"Why not just ask ChatGPT or Claude to write your HR documents?"
It's a fair question, and we'd rather answer it honestly than pretend AI doesn't exist. You can open ChatGPT or Claude right now, type "write me a disciplinary policy", and get something back in thirty seconds that looks professional. So why pay for an official template?
Because looking professional and being legally sound are two different things. And in HR, the gap between them is where tribunal claims live.
01
The problem isn't the writing. It's the law.
AI tools are genuinely good at writing. The sentences will be clean, the structure sensible, the tone convincing. That's exactly what makes them risky for HR documents, the output looks finished, so nobody checks whether it's right.
Most large AI models are trained predominantly on American content. Ask for an employment policy and you'll often get US concepts dressed in UK spelling: at-will employment thinking, references to procedures that don't exist here, or terminology that has no legal meaning in a UK tribunal. Unless you already know UK employment law well enough to spot the errors, you won't spot the errors. And if you knew it that well, you wouldn't need to ask ChatGPT.
AI tools also invent things. They're language models, not lawyers, they predict what a plausible answer looks like, which means they can fabricate legal principles, cite tests that don't exist, or misapply the ones that do. AI-generated letters routinely get statutory redundancy and notice pay calculations wrong. Those aren't style errors. They're the kind of mistakes that cost real money and end up in front of a judge.
02
You're handing employee data to a public tool
There's a risk that is real before the document is even written. To get a useful disciplinary or grievance letter out of AI, you have to tell it about the situation which usually means typing in the employee's name, the allegations, salary details, absence history, or health information.
That's special category personal data going into a public AI platform, often one that processes it outside the UK and may use it to train future models. Do that without a lawful basis and proper safeguards and you're looking at a UK GDPR breach, a confidentiality problem, and potentially an ICO complaint all before anyone has even read the letter. The employee whose disciplinary details ended up in a chatbot has a grievance of their own now.
A template flips this on its head. The document comes to you complete, and the personal details never leave your business. You fill in the highlighted fields yourself, on your own system.
03
Employment law doesn't stand still
UK employment law is changing faster right now than at any point in a generation. The Employment Rights Act 2025 is reshaping day-one rights, dismissal rules, and statutory entitlements, with changes phasing in over the next two years. ACAS codes get updated. Case law shifts what "reasonable" means in practice.
An AI model has a training cut-off. It doesn't know what changed last month, and it won't tell you it doesn't know, it will confidently give you the position as it stood whenever its training data ended. A policy built on last year's law can be worse than no policy at all, because it commits you in writing to a process that no longer protects you.
Our templates are drafted against current UK legislation and the ACAS codes of practice, and we update them when the law moves. That's not a feature. It's the whole point.
04
When something goes wrong, who stands behind the document?
This is the question that matters most. If a dismissal ends up at tribunal, your documents become evidence. A judge will look at your policy, your letters, and your process, and ask whether they were fair and legally compliant.
If AI wrote your policy, nobody stands behind it. There's no professional who reviewed it, no accountability, and no way to show it reflected the law at the time. You carry the entire risk of every sentence, including the ones you didn't know were wrong.
There's also a judgement problem an AI can't solve. It doesn't know your workplace, the employee's history, previous warnings, what you've allowed to slide before, or how the ACAS codes of practice apply to your specific situation. Fairness in employment law is contextual, and context is precisely what a chatbot doesn't have. It will write you a confident letter for a process that, in your circumstances, isn't fair.
Our documents are written and reviewed by HR professionals who deal with real disciplinaries, real grievances, and real tribunals. When you use one, you're not just getting words on a page, you're getting the judgement of people who know how these documents perform under pressure, with guidance notes that tell you when a situation needs advice rather than a template.
05
Documents don't work in isolation
An HR document suite is a system. Your disciplinary policy needs to match your disciplinary letters. Your absence procedure needs to align with your sickness policy and your return-to-work forms. Definitions, timescales, and processes have to be consistent across all of them, because inconsistency is exactly what a claimant's solicitor looks for.
Ask an AI for documents one at a time and you get one-off answers that don't know about each other. One policy gives five days for an appeal, the letter says seven. One document defines gross misconduct one way, another defines it differently. Each document might look fine alone; together they contradict each other.
Our templates are built as a coherent suite. They reference each other correctly, use consistent language, and follow the same process from policy through to final letter.
06
So is AI useless for HR? No and we won't pretend it is
AI is a brilliant thinking partner. Use it to summarise a long policy, prepare questions for a meeting, or get your head around a topic before you speak to a professional. Used that way, it saves time and sharpens your thinking.
Just don't use it as the source of truth for documents that carry legal weight. The right tool for that is a template drafted for UK law, kept current, reviewed by humans, and designed to work as part of a complete system with clear guidance on how to adapt it safely to your business.
07
AI arguing against AI — with nobody checking the facts
As HR practitioners we encounter a fair amount of AI slop being produced by employers that can look convincing but doesn't hold up when challenged. The detail is wrong, and that creates problems that didn't need to exist.
Increasingly, employees use AI to challenge documents that employers wrote with AI in the first place — so you get AI arguing against AI, with nobody checking the facts.
We Value Your Privacy
We use cookies to keep the site working and, with your permission, to understand how it is used and to personalise our marketing. Read our Privacy Policy.