HR: Job Postings, Resume Screening, Onboarding

In small and medium-sized businesses (SMBs), human resources rarely fall to a specialist. The founder drafts job postings between meetings, or a generalist staff member discovers resume screening the day sixty applications land in their inbox. ChatGPT won’t turn you into an HR director, but it professionalizes your processes without requiring a new hire.

Write Attractive Job Postings

Job postings from SMBs suffer from two opposite problems. Some run three lines—“Looking for a salesperson, send your CV”—and attract only mass applications. Others copy a corporate template ill-suited to a twelve-person company and ring false from the first sentence. The fix is to give ChatGPT the material those two extremes skip: your actual size, your concrete responsibilities, and your salary range.

“Write a job posting for a [job title] role in an SMB of [size] in the [industry] sector, based in [city]. Key responsibilities: [3–5 responsibilities]. Desired profile: [skills and experience]. What we offer: [concrete benefits—no foosball tables]. Salary: [range]. Tone: professional but authentic, reflecting the culture of a dynamic SMB. The posting should be appealing without overselling.”

Then run three checks. First, screen for discriminatory criteria: ChatGPT can flag risky phrasing, but you must always reread to eliminate any reference to age, gender, origin, or any other factor the law prohibits from consideration. Second, watch for internal jargon, since a candidate won’t know what your homegrown acronym means—ask for a rewrite aimed at an external reader. Third, hunt for overpromises like “hyper-growth company with unlimited prospects”: candidates spot that tone in a second, and it costs you the best profiles.

Screen Applications Without Improvising

Screening becomes defensible once it rests on a rubric written before you read the first resume. Ask for it explicitly:

“I’m hiring for a [role]. Here are the 5 essential criteria: [list]. Create a 5-point evaluation rubric for each criterion, with descriptions for each level (1 = insufficient, 5 = excellent).”

Next, paste the resume text or import it, then request an evaluation criterion by criterion, with each score justified, plus an overall score and a recommendation: interview, waitlist, or reject. This analysis remains preliminary. The final decision is yours, and this rule isn’t just formal: a model reproduces biases from its training data, and fully automated screening silently filters out atypical profiles who often make the best hires in SMBs. It’s also a legal requirement: the GDPR (Article 22) gives everyone the right not to be subject to a decision based solely on automated processing that significantly affects them, and Recital 71 specifically cites online recruitment without any human intervention. The EU AI Act further classifies systems designed to analyze and filter job applications as high-risk; their obligations apply starting December 2, 2027, as set by the Omnibus Regulation of July 2026.

Then there’s the part everyone postpones: responding to rejected candidates. It takes two minutes with the right prompt:

“Write a kind rejection email for a candidate applying for the [job title] role. We appreciated [positive point]. The reason for rejection is [reason—phrased positively]. Encourage the candidate to apply again if a future opportunity better matches their profile.”

Structure Onboarding

A hire is won as much after signing as before. Many SMBs leave new hires to fend for themselves the first week, then wonder why they quit after three months. A written onboarding path changes that trajectory:

“Create a 4-week onboarding program for a new [role] in an SMB of [size]. Week 1: discovery. Week 2: product training. Week 3: guided autonomy. Week 4: first objectives. For each week, detail day-by-day activities, meetings to schedule, and expected deliverables.”

Complete it with a welcome booklet gathering the company description, your values, and practical information—hours, remote work policy, tools used, key contacts—on four pages maximum, in a warm and practical tone.

Other HR Areas

The same reflex applies to annual reviews, where ChatGPT produces evaluation rubrics, question frameworks, and summaries. It turns your existing procedures into internal training materials, drafts your memos and internal newsletters, and checks your documents for consistency with labor law—though this last check must always be confirmed by a lawyer.

Two Reflexes

GDPR first: a resume is personal data; don’t send it without a legal basis, inform candidates, and reserve this work for a Business or Enterprise plan. Tone next: never publish a job posting straight from ChatGPT, because a craft-based SMB doesn’t address candidates like a tech startup.

Your Turn

Test this on your next hire. Write the posting with the prompt above, build the corresponding evaluation rubric, then test the analysis on two or three applications, keeping the final decision for yourself.

Official sources: Regulation (EU) 2016/679, Article 22 (CNIL, France’s data protection authority) · Regulation (EU) 2024/1689, Annex III · Regulation (EU) 2026/1744 (EUR-Lex) · ChatGPT Business overview (OpenAI Help Center)

Key Takeaways

• ChatGPT professionalizes your HR processes even without a dedicated specialist.
• Job postings, resume screening, onboarding: every step gains significant time.
• The final decision always remains human, especially in hiring.
• Comply with GDPR: use a Business or Enterprise plan for resumes, inform candidates, and never issue a purely automated rejection.