When you are one of country’s longest established and most trusted payroll providers, how do you go about incorporating artificial intelligence into your offering?
With an innovative, trust-first approach that aims to solve the time-consuming tasks that payroll teams face every day and helps helps support compliance with local regulations, says Richard Kenyon, Associate Director of Datapay AI Labs.
Since it was established 60 years ago, Datacom has evolved its payroll system to meet changing business needs and technological advancements. Datapay – which was custom-built for the local market, filling a gap not met by global providers – is a cloud-based solution tailored for medium to large enterprises, used by government and private sector customers and valued for its flexibility, security, and robust support from local payroll experts.
With the rise of large language models (LLMs), the Datapay team saw an opportunity to introduce an AI assistant designed as a natural language interface, allowing payroll professionals to ask complex questions in plain English and receive concise, authoritative answers based on legislation.
For example, using the Datapay AI-powered Payroll Assistant, a payroll manager can ask: “How do I pay someone correctly who’s just returned from long-term leave with public holidays in the mix?” and receive a tailored, accurate response in seconds, rather than the hours or days such queries often require.
Users can inquire about legislative requirements, draft correspondence, or rewrite policies, all within a secure environment.
“People now understand that there are two types of AI,” says Kenyon. “There’s consumer AI, like ChatGPT, which is good for finