25 ChatGPT Prompts for HR and Recruiting (Job Descriptions, Interviews, Offers)

Most HR and recruiting prompt lists tell you how to write a "compelling job description" using ChatGPT. The result is usually a job description that sounds exactly like every other AI-generated job description: bullet points about being "passionate" and "detail-oriented" working in a "fast-paced environment."

This list is built around the prompts that produce HR copy people actually finish reading. Job descriptions, interview kits, candidate communication, offer letters, and internal docs. None of it sounds like a 2014 corporate handbook because that's what HR has been trying to escape for ten years.

Why most HR prompts return generic copy

HR writing fails when it's written for the org chart instead of for a real person reading it. Job descriptions written for hiring managers list internal duties. Job descriptions written for candidates name what the candidate will actually do, what skills they'll grow, and what their day will look like. Same content, different framing. Candidates open the second one and skip the first.

The prompts below force ChatGPT into the candidate-first or employee-first framing. Each one names who's reading, what they need to learn, and what specific output to produce. Without those constraints, ChatGPT defaults to template-speak. With them, you get copy that does the job.

How to use these ChatGPT prompts in HR

Paste the prompt verbatim. Replace bracketed placeholders with the role specifics, the company stage, the candidate experience level, and any unique constraints. If the response sounds corporate, paste this follow-up: "Cut anything that sounds like HR jargon. Write it like a person explaining the role to a friend who'd be good at it." That single line strips out 90% of the template language.

A real concern about employee privacy: do not paste candidate names, employee names, salary specifics for individuals, performance review content, or anything that could identify a real person to ChatGPT. Use descriptions instead: "a senior engineer who's been with us 3 years and is asking about promotion path." That gives ChatGPT enough to work with and protects the people involved.

Job Descriptions (Prompts 1-5)

1. Job description that doesn't sound like every other JD

Prompt: "Write a job description for [role title] at a [company stage: startup / scaleup / public]. The role exists because [why it exists, in 1-2 sentences]. Day-to-day: [list 3-5 actual tasks, not duties]. Skills required: [list 3-4 hard requirements]. Company stage details that matter: [team size, what stage of growth]. Avoid 'rockstar,' 'ninja,' 'passionate,' 'fast-paced.' Open with what the person will work on, not the company description."

The "open with what they'll work on" rule fixes the default JD structure (which buries the actual job under three paragraphs of company-speak).

2. JD bias check

Prompt: "Audit this job description for biased language: [paste]. Specifically flag: gendered words ('rockstar,' 'guru,' 'aggressive,' 'nurturing'), age-coded language ('digital native,' 'recent graduate'), unnecessary requirements that filter out qualified candidates (years of experience that don't match seniority, degree requirements for non-degree work), and ableist language. For each flag, suggest a neutral alternative. List in priority order."

Run this on every JD before posting. The fixes are usually 3-5 minutes and the impact on applicant pool diversity is well-documented.

3. Salary range copy that doesn't anchor low

Prompt: "Write a salary range section for a [role] in [city or 'remote-US']. The internal band is [paste range]. Output: a 2-3 sentence section that 1) shows the range, 2) explains briefly what determines where in the range a hire lands, and 3) names what's included beyond base. No 'competitive salary' nonsense. No ranges with implied ceilings ('up to X')."

Pay transparency is mandatory in many states now. This produces compliant, candidate-friendly copy.

4. JD for a role that's hard to describe

Prompt: "I need a JD for a role that's a hybrid of [function A] and [function B]. The hire needs to [primary outcome] but with the [secondary skill] required to [why secondary matters]. Output: a 200-word JD that names this hybrid clearly so candidates self-select. Don't make it sound like two jobs in one (which scares people off). Make it sound like one role with a clear shape."

Hybrid roles are the hardest JDs. This prompt is the one that gets the most use.

5. JD translation to candidate-friendly language

Prompt: "Here is an internal JD that uses our company's jargon: [paste]. Rewrite it in candidate-friendly language. Replace internal acronyms and tool names with their generic equivalents (and add the specific tool in parentheses if it's a known one). Translate org-chart language ('reports to VP of...') into 'works closely with' framing. Keep all technical requirements; just translate the soft language."

