PortuguêsEspañol

AI Prompts for HR in 2027: 18 Tested Prompts for Job Posts, Interviews, Reviews and Rejections

OBy Onne Company · Published
Quick answer: The AI prompts for HR that save the most time are job descriptions with a bias check, resume review with quoted evidence, structured interview guides with rating anchors and respectful rejection emails. Below are 18 tested prompts that work in ChatGPT, Claude, Gemini or Copilot, with real outputs.
AI Prompts for HR in 2027: 18 Tested Prompts for Job Posts, Interviews, Reviews and Rejections
In this article
  1. Before you paste: the rule behind every HR prompt
  2. Prompts for job postings and sourcing
  3. Prompts for screening
  4. Prompts for structured interviews
  5. Prompts for offers, rejections and onboarding
  6. Prompts for performance and development
  7. Prompts for engagement and policies
  8. Prompts for offboarding
  9. Common mistakes with HR prompts
  10. Want to skip this part?
  11. Frequently asked questions
  12. Sources

Here are 18 AI prompts for HR, from the job posting to the exit interview. They work as ChatGPT prompts for HR, and just as well in Claude, Gemini or Microsoft Copilot. Replace everything in [BRACKETS], and never paste a resume with a home address, a Social Security number or medical details into a personal AI account. Use a business plan where your data is not used for training.

Before you paste: the rule behind every HR prompt

Every HR prompt that works gives the AI three things it cannot guess:

  1. Real details. Pay range, schedule, benefits, the competency names your company uses. Without them, the AI invents a 401(k) match you do not offer.
  2. The law as a guardrail. "Flag anything that could discourage applicants based on age, disability, sex, race, religion or national origin." Without it, the AI copies the bias from your old posting.
  3. The AI's limit. "Do not decide. Organize." Without it, the AI ranks people by writing style.

The full map of where AI fits in HR, and where it creates legal risk, is in the AI for HR guide.

Prompts for job postings and sourcing

1. A job description with a bias check

When to use: every new req, or an old one that keeps pulling the wrong applicants.

Write a job description for [JOB TITLE] at [COMPANY: industry and size], [ON-SITE/HYBRID/REMOTE] in [CITY, STATE].
Use only these real details: pay range [RANGE], benefits [LIST], schedule [SCHEDULE], tools [SYSTEMS].
Deliver: title, 2-line summary, up to 6 responsibilities, up to 4 true requirements (only what is essential to do the job), nice-to-haves and benefits.
Before delivering, flag any wording that could discourage applicants based on age, disability, sex, race, religion or national origin, and remove it. Note if the role's states require a pay range in the posting.

2. A LinkedIn post that sounds like a recruiter, not a brochure

Turn this job description [PASTE] into a LinkedIn post under 120 words, written by me (the recruiter) in first person. First line: the real problem this hire will solve. Include the pay range and how to apply. No "join our amazing team," no more than one emoji.

Real output for a Payroll Coordinator role in Columbus, Ohio:

Our payroll runs for 340 people every other Friday, and right now one person carries all of it. I'm hiring a Payroll Coordinator to share that load and own the questions employees bring to HR every week ("why did my withholding change?"). On-site in Columbus, Monday to Friday, 8 to 5. US$ 52,000 to US$ 60,000, medical, dental and vision, 401(k) with a 4% match, 15 PTO days. You'll work in ADP Workforce Now, so experience there helps. If you've run multi-state payroll or cleaned up a messy garnishment process, I'd like to hear about it. Apply through the link in the comments or send me a message.

3. Knockout questions that only knock out the right people

Write 5 objective screening questions for [JOB TITLE] that only filter out applicants who do not meet these true requirements: [LIST]. No questions about age, graduation year, family, health, religion, citizenship beyond work authorization, or criminal history. For each question, state the answer that disqualifies and why it is job-related.

4. Outreach message to a passive candidate

Write a LinkedIn InMail under 90 words to [ROLE AT CURRENT COMPANY] for our [JOB TITLE] opening. Mention one specific thing from their profile: [DETAIL]. Say the pay range [RANGE] up front. End with a low-pressure question, not "let's hop on a call." No flattery.

