Ryan is a payroll assistant at Watkins & Co — Sarah’s 45-person accountancy practice across three sites. He’s quick, enthusiastic and has fully embraced AI to speed up his work. Sarah thought he was a find.
For six months Ryan used an AI tool to process the monthly payroll. He fed in the data, checked the output looked about right and hit send. Every single month.
Last week someone finally looked properly.
Month 1: £800 in overpayments. Month 2: £1,200. Month 3: £1,400. Month 4: £1,800. Month 5: £2,100. Month 6: £2,300.
Total: £9,600. Six months. Nobody noticed.
Not a single error was enormous on its own. But together they add up to a serious recovery problem, a difficult conversation with multiple employees, and a payroll process that has to be rebuilt from scratch.
Ryan didn’t cut corners deliberately. He just used the tools available to him because nobody told him not to. There’s no AI use policy at Watkins & Co. No sign-off process for automated outputs. No second pair of eyes on anything Ryan produces.
The AI didn’t make a mistake. The process did.
✅ 3 THINGS TO PUT IN PLACE BEFORE AI TOUCHES YOUR PAYROLL
1. An AI use policy — which tools are approved, what they can be used for and who is responsible for checking outputs
2. A mandatory review process — no automated payroll output should be submitted without human sign-off
3. An overpayment recovery procedure — so when it happens you know exactly what you can recover, how and when
AI is not going away. But right now most businesses are letting staff use it without any framework — and the mistakes land on the employer’s desk, not the tool’s.
This is where People Vision HR comes in.
We help you build a simple AI use policy your team will actually follow, alongside the payroll checks and HR procedures that stop a small error becoming a £9,600 problem. Practical, plain English, built around your business.
Before Ryan’s next six months looks like his last six.