Internal JDs make sense to internal people. Candidates need translated versions.

Interview Question Banks (Prompts 6-11)

6. Behavioral interview kit for a specific role

Prompt: "Generate 6 behavioral interview questions for [role] focused on [3 competencies, e.g., ownership, written communication, prioritization]. For each question: the prompt, what a strong answer would include, what a weak answer might sound like, and 1 follow-up that probes deeper. Avoid 'tell me about yourself' style questions."

The "what a weak answer might sound like" is what makes the interviewer calibrated.

7. Technical screening for a non-technical interviewer

Prompt: "I'm a non-technical hiring manager screening candidates for [role]. Generate 5 questions I can ask in a 30-min phone screen that distinguish 'real' from 'fake' technical knowledge without me needing to evaluate the technical depth. Each question should produce an answer that's easy for me to assess on a 'sounds confident vs. evasive' axis."

Lets a recruiter or non-technical lead screen out obvious mismatches before scheduling a technical loop.

8. Reverse interview prep for a candidate's questions

Prompt: "I'm interviewing for [role] at [company stage]. The candidate asks me [paste candidate's question]. Write a 4-sentence answer that's: honest, specific, doesn't oversell, and lands the interesting parts of working here. If the question has any landmines (compensation, layoffs, why is this role open), give me the version that's transparent without legally exposing the company."

The "transparent without legal exposure" line is what hiring managers actually need.

9. Take-home assignment design

Prompt: "Design a 90-minute take-home assignment for a [role] candidate. Skills it tests: [list 3]. Output: 1) the prompt I send the candidate, 2) what a strong submission looks like, 3) red flags in submissions, 4) follow-up questions to ask in the readout. The assignment should be doable in 90 minutes; don't sneak in 4 hours of work."

The "doable in 90 minutes" check is what stops scope creep and candidate dropout.

10. Reference check questions that get real answers

Prompt: "Write 6 reference check questions for a [role] candidate. Goal: get past 'they were great' platitudes. Mix: 2 questions about specific situations the candidate handled, 2 questions that invite balanced feedback ('what would they need to develop in their next role'), 1 'would you hire them again' calibration, 1 'what's something I should know that I haven't asked.' Frame so the reference feels safe being honest."

The 'safe being honest' framing is the entire game in reference calls.

11. Interview debrief template

Prompt: "Write a structured debrief template for the hiring panel after interviewing [role] candidates. Sections: signal alignment (what each interviewer thought we were testing for), strengths observed (specific examples), gaps observed (specific examples), risks if hired, recommendation (clear: yes / no / depends on what). Cap each section at 100 words. Avoid 'culture fit' as a category; replace with specific behaviors."

The "no 'culture fit'" replacement is one of the most impactful hiring process changes a team can make.

Candidate Communication (Prompts 12-17)

12. Outreach to a passive candidate

Prompt: "Write a cold outreach message on LinkedIn to a candidate currently at [company], position [title]. Our role: [role title at our company]. Make it personal: reference 1 specific thing about their public profile (a project they shipped, a post they wrote, a conference talk). Don't say 'I came across your profile.' Don't say 'reaching out because.' 60 words max. End with a question, not a CTA."

The banned openers are the ones that get auto-deleted.

13. Rejection email after first round

Prompt: "Write a 4-sentence rejection email to a candidate we screened but won't move forward. Tone: respectful, concrete, not corporate. Include 1 specific thing we appreciated about their candidacy if there's anything genuine to say (don't fake it). Don't promise to keep them in mind 'for future opportunities' unless we actually will. End with one practical thing (e.g., feel free to apply for X role in 6 months when they have more Y experience)."

The "don't fake it" rule keeps the rejection respectful instead of generic.

14. Rejection email after final round

Prompt: "Write a 6-sentence rejection email to a candidate we interviewed for [role] across [N] rounds. They were strong but we hired someone else. Acknowledge the time they put in. Be specific about something they did well in the interviews. If we want to stay in touch for a future role, say so concretely. If we don't, don't pretend. Offer to give them feedback on a 15-min call if they want it."

