TECHNOLOGY
AI comes to payroll
Many people still view payroll as a last-century, manual and administrative process, involving a lot of data entry, clicking and generally being very operational. It’s 2025, and payroll is very different, as Mark Paraskeva, Chief Executive Officer and Founder, Paiyroll, tells us
B elow, I discuss three areas in which artificial intelligence (AI) is currently having an impact. Intelligent spreadsheet upload Pay professionals have known for a long time that uploading spreadsheets can significantly accelerate processing, reduce manual data entry and enhance accuracy. However, current spreadsheet uploaders only work if the spreadsheet is formatted exactly according to the payroll software. Typically, this means that existing spreadsheets cannot be used and must be edited or manually created from scratch. In contrast, AI enables payroll systems to read any spreadsheet format. This allows existing spreadsheets, such as those typically produced by a human resources or time and attendance (‘T&A’) system, as well as existing company / client spreadsheets, to be uploaded, dramatically reducing manual work for pay professionals. AI-automated payroll import One of the most complex challenges in payroll is switching to new payroll software. In addition to transferring all the data and configuration, parallel runs using the new system alongside the current software are essential. All this requires a significant amount of time. As payroll is an ongoing process, many pay professionals choose to switch at the beginning of the tax year, with less
data to migrate and less risk. Now, AI can automate payroll switching using a full payment submission (FPS) and an employee CSV file. AI can literally read and merge all the data, fill in missing gaps (like automatic enrolment) and create a parallel run in seconds. This can be done at any point in the tax year. Some of our clients have used AI to switch thousands of payrolls mid-year and verify each with a parallel run. Mid-year parallel run payroll import is the holy grail of payroll. “Mid-year parallel run payroll import is the holy grail of payroll” Data validation Pay professionals know that data quality is king when it comes to payroll. Importantly, AI can check and validate many data fields. Here are just a couple of examples: 1. In cases where incorrect National Insurance (NI) numbers are provided, employees may miss pension contribution years. AI verifies that NI numbers have the correct format and, if HM Revenue and Customs (HMRC) sends an NI number correction through the DPS (data provisioning service), it’s automatically
updated in the employee record. 2. The most crucial step in payroll is ensuring each employee is paid accurately. However, bank details are sometimes mistyped and are only identified when bank files are uploaded for payment. AI can be used to validate bank details directly at the point of entry at the start of the payroll cycle. This allows ample time for the data to be checked, flagged to employees for correction and amended, before payday. What this means for the role of pay professionals AI is powerful, but it's the pay professional who must specify how results should be computed. For example, starting or leaving salaries may need to be pro-rated using a salary scheme based on annual (365 days), working days (260 days), monthly calendar days or monthly working days. The same is true for holiday schemes: if an organisation has irregular-hour workers, it may want to pay 12.07% rolled-up holiday pay or holiday paid as it's taken, based on average pay over the 52 weeks preceding the holiday. There may be workers with a salary plus variable pay (e.g. commission or overtime), alongside pure salaried workers, which will require completely separate holiday schemes. The role now requires the pay professional to understand and specify the schemes. Workers can then book a holiday, with managers approving via
| Professional in Payroll, Pensions and Reward | November 2025 | Issue 115 64
Made with FlippingBook - Online magazine maker