It's Friday afternoon and a client MD rings. He needs a disciplinary invitation letter by Monday, an updated hybrid working section for the handbook and, while he's got you, a quick one: is it all right if his managers write job adverts in ChatGPT? HR consultancies have a double problem with AI. You need it to keep up with the work, and your clients increasingly expect you to tell them how they should be using it.
Start HR consultancy AI training by setting firm rules on employee data, then use AI for first drafts of policies, letters and manager guidance built from your own precedents. Add casework organisation and plain-English advice next. Once your own practice is solid, turn it into a client service: AI policies, manager guidance and workforce planning.
Why HR consultancies need their own AI training
Your clients' staff are using AI already, policy or no policy. CIPD's Labour Market Outlook for autumn 2025 found employees in 76% of UK organisations use AI tools at work, and 54% of employers said staff use free versions of tools like ChatGPT, Claude and Gemini. That's a lot of employee information heading somewhere nobody has checked.
HR people are more confident with the tools than with the strategy. CIPD research from early 2026 found 67% of HR practitioners felt confident learning to use AI tools, but only 40% of HR leaders felt confident designing reskilling pathways for the change AI brings. That second job is the one clients will pay a consultancy to do.
Training is patchy too. A General Assembly survey of US and UK HR professionals found 82% use AI at work, while 26% had received no formal AI training at all. A consultancy that can show its working has a genuine edge.
- 01Protect the dataSpecial category data stays out.
- 02Draft from precedentsYour templates, AI's speed.
- 03Organise caseworkChronologies, not conclusions.
- 04Translate the advicePlain English for managers.
- 05Sell the policyYour rules become a service.
- 06Plan the workforceRoles, skills, reskilling.
What should an HR consultancy learn about AI first?
1. Treat employee data as the line you don't cross
HR files hold the most sensitive information a business has: health, grievances, disciplinary history and things an employee told one person in confidence. Much of it is special category data under UK GDPR. None of it belongs in a free AI tool.
Set your rules before anybody builds a habit. Approved tools only, business accounts with appropriate data settings, names and identifying details removed before any case material is used, and a short list of things that never go in at all. Every consultant should be able to say those rules back to you without looking.
Clients will ask how you handle this. A one-page answer you can email them is the start of trust and, handily, the start of a product.
Check your progress
2. Draft policies and letters from your own precedents
This is where the hours come back. Give AI your consultancy's approved template, the client's specifics and the tone you want, and you'll get a first draft of a policy section, a contract variation letter or a hearing invitation in a couple of minutes.
Work from your precedents, not the internet's. A model asked to write a disciplinary policy from scratch will produce something that sounds right and may not match current UK law or the client's own procedure. Employment law moves, sometimes at speed. Your template carries the judgement. AI does the typing.
HR prompt to try
Using our template below, draft a letter inviting an employee to a disciplinary hearing. Insert the details in square brackets only, keep the template's legal wording unchanged and use a calm, neutral tone. List any details you still need from me before this can be sent.
Check your progress
3. Use AI to organise casework, never to decide it
Investigations produce paperwork: witness notes, email chains and timelines that refuse to stay in order. With identifying details removed, AI can build a chronology, group statements by allegation and list the questions nobody has answered yet. On a long case that's a real help.
It must not decide what happened. Findings, sanctions and judgements about credibility belong to the investigator, and the reasoning has to be theirs to defend at a tribunal. Train consultants to ask AI for structure and gaps, and never for conclusions.
HR prompt to try
From these anonymised witness notes, build a dated chronology of events. For each allegation, list which notes refer to it and any contradictions between them. Do not reach any conclusion about what happened.
Check your progress
4. Turn expert advice into something a line manager will read
Your advice is usually right and occasionally ignored, because the manager stopped reading at the second subordinate clause. AI is good at rewriting advice in plain English at different lengths: a two-line message for the supervisor on the shop floor, a one-page note for the MD, a script for a difficult conversation.
Keep the substance identical. Check that simplifying hasn't shifted the meaning, especially around process and timings. Then send the version your reader will actually finish.
