How can HR consultancies use AI?
HR consultancies get the most from AI on repeated document work: keeping policies and staff guides current, checking onboarding paperwork for gaps, and drafting first versions of standard letters, job adverts and onboarding packs. Keep people on every output and take extra care, because employee data is personal data.
This is general information, not legal advice. HR work often involves sensitive personal data, so check your own obligations or ask a data protection adviser.
Where should an HR consultancy start with AI?
Start with document work that is repetitive and easy to check. Four good first jobs:
- Policies and staff guides. Handbooks, procedures and how-to guides go out of date the moment the way people work changes. AI is good at redrafting a guide when you tell it what changed.
- Checking paperwork. New starter forms, right-to-work documents and contracts all need checking for missing or inconsistent items before anyone relies on them.
- First drafts of standard letters. Invitations, outcomes and reminders follow a pattern. AI drafts, an HR professional edits and sends.
- Job adverts and onboarding packs. Give AI the role, the pay band and the team, and it can put together a job advert straight away. Give it a new starter's role and start date, and it can assemble a first onboarding pack: a welcome letter, a first-week schedule, a checklist of forms and a list of who to meet. A person reads and adjusts each one before it goes to the candidate or the new starter.
Avoid, at first, anything that makes or influences a decision about a named person. That needs a human at every step.
What does this look like when it is built?
Two things I have built map straight onto HR work, shown here as illustrations of the method rather than HR projects I have delivered:
- Guides that keep themselves up to date. When the way you work changes, the change is spotted, the guides and help pages are rewritten, and you approve before it goes live. Typically about 2 hours a week per person (see the example).
- New customer checks. Everything a new customer sends is checked, and anything missing or wrong is flagged before it causes a problem (see the example). The same pattern fits checking a new employee's paperwork.
Both stop at a person's approval. Nothing goes out without your OK.
What about data protection?
Employee records are personal data, some of it special category, so UK GDPR applies whichever AI tool you use. The ICO publishes guidance on AI and data protection. Its page says it was last updated on 15 March 2023 and is under review because of changes made by the Data (Use and Access) Act, so treat it as a starting point.
Practical habits that follow: use a business plan where training on your data is off by default, do not paste identifiable records into a personal account, and tell staff and clients where AI is used. Can AI tools see my emails and files once connected? has a plan-by-plan comparison.
Questions people also ask
Can AI write a job advert or an onboarding pack?
Yes, and it is one of the quickest wins. A clear brief (the role, the team, what the person will do in their first weeks) gives you a usable first version in minutes. Check the wording for anything that could discourage a group of applicants, and check dates, names and contract terms before anything is sent.
Can AI write HR policies?
It can draft them well from a clear brief, but a person with HR and legal knowledge must check them. Treat the output as a first version, and keep a record of who approved the final text and when.
How can AI help with staff handbooks and procedures?
It can redraft sections when something changes and keep wording consistent. The useful habit is a single source of truth that the AI updates from, with a person approving each change before it replaces the old text.
Is it safe to put employee information into ChatGPT or Claude?
Only on a plan where your data is not used for training by default, and only what the task needs. Avoid identifiable special category data such as health information unless your data protection adviser agrees.
Should AI be used in hiring decisions?
Use extreme care. Using AI to draft job adverts or summarise documents is low risk. Using it to rank or reject people raises fairness and data protection issues, so keep a human making the decision and check ICO guidance first.
See it working
Things I've built that do this, and the time they give back.
Guides that keep themselves up to date
When the way you work changes, your how-to guides and help pages update themselves.
Typically 2 hrs a week per person
See how it works →New customer checks
Everything a new customer sends you is checked, and anything missing or wrong is flagged before it causes a problem.
See how it works →