Prompts for screening

5. Resume review with evidence, not a score

When to use: a few resumes, read one at a time. For hundreds, use the AI screening agent for recruiters.

Compare the resume below (name, address and graduation years removed) with these criteria: [LIST].
For each criterion, answer "yes", "no" or "not stated", with the exact quote from the resume.
Do not evaluate personality, writing style, employment gaps or anything not on the list.
At the end, list 3 interview questions about what was "not stated".
Resume: [PASTE]

6. Finalist comparison for the hiring manager

Build a table comparing these [NUMBER] finalists for the [DEPARTMENT] hiring manager: relevant experience in years, results with numbers, what to confirm in the final interview. Use only my notes below. No adjectives, no ranking, no recommendation.
[PASTE NOTES]
Advertisement

Prompts for structured interviews

7. A 45-minute structured interview guide

When to use: when several managers interview for the same role. Standardizing questions is the cheapest way to cut bias in the human part of hiring.

Build a 45-minute structured interview guide for [JOB TITLE] assessing: [3 TO 4 COMPETENCIES].
For each competency: 2 behavioral questions in STAR format, 1 follow-up probe, signs of a strong answer, signs of a weak answer and a 1-to-4 rating anchor.
Add a 5-minute opening and a closing for the candidate's questions. No questions about age, family plans, health, religion, national origin or hypotheticals like "where do you see yourself in 5 years".

Real output, one competency for a Customer Success Manager role:

Competency: handling an unhappy customer
Q1: Tell me about a time a customer was close to canceling. What was the situation, what did you do in the first 48 hours and what happened? Q2: Describe a time you had to tell a customer "no" on something they cared about. How did you say it and what was the result?
Probe: What would you do differently with that same customer today?
Strong answer: names the customer's real problem, describes specific actions with dates, gives a measurable outcome (renewed, expanded, churned and why).
Weak answer: talks about "always putting the customer first" with no example, or blames the product team for everything.
Rating: 1 = no example; 2 = example without own actions; 3 = clear actions and outcome; 4 = clear actions, outcome and what they changed in their process afterward.

8. Interview scorecard summary

Here are the scorecards from [NUMBER] interviewers for [CANDIDATE ID]: [PASTE]. Summarize where interviewers agree, where they disagree by more than one point, and which ratings have no example written next to them. Do not give an overall recommendation.

9. Reference check script

Write a 15-minute reference check script for [JOB TITLE] focused on [2 COMPETENCIES]. 6 questions max, each asking for a specific example. Include how to open the call and what not to ask (health, family, protected characteristics, anything outside the job).

Prompts for offers, rejections and onboarding

10. Rejection email after an interview

Write a rejection email to a candidate who interviewed for [JOB TITLE]. Thank them for [SPECIFIC THING FROM THE PROCESS], say we chose candidates with more experience in [GENERAL AREA], and invite them to apply again for [TYPE OF ROLE] if it fits. Under 110 words, warm and direct. No details from interview notes, no "we'll keep your resume on file" unless true.

Real output:

Hi Jordan, thank you for the time you gave us over two interviews, and for the walkthrough of how you rebuilt your team's onboarding checklist. It was one of the clearest answers we heard. We've decided to move forward with a candidate who has more experience running multi-state payroll, which is a big part of this role in the first year. I know that's not the news you wanted. If a People Operations or HR generalist role opens here, I'd be glad to see your name again. Best of luck with your search, Dana

11. Offer call talking points

Write talking points for a verbal offer call for [JOB TITLE]: base pay [AMOUNT], bonus [DETAILS], start date [DATE], benefits highlights [LIST]. Include how to answer "is there room on the salary?" given our limit of [MAXIMUM]. Do not ask about their current or past salary.

That last line matters: many states and cities ban salary history questions.

12. 30-60-90 day onboarding plan

Create a 30-60-90 day plan for a new [JOB TITLE] reporting to [MANAGER ROLE]. The manager's real priorities: [LIST]. Tools they will use: [LIST]. For each period: 3 goals with a way to measure them, people to meet and one early win. Format as a table.