Check your progress
5. Package AI policy and guidance as a client service
Once your own house is in order, you're well placed to help clients sort theirs. Most small and mid-sized employers have staff using AI and no written rules. They need an acceptable use policy, guidance for managers and a sensible conversation about what's allowed.
AI can help you turn your client interview notes into a first draft, but the value is your judgement about their risk, sector and culture. Use our AI policy guide with 15 questions as a starting checklist, and read who's liable when AI gets it wrong before you write the paragraph on accountability.
Check your progress
6. Help clients plan roles and skills around AI
The bigger conversation is about jobs: which tasks in each role are changing, which skills the team needs next and how to reskill people rather than replace them. That's HR's home ground, and AI can help with the legwork by breaking job descriptions into tasks and mapping them against current skills.
Our AI skills matrix template is a good place to start. Keep people in the room. A workforce plan written entirely by a model, about the people in a business, tends to land about as well as you'd expect.
HR prompt to try
Break this job description into its main tasks. For each task, say whether AI tools are likely to change it in the next two years, how, and what new skill the employee would need. Present it as a table. Don't suggest removing the role.
Check your progress
Which HR tasks are safe to use AI for?
A simple traffic-light guide for your consultants. Print it, argue about the amber row and then agree it.
| HR task | AI's role | Rating |
|---|---|---|
| Job adverts and role profiles | Drafting from approved templates | Green |
| Policy and handbook sections | First drafts from precedents, reviewed by a consultant | Green |
| Manager guidance and FAQs | Plain-English versions of your advice | Green |
| Investigation chronologies | Organising anonymised notes only | Amber |
| Sickness, health or grievance details | Kept out unless there's a lawful basis, safeguards and client agreement | Red |
| Outcomes, sanctions and dismissals | Never decided by AI | Red |
Green means go, with normal checks. Amber means anonymise and review carefully. Red means a person, a process and no shortcuts.
A practice task for your consultants
Choose one policy you update often, such as hybrid working or annual leave. Ask two consultants to update it for a fictional client: one by hand, one using AI with your template. Compare time taken, accuracy against current guidance and how much editing each version needed.
Then write a one-page note for clients explaining how your consultancy uses AI and what it never does. Test it on a friendly client and see which questions come back. Those questions are your FAQ.
Train your consultancy with us
Our AI Policy Workshop helps consultancies write their own rules and gives you a structure to reuse with clients. For hands-on skills, Basic AI Training covers safe everyday use, and if you run recruitment for clients, AI for Recruiters covers adverts, sourcing and screening. The wider plan for client teams is in how to train your staff to use AI.
Based in Witney, Oxfordshire. Explore UK training options or take the free two-minute AI skills assessment to see where your team stands.
Talk to us about your consultancyAI training for HR consultancies: frequently asked questions
Can HR consultants use ChatGPT with employee data?
Only with care and the right setup. Use an approved business tool, remove names and identifying details, and keep special category data such as health information out unless you have a lawful basis, proper safeguards and client agreement. When in doubt, leave it out.
Can AI write a disciplinary policy?
It can draft one, but don't rely on that draft. Start from your consultancy's approved precedent, have AI adapt it to the client and have a qualified consultant check it against current law and the client's procedures.
Should HR consultancies offer AI policy services?
It's a natural fit. Clients need acceptable use policies, manager guidance and training, and they already trust their HR adviser on people risk. Use AI in your own practice first so your advice comes from experience.
Will AI replace HR consultants?
It will replace some of the typing. Clients pay for judgement about people, risk and fairness, and demand for that is rising as employers work out how AI changes their roles. Our guide to whether AI will replace staff goes further.
What should an HR consultancy learn about AI first?
Data rules first, then drafting from your own templates, then casework organisation and plain-English advice. Client services come once your team uses AI confidently and safely on its own work.
Sources and further reading
- CIPD Labour Market Outlook, Autumn 2025
- CIPD: Futureproofing your skills, AI is changing how HR skills are applied (2026)
- General Assembly: The AI gap in HR (2025)
Statistics, standards and programme links checked on 5 October 2026.