14 AI agent prompts for HR & People

Copy-paste, beginner-friendly prompts to help you onboard new hires, answer policy questions and write job posts. Each one is based on a real-world example and works with ChatGPT, Claude or an AI agent. 13 are free below, and 1 more is in AgentPlaybook Pro.

Free HR & People prompts

Create an onboarding plan for a new [role] at my [company type]: first-day checklist, week-1 schedule, 30-60-90 day goals, and a welcome email. Tools they'll use: [list].

Write a job description for [role] at [company]: summary, responsibilities, must-haves, nice-to-haves, salary range [x], and 5 screening questions with what a good answer looks like.

Here are anonymous employee survey comments: [paste]. Group them into themes, estimate sentiment for each, quote representative comments, and suggest 3 actions leadership could take.

Here's this pay period's payroll data: [paste or attach]. Here's last period's for comparison: [paste]. Flag anyone with missing hours, rates, bank details, or tax info, plus any pay that changed by more than [X]% from last time. Put it in a table with a suggested question to ask for each.

Here are my open roles with posting date, target fill date, and candidates at each stage: [paste]. Flag roles at risk of missing their date and explain why. For each, suggest 2 actions (for example, widen the search, reword the post, or follow up with stalled candidates) and draft the post edit or outreach message.

Here is the job description for [role]: [paste]. Create a structured interview plan for [45/60] minutes with: sections and timings; one realistic hands-on task where the candidate can use AI tools, based on our real work: [describe]; sample questions and follow-up probes for each section; and a 1-4 scoring rubric for three AI-fluency skills: Framing (how well they define the problem and give the AI context), Steering (how they guide, refine and redirect the AI), and Judging (how they review and validate what the AI produces). Show what a strong vs weak answer looks like, and flag anything in the plan that's too long or could be unfair to candidates.

Build a talent market intelligence brief for this role. Job description: [paste]. Target locations: [list]. Our pay band if known: [paste or say unknown]. Research comparable titles, typical experience, and public compensation ranges for each location. Assess how deep the talent pool looks (scarce / moderate / plentiful) and name 3–5 competitor or peer employers who hire similar roles. Finish with 5 sharp questions I should ask the hiring manager in the intake meeting. Cite the kind of public source you'd use for each pay figure (job boards, salary sites, government data). Mark anything you're unsure about; don't invent exact salaries.

You are a first-round screening voice agent for [role]. Must-ask questions and knock-out criteria: [paste]. Scorecard dimensions: [e.g. role fit, communication, experience depth, culture signals]. After the candidate answers, produce: a 5-bullet summary, scores 1–5 per dimension with evidence quotes, stack-rank vs other candidates in this batch [paste prior scores if any], and a recommend (advance / maybe / reject) for a human recruiter. Never invent resume facts. Never make an offer, discuss compensation bands beyond [public range], or reject legally protected characteristics. Save only as a draft scorecard until a recruiter reviews.

Help me prepare this pay period's payroll before I submit it. Pay period: [dates]. Employees and rates: [list]. Hours and tips: [paste spreadsheet rows]. Bonuses, reimbursements or one-time deductions: [list]. Time off taken: [list]. Check everything and give me: (1) a summary table per employee with gross pay, (2) a list of anything that looks wrong or missing (overtime, missing hours, PTO over balance, new hires without details), (3) total labor cost compared with [budget], and (4) the questions I need to answer before running payroll. Do not submit anything; I approve payroll myself.

You are an employee service assistant helping me offboard someone. Employee: [name, role, team, manager]. Last day: [date]. Systems and apps they use: [list]. Company equipment: [laptop, phone, badge, other]. Our offboarding policy: [paste steps/approvals]. Step 1: build a checklist grouped by team (IT, HR, facilities, manager) with an owner and a due date for each item, based only on our policy. Step 2: list which access removals need approval and from whom. Step 3: draft short messages to each owner asking them to complete their item. Step 4: draft the employee's farewell/next-steps note (equipment return, final pay questions to HR, any letter they requested). Do not remove access or send anything yourself, and mark any step not covered by the policy as 'confirm with HR'.

Here is a training document we use: [paste or attach the guide, checklist or slide text]. It was written for [original country/market] in [language]. Please adapt it for [new country/market] in [target language]. Local details I know: [currency, units, laws or rules that differ, local holidays, job titles, anything else]. 1) Translate it in plain, friendly language a new hire would understand, keeping the same headings and order. 2) Replace examples, names, prices and units with local equivalents and list every change you made in a table (original, new, why). 3) Flag any step that may conflict with local laws or customs, and mark it 'check with a local manager' instead of rewriting it. 4) List any words or brand terms you kept in the original language on purpose. 5) Add a short 5-question quiz at the end in the target language to check understanding. Only use the local details I gave you; don't guess at local laws or rules.

Help me rediscover past applicants for a role I'm hiring for now. New role: [paste the job description, location, pay range and must-have skills]. Past applicants: [paste or upload a list or spreadsheet with name, role applied for, date, key skills, notes from earlier interviews, and why they weren't hired]. Do this: 1) List the must-haves and nice-to-haves from the new role. 2) Score each past applicant against them (strong / possible / no fit) and give a one-line reason based only on what's in their record. 3) Skip anyone marked "do not contact" or who was let go for conduct. 4) Show the top [5-10] as a table: name, original role, date applied, fit score, reason, anything to check. 5) For each strong match, draft a short, warm message that mentions their earlier application and invites them to apply or chat. Don't guess at skills that aren't in the record, and don't use age, gender, race or other protected traits in any ranking.

You are helping me turn my business know-how into workflow documents an AI agent can follow. My business: [e.g. specialist recruitment for finance roles in Sydney]. The jobs I want agents to take on: [e.g. finding new clients, market updates, keeping candidates warm, running a search]. Start with this one job: [name it]. Interview me one question at a time until you understand: the trigger that starts it, every step I take, the tools and data I use, how I decide (with real examples of good and bad cases), the hard rules (e.g. only candidates with the right to work in [country]), the tone of anything sent under my name, and what 'done well' looks like. Then write the workflow doc with: purpose, inputs, numbered steps, decision rules, things the agent must never do, a draft-for-approval point before anything goes to a client or candidate, and a checklist I use to review each output. Finish with a log template where I record each mistake and the rule we add to stop it happening again. Then ask which job to document next.

More in AgentPlaybook Pro

• Policy Q&A assistantAnswer employee policy questions from your handbook.

See Pro pricing →

Guides that include HR & People prompts

Other categories