The 'offer feedback' option separates good employer brands from disposable ones.

15. Status update during a slow process

Prompt: "Write a 3-sentence status update to a candidate whose interview process has stalled internally for [time period]. Tone: respectful, not apologetic, factual. Tell them where the process actually stands (not 'we're still considering'). Give a specific next-step time if you have one. If you don't have an update, say so directly."

Most candidates ghost when communication ghosts. This prevents that.

16. Offer letter cover email

Prompt: "Write the email body that accompanies an offer letter for [role] to [candidate first name placeholder]. Tone: warm, but not gushing. Cover: brief congrats, the highlights of the offer (role, comp, start date, key benefit), an invitation to ask questions before signing, a clear 'we want this to be the right fit' framing. Under 150 words. Don't apply pressure on timing."

The "don't apply pressure" is the candidate-experience tell.

17. Counter-offer response

Prompt: "A candidate we offered came back asking for [their ask: more comp / different start / different title / equity adjustment]. Our flexibility on this is [paste guidance]. Write a 4-5 sentence reply that acknowledges their request, names what we can and can't do specifically, explains the constraint briefly without sounding defensive, and confirms whether the original offer still stands or has been adjusted. Avoid 'unfortunately.'"

Counter-offer replies are weirdly hard to write. This formula gets it right consistently.

Onboarding and Internal Comms (Prompts 18-21)

18. First-week onboarding plan

Prompt: "Write a first-week onboarding plan for a new [role] hire at [team size + stage] company. Day-by-day breakdown (Mon-Fri). Each day: 1 person to meet, 1 doc to read, 1 small task to complete (something they can ship in under 2 hours), 1 'don't worry about this yet' callout to manage overwhelm. Tone: welcoming but specific, not generic."

The "don't worry about this yet" callouts are the secret to first-week experience.

19. Manager check-in script for new hires

Prompt: "Write a 30-min check-in script for a manager to use with a new hire at the 30/60/90 day marks. Each version has 5 questions: 1 about whether the role matches expectations, 1 about energy and overwhelm, 1 about gaps in onboarding, 1 about emerging strengths or interests, 1 open-ended. Avoid 'how's it going' as the opener."

Saves 30-min weekly meetings worth of unstructured 'how's it going.'

20. Layoff or org change announcement

Prompt: "Help me draft an internal announcement about [type of change: org restructure, layoffs, leadership change]. Tone: factual, not euphemistic, not cold. The announcement should: 1) state what's happening clearly in the first sentence, 2) explain why if there's a real reason (not 'positioning for growth'), 3) name what changes for people specifically, 4) name what stays the same, 5) include time and place for follow-up Q&A. Maximum 300 words. No 'difficult decision' opening."

These are the hardest internal comms to write. The 'no difficult decision opener' is what separates respect from corporate-speak.

21. Promotion announcement

Prompt: "Write a 100-word internal announcement about [employee placeholder]'s promotion to [new title]. Reference 2 specific things they accomplished (placeholders for now). Make it clear what's changing about their scope. Don't open with 'thrilled to announce.' End with one personal note about working with them (placeholder for the manager to fill in)."

Generic promotion announcements signal a company that doesn't pay attention. Specific ones build culture.

Performance and Development (Prompts 22-25)

22. Performance review draft

Prompt: "Help me draft a performance review for [role] employee. Inputs I'll paste: 5-7 specific accomplishments and 2-3 areas of improvement (anonymized, in my own notes). Structure: 1 paragraph on impact, 1 on strengths, 1 on growth areas, 1 on goals for next period. Be specific (cite the accomplishments). Be honest about growth areas without sandbagging. Avoid 'consistently' as filler. End with 1 question for the employee to reflect on."

The 'avoid consistently' instruction strips out filler. Performance reviews live or die on specificity.

23. Constructive feedback for a difficult conversation

Prompt: "Help me prepare for a difficult feedback conversation. The pattern: [describe the behavior, anonymized, in 1-2 sentences]. The impact: [describe what it's affecting]. The desired change: [what 'better' looks like]. Output: a 5-sentence script for me to open the conversation with. Use SBI format implicitly (situation, behavior, impact). Don't sandwich. End with a clear ask."

