The problem

An HR team in Rabat receives 300 applications for one open role. Reading every PDF takes days, good profiles get lost in the pile, and candidates wait weeks without an answer — which hurts the employer brand.

The solution, step by step

  1. Centralize applications: every CV lands as a PDF in one Google Drive folder (from email or a form).
  2. An n8n workflow watches the folder, extracts the text of each new PDF.
  3. ChatGPT receives the CV text plus the job description and returns a structured evaluation: score out of 100, matched skills, missing skills, seniority estimate.
  4. Important guardrail: instruct the model to ignore name, gender, age and photo and to justify every score with quotes from the CV.
  5. Results are written into Airtable: one row per candidate, sortable by score, with the justification visible.
  6. HR reviews the top 20 manually, and a polite automatic acknowledgment email is sent to every applicant.

    7. Result and estimated gains

    • Screening time drops from 4 days to about 1 hour for 300 applications.
    • Every candidate gets an answer, protecting the company's reputation.
    • More consistent first-pass evaluations, with a written justification for each score that HR can verify.