. This eliminates manual CV review, speeds up your hiring pipeline, and ensures every applicant is evaluated consistently.

Workflow Steps

1. Application Form (Form Trigger) — No auth required. This node presents a custom-branded web form to collect the candidate's Full Name, Email, and CV (PDF upload). The form is styled with a dark, glassmorphism theme and includes animated backgrounds. Once submitted, the workflow triggers automatically.

2. Log Candidate Submission (Google Sheets) — OAuth2. Immediately after form submission, this node appends the candidate's name and email to the 'CVs' Google Sheet. This ensures the application is recorded even if the subsequent analysis steps fail.

3. Extract CV Text (Mistral AI) — API Key auth. This node takes the uploaded PDF file from the form and uses Mistral's OCR API to extract all text content from the CV. It dynamically selects the binary file uploaded in the previous step.

4. AI Qualification (LLM Chain) — This is the core analysis node. It uses a detailed prompt that includes:

   • A system message defining the AI as a professional, objective hiring assistant.
   • The full job description for a "Senior Frontend Developer" with core requirements and preferred qualifications.
   • Clear evaluation logic: candidates missing core requirements cannot score above 0.6; meeting all core requirements yields at least 0.75; preferred qualifications add bonuses up to 1.0.
   • The extracted CV text as context.
   • An instruction to output raw JSON only.

5. Gemini 2.5 Flash Lite (Language Model) — API Key auth (Google AI Studio). This node provides the actual AI model (Google Gemini 2.5 Flash Lite) with a temperature of 0.4 for balanced, deterministic outputs. It powers the LLM Chain above.

6. JSON Output Parser (Output Parser) — This node ensures the AI's response is valid JSON with the exact schema: qualificationRate (number) and explanation (string). It prevents malformed responses from breaking downstream nodes.

7. Add CV Analysis (Google Sheets) — OAuth2. After analysis, this node updates the candidate's row in the 'CVs' sheet by matching on the Email column. It writes the qualificationRate and explanation from the AI's output into the corresponding columns.

8. Create 'CVs' Spreadsheet (Google Sheets) — OAuth2. This node runs once when the workflow is first triggered manually. It creates a new Google Sheet named 'CVs' with columns: FullName, Email, QualificationRate, QualificationDescription.

Setup Instructions

Prerequisites

• A Google account (for Google Sheets)
• A Mistral AI account (free tier available) — Get your API key here
• A Google AI Studio account — Create an API key here

Step-by-Step

1. Google Sheets Credentials: In the three Google Sheets nodes, create or select an OAuth2 credential for Google Sheets. Grant access to create and edit spreadsheets.
2. Mistral AI Credential: In the "Extract CV Text" node, create a new credential and paste your Mistral API key.
3. Gemini Credential: In the "Gemini 2.5 Flash Lite" node, create a new credential and paste your Google AI Studio API key.
4. Customize Job Description: Open the "AI Qualification" node and edit the <job_requirements> section in the prompt to match your actual job posting.
5. Activate the Workflow: Toggle the workflow to "Active" status. The "Application Form" node will generate a public URL you can share with candidates.
6. Test: Use the "Start Here" manual trigger to initialize the Google Sheet, then submit a test application via the form URL.

Use Cases & Variations

• Custom Job Roles: Replace the Senior Frontend Developer description with any role (e.g., Marketing Manager, Data Scientist, Sales Rep) by editing the prompt in the AI Qualification node.
• Different Storage: Replace Google Sheets with Airtable, Notion, or a database like PostgreSQL for storing applications and results.
• Multi-Round Screening: Add a Slack or email notification node to alert the hiring team when a candidate scores above a certain threshold (e.g., >0.8).
• Interview Scheduling: Connect the output to a Calendly or Google Calendar node to automatically invite high-scoring candidates for an interview.
• Batch Processing: Modify the trigger to accept a CSV upload of multiple CVs for bulk analysis.
• Enhanced Analysis: Add more output fields like "years of experience", "top skills", or "red flags" by updating the JSON schema in the Output Parser and the prompt.

{
  "meta": {
    "instanceId": "f230a542ff7e8110500e02e584e98cd66f8a922485dc0d875a2e022900b4e5da"
  },
  "nodes": [
    {
      "id": "821d52da-591f-44db-b218-ee3974855a05",
      "cid": "Ikx1Y2FzIFBleXJpbiI",
      "name": "Sticky Note - Create Spreadsheet1",
      "type": "n8n-nodes-base.stickyNote",
      "notes": "© 2025 Lucas Peyrin",
      "creator": "Lucas Peyrin",
      "position": [
        832,
        560
      ],
      "parameters": {
        "color": 7,
        "width": 272,
        "height": 272,
        "content": "Initializes a new Google Sheet with columns for applicant data. Can be replaced with Airtable, Notion, etc."
      },
      "typeVersion": 1
    },
    {
      "id": "72c7af0e-d8f8-4498-a849-b95c323a6cd6",
      "cid": "Ikx1Y2FzIFBleXJpbiI",
      "name": "Sticky Note - Append Row1",
      "type": "n8n-nodes-base.stickyNote",
      "notes": "© 2025 Lucas Peyrin",
      "creator": "Lucas Peyrin",
      "position": [
        1520,
        208
      ],
      "parameters": {
        "color": 7,
        "width": 320,
        "height": 288,
        "content": "Adds the applicant's name and email to the Google Sheet right after they submit the form to have it logged even if the workflow fails."
      },
      "typeVersion": 1
    },
    {
      "id": "b8076a35-fb6e-481d-9d92-3a95e9c3c732",
      "cid": "Ikx1Y2FzIFBleXJpbiI",
      "name": "Sticky Note - Extract Text1",
      "type": "n8n-nodes-base.stickyNote",
      "notes": "© 2025 Lucas Peyrin",
      "creator": "Lucas Peyrin",
// ... truncated (copy to see full JSON)