📢 Job ads that attract the right people
A poor job ad wastes everyone's time — yours and every wrong-fit applicant's. Most job ads fail because they describe what the role does but not why someone would want it, or they use jargon that puts off exactly the people you're trying to attract.
AI can help you write job ads that are clear, human, and well-structured — but you need to give it the real details about the role and your organisation's culture.
Before (typical)
"The successful candidate will be responsible for the coordination and management of administrative functions including but not limited to..."
After (AI-assisted)
"You'll be the person who keeps things moving. Part admin, part problem-solver, you'll work across the team making sure..."
Write a job ad
Write a job advertisement for a [job title] at [organisation name], a [brief description of org, e.g. community arts trust in Raglan].
Role overview: [2-3 sentences about what the person will do]
Key skills needed: [list 4-5]
Culture: [describe your team / working environment]
Pay: [$X to $Y / negotiable / not specified]
Hours: [full-time / part-time / hours per week]
Make it sound like a real person wrote it — not a policy document. Lead with why someone would want this job, then what they'll do, then what we're looking for. Under 350 words.
Then ask: "Write 3 alternative opening paragraphs — one formal, one conversational, one that leads with impact." Pick the one that fits your org.
Inclusive language check
Review the following job advertisement for language that might unintentionally discourage applicants from underrepresented groups. Flag any gendered language, unnecessary jargon, or requirements that may not be essential. Suggest alternatives.
[paste job ad]
Why this matters: Research consistently shows that certain words (e.g. "rockstar", "aggressive", excessive credential lists) deter women and BIPOC candidates. AI can catch patterns humans miss.
Interview questions from a job ad
Based on the following job ad, write 8 structured interview questions. For each question, explain what it is designed to reveal about the candidate. Mix behavioural questions ("tell me about a time...") and situational questions ("what would you do if..."). Avoid anything that could be discriminatory.
[paste job ad]
Tip: Add "Include 2 questions specific to [sector/org type]" for more targeted results.
Exercise H.1
Write a real (or realistic) job ad
Use a role you're hiring for now — or a role you know well from your organisation. If you're not currently hiring, use a recent vacancy as a test case. The goal is to get a first draft in under 5 minutes.
- 1Fill in the job ad prompt with real details about the role
- 2Run it. Read it as if you were a candidate — would you apply?
- 3Run the inclusive language check on the output
- 4What would you change? Make one edit to the prompt and run it again
🚪 Onboarding that doesn't get ignored
Most onboarding documents are walls of text that new starters read once and never look at again. The information is usually there — it's the format that fails. AI can help you restructure existing content into something people will actually use.
The reformat trick: If you have an existing staff handbook or policy document, paste a section into Claude and ask it to rewrite it as a friendly, plain-English summary with headers and bullet points. You don't have to rewrite everything — just the bits people always ask questions about.
First-week checklist
Create a practical first-week checklist for a new [job title] joining [type of organisation]. Include:
- Day 1 essentials (who to meet, what to set up, where things are)
- Week 1 learning goals (systems to learn, documents to read)
- Week 1 tasks (things they should actually do, not just observe)
- Questions to ask their manager by end of week 1
Format as a clean checklist they can tick off. Friendly, practical tone.
Make it yours: Add "Our systems are [X, Y, Z]" and "Our team norms are [list]" to get something specific to your org.
Rewrite a policy as plain English
Rewrite the following policy section as plain-English guidance a new staff member can understand without a law degree. Keep the key requirements but make it readable. Add a "what this means in practice" section with 3 concrete examples.
[paste policy section]
Important: Don't use AI output as the legal policy itself — just as a companion guide for staff understanding.
Welcome email from the manager
Write a warm, personal welcome email from a manager to a new team member starting on [date]. The new person's name is [name], their role is [role]. Our team works on [brief description]. Include: a genuine welcome, what their first day will look like, one thing to look forward to about joining. Under 200 words. Don't make it sound like a template.