Prompts for performance and development

13. Performance review draft from facts

Turn these notes into a performance review draft in the manager's voice: goals [LIST], results with numbers [LIST], specific examples [LIST], peer feedback [PASTE]. Sections: strengths, areas to improve, goals for next period. Use only these facts. Every statement needs an example. No generic phrases like "strong team player."

14. PIP structure

Draft a 60-day performance improvement plan for [JOB TITLE]. Gaps: [SPECIFIC GAPS WITH EXAMPLES]. For each gap: measurable expectation, support the company will give, check-in dates. Neutral tone, no judgment of character. Do not include anything about health, leave or complaints the employee has made.

After the draft, you check one thing the AI cannot: whether the PIP follows closely on a leave request, an accommodation request or a complaint. That timing is what turns a PIP into retaliation evidence.

Prompts for engagement and policies

15. Coding survey comments by theme

Below are [NUMBER] anonymous comments from our engagement survey (names and team names removed). Group them into 6 to 10 themes. For each theme: a one-line description, how many comments, 2 representative quotes and whether it is mostly positive or negative. Do not guess who wrote anything. Flag any comment that mentions harassment, discrimination or safety in a separate list.

16. Handbook policy in plain English

Rewrite this policy [PASTE] at an 8th-grade reading level for employees. Keep every rule and every number exactly as written. Add a 3-question FAQ at the end using only what the policy says. If anything is unclear in the original, list it for me instead of guessing.

Prompts for offboarding

17. Termination meeting script

Write a script for a 10-minute termination meeting for [JOB TITLE]. Decision is final; reason: [BRIEF, FACTUAL]. Include: opening line, the decision in one sentence, final pay and benefits details [LIST, including COBRA notice timing], return of equipment, what happens today. Calm, respectful, no debate about the decision, no extra reasons.

18. Exit interview themes across a quarter

Here are notes from [NUMBER] exit interviews this quarter, with names removed: [PASTE]. List the top 5 reasons people left, how many mentioned each, which teams or managers come up more than twice and 3 questions leadership should answer. Only what is in the notes.
Advertisement

Common mistakes with HR prompts

  • Short prompts. "Write a job description for an accountant" gives you a posting for nobody. Real details make the difference.
  • Asking the AI to pick the best candidate. It will, confidently, based on how the resumes are written. Ask for evidence and decide yourself.
  • Trusting legal answers. Use AI to explain a rule you already confirmed, not to find out what FMLA or your state's final pay law says.
  • Personal accounts with employee data. Business plan, training off, identifiers removed.
  • Sending the first draft. Ask for a second version ("cut it in half," "less formal") and read it as the person who will receive it.

If you notice you run the same five prompts every week, that is a sign they can become an automatic routine, like the screening agent.

Ready-made AI agents for HR professionals

Want to skip this part?

Ready-made AI agents for HR: nothing to build, no prompts to write, they already work for you.

  • Screens hundreds of applications with quoted evidence per criterion and a shortlist for your review
  • Sends respectful, on-time replies to every candidate who did not move forward
  • Answers candidate questions about the role and the process around the clock
I want the ready-made agents

Frequently asked questions

What is the best ChatGPT prompt for HR?

The job description prompt with a bias check saves the most time with the least risk. It gives the AI your real pay, schedule and benefits and tells it to flag wording that could discourage protected groups.

Can I paste resumes into ChatGPT?

Only on a business plan where your data is not used for training, and after removing addresses, graduation years and any sensitive details. Never in a personal free account.

Can AI write interview questions?

Yes, and it is one of the best uses. Ask for behavioral questions with a follow-up probe and a 1-to-4 rating anchor, and ban personal or hypothetical questions in the prompt.

Can AI answer HR policy and leave questions?

Only from your own handbook and plan documents, with a rule to send anything unclear to HR. General answers about FMLA or state leave laws are often wrong.

Sources

O

Onne Company · the Núcleo Aprendizado team

We test, compare and explain what is worth learning, with a focus on what gets results at work. How we work.

Read next