The "no sandwiching" rule is what makes the feedback land instead of confuse.

24. Career growth conversation prep

Prompt: "I have a career-growth 1:1 with a [role] employee at [tenure]. They've signaled interest in [direction]. Generate: 4 questions to ask them, 3 questions they're likely to ask me, and 2 frameworks for thinking about their growth (e.g., skills vs. scope, manager track vs. IC track). Tone: collaborative, not directive. End with a 1-sentence reminder of what's NOT a promise."

The 'what's not a promise' guard rail keeps career convos honest.

25. PIP (performance improvement plan) draft

Prompt: "Help me draft a PIP for [role] employee. Inputs (anonymized): the pattern, the impact, what specific behaviors need to change, what 'success' looks like in measurable terms, what support we'll provide. Output: a 1-page PIP with: summary of the issue, specific expectations (3-5, measurable), check-in cadence, support resources, and consequences if expectations aren't met. Tone: factual, not punitive. No legalese. (Reminder: have legal review before sending.)"

PIPs are legally sensitive. Use the output as a structural draft only; don't send without HR + legal review.

Tips for getting better HR copy from ChatGPT

A handful of patterns matter most.

Always specify the company stage and team size. A 10-person seed-stage startup has different needs than a 500-person scaleup. Without that input, ChatGPT defaults to mid-market enterprise tone, which fits neither extreme.

Cut "passionate," "fast-paced," "team player" everywhere. These words have lost their meaning. The prompts above mostly ban them. Consider banning them in your full org's writing.

Treat first drafts as drafts, especially for sensitive comms. Layoff announcements, PIPs, terminations, and internal restructures should never go out as raw AI output. ChatGPT can give you the structural skeleton; the actual words need a human pass for tone calibration to your culture.

Run anything candidate-facing through a second human. A second pair of eyes catches the unintentionally cringey line that AI sometimes produces.

Check legal compliance. Pay transparency, accommodations, classification language, and termination phrasing have legal implications. ChatGPT does not know your jurisdiction's specifics. Double-check.

FAQ

Will candidates know my emails are AI-written?

Often, if you ship the first draft. With a 10% human edit (cut a corporate phrase, add one specific detail about their background or our role, fix the tone of one sentence), almost never. The prompts above are calibrated to land closer to that second version.

Should I use ChatGPT to write performance reviews?

For the structural draft and the language tightening, yes. For the actual judgment calls about an employee's performance, no. The substance has to come from you and the manager. ChatGPT just helps the writing be specific and concrete.

Is it OK to paste candidate resumes into ChatGPT?

Generally no. Resumes contain PII (name, contact, work history, sometimes address). Don't paste them. Use descriptions: "a candidate with 5 years of experience at consumer apps, currently at [stage] company, interested in [reason]." Strip names and identifying details.

Can ChatGPT help with diversity and inclusion language?

For surface-level audits (gendered words, age-coded language, unnecessary requirements), yes. For deeper DEI work (programs, policies, equitable evaluation rubrics), no. That work needs domain expertise that ChatGPT doesn't have.

What about HR analytics and people data?

ChatGPT can help interpret tables of HR data if you paste them in. But: scrub the data of PII first, treat the analysis as one input among many, and verify any specific numbers it cites against your source data. Don't outsource judgment to it.

What to try this week

Pick three prompts. Use them on real work this week (a JD you owe, an interview kit you're scoping, a rejection email you've been dreading). Notice which ones produce shippable output and which ones miss. Keep the ones that work.

The HR teams that get value from AI aren't using more prompts. They're using fewer, better-crafted ones over and over.

For employer-brand content, social posts, and recruiting marketing, the 30 ChatGPT prompts for marketing covers ad copy and content briefs. If you hire engineers and want to understand the work better, the 25 ChatGPT prompts for software developers gives you technical context. And if you handle agent hiring at a brokerage, the 25 ChatGPT prompts for real estate agents covers what they actually do day-to-day.

Bookmark this page. We update it as the patterns we use shift.