The HR Prompt Library: Eight Prompts You Can Use Today

HR teams handle the most sensitive workflows in the company: hiring, feedback, conflict, reorganizations. Yet they often receive the most generic AI advice. "Use ChatGPT to save time" says nothing about preparing a difficult conversation or rewriting a policy nobody reads.

Here are eight prompts built for real HR work. Two rules before you start: adapt the bracketed placeholders to your context, and never paste personal or identifying employee data into a public AI tool. Anonymization is not optional, it comes first.

1. The job description that attracts the right people

Act as an experienced HR advisor. Write a job description for a [role] at a [size, sector] company. Context: [team, main challenges, why the role is open]. Structure: mission, 5 key responsibilities, essential skills, desirable skills, what we offer. Tone: direct and human, no corporate jargon. Before writing, ask me 3 questions if anything essential is missing.

When to use it: when you open a role and the existing description is a copy of a copy.

What good output looks like: a description that sounds like your company, with specific, verifiable responsibilities and no filler.

2. The structured interview guide

Build a structured interview guide for a [role]. The 4 competencies that matter most: [list]. For each competency, give 2 behavioral questions based on past situations, 1 follow up question, and 3 concrete signals of a strong answer. Add 2 warning signs to watch for across the interview. Format the result as a table I can print.

When to use it: before any hiring round, so every candidate is assessed against the same criteria.

What good output looks like: questions anchored in the candidate's past, not hypotheticals, and observable signals rather than gut feel.

3. The onboarding plan

Create a 30, 60 and 90 day onboarding plan for a [role] joining a team of [number] people. Their manager has [little / plenty of] time available. Include: goals week by week, key people to meet, quick wins, and 3 checkpoints with the questions the manager should ask. Keep everything on one page.

When to use it: every new hire, especially when the manager is stretched thin.

What good output looks like: a plan usable from day one, with checkpoints the manager can realistically keep.

4. The difficult conversation rehearsal

I need to prepare a difficult conversation with an employee about [situation, no personal data]. Play the role of the employee and respond as they realistically would. After each of my replies, give me brief feedback on my tone and clarity, then continue the role play. Start by asking me how I plan to open the conversation.

When to use it: before a high stakes conversation, when you want to test your approach without consequences.

What good output looks like: a role play that challenges you credibly, and feedback that points at specific phrasings to improve.

5. The policy in plain language

Rewrite this internal policy so any employee understands it on first read: [paste text]. Rules: short sentences, no legal jargon, concrete examples for the 3 most common situations, and a 5 line summary at the top. Flag anything that remains ambiguous so I can clarify it with legal.

When to use it: remote work, expenses, leave, any document nobody reads because it is unreadable.

What good output looks like: a text a new employee understands alone, plus a list of grey areas to settle before publishing.

6. The employee survey analysis

Here are anonymized answers from our internal survey: [paste answers]. Identify the 5 main themes, rank the themes by frequency, and quote 2 verbatim answers per theme. Separate clearly what people say from what you infer. End with 3 questions the leadership team should discuss.

When to use it: after every survey, when open answers pile up unread.

What good output looks like: themes backed by real quotes, and a clean line between facts and interpretation.

7. The performance feedback draft

Turn these rough notes into clear, fair performance feedback: [paste notes]. Structure: what is working, what needs to change, expected next steps. Every point must refer to an observable behavior, not a personality trait. Tone: honest and respectful. If any point lacks a concrete example, ask me for the detail before writing.

When to use it: before annual or interim reviews, when your notes are fragments.

What good output looks like: feedback the employee can receive without feeling attacked, because every point refers to a fact.

8. The internal announcement

Draft an internal announcement about [change]. Audience: [who is affected]. Likely worries: [list]. Structure: what changes, why now, what it means for each person, where to ask questions. Length: under 250 words. Tone: honest, no spin. Do not oversell any benefit.

When to use it: reorganizations, new tools, policy changes, anything that will generate questions at the coffee machine.

What good output looks like: an announcement that names the worries instead of dodging them, and says where answers live.

These prompts do not replace HR judgment. They prepare the ground: you validate, adjust and own the result. That is exactly the Knowledge dimension of the MAKIA framework: capability does not come from the tool, it comes from how people use it.

Try this week

Pick one real HR document, a job description or an internal policy. Run the matching prompt on it, compare the output with the current version, and note three differences with a colleague. It takes less than 30 minutes and shows you exactly where AI helps, and where your judgment still does the work.