Personal touch: Add one real detail about your team's culture and the output shifts from generic to genuine.
📣 Internal communications people actually read
Internal comms has a trust problem. Staff have learned that "all-staff update" often means dense, top-down text they're supposed to care about but rarely do. The fix isn't better writing — it's shorter, clearer, more human writing. AI is good at this.
Staff update / all-staff email
Write a short all-staff update email from [sender name/role] at [organisation]. Key points to cover:
- [point 1]
- [point 2]
- [point 3]
Tone: warm, transparent, direct. Length: under 200 words. Structure: 1-sentence opener, 3-4 bullet points, 1-sentence closer with any action needed. This isn't a press release — it's a conversation with colleagues.
Tip: "Under 200 words" is the most important constraint — if it's longer, staff tune out.
Change announcement
Write a staff communication announcing [change — e.g. new process, restructure, policy update, team change]. The audience is [description of staff]. Key concerns they might have: [list 2-3]. We want to be transparent about [X] while emphasising [Y]. Acknowledge the impact on staff before explaining the rationale. Under 250 words.
Order matters: AI will put rationale first if you don't specify. People need to feel heard before they'll listen to reasons.
FAQ from a policy or change
Based on the following policy/announcement, generate a FAQ that answers the 8 questions staff are most likely to ask. Be honest — if something isn't decided yet, say so. Write the questions from a staff member's perspective, not management's.
[paste policy or announcement]
Then add: Your actual answers to the questions AI generates. The questions are what AI is good at — the answers need to be yours.
Performance review prompt starters
Generate 10 opening prompts a manager can use to start a constructive performance review conversation. Mix prompts for: strong performers, people who are struggling, and people who've had a difficult year. Each prompt should open a two-way conversation, not deliver a verdict. Keep each under 25 words.
How to use: These are conversation starters, not scripts. Adapt to the person in front of you.
Exercise H.2
Write a comms piece your team actually needs
Think of something your organisation needs to communicate to staff or the public in the next month — a change, an update, an announcement, or a recurring message you always end up rewriting from scratch. Draft it now using AI.
- 1Pick the comms piece: _______________
- 2Choose the prompt template that fits best, or write your own
- 3Run it. Does it sound like your organisation, or too generic?
- 4Add one specific detail that only your org would know — how does the output change?
⚖️ Where to draw the line in HR
HR work touches sensitive information more than almost any other function. Before using AI in your people processes, understand where the clear limits are.
Never put individual staff data into free AI tools
Names, performance issues, health information, disciplinary matters, pay details — none of this belongs in ChatGPT's free tier. If you need AI for sensitive HR work, use an enterprise plan with privacy guarantees, or anonymise all data first.
AI shouldn't make hiring decisions
Using AI to screen CVs or rank candidates creates legal exposure around algorithmic bias. AI can help you write job ads and interview questions — it shouldn't be deciding who gets through to interview.
Performance feedback needs a human hand
AI can generate conversation starters and frameworks — but the actual feedback about a specific person's performance must come from you. An AI-written performance review that the manager hasn't reviewed line by line is a liability.
Where AI genuinely shines in HR: Job ad drafting, inclusive language checking, policy plain-English rewrites, onboarding document formatting, staff update templates, FAQ generation. These are all structural / editorial tasks with no individual's sensitive data involved.
Exercise H.3 — The HR AI audit
Map what's safe vs what's not for your org
Every HR team has different risk exposure depending on org size, sector, and data sensitivity. Take 5 minutes to map your own situation.
- 1List 5 HR writing tasks you do regularly
- 2For each: does it involve personal data? If yes, mark it "proceed carefully"
- 3For the "safe" ones — write a prompt you could use right now
- 4For the "careful" ones — what would you need to anonymise before using AI?
HR & Comms module complete
You've got templates for job ads, onboarding, internal comms, and a clear picture of where AI fits in people work — and where it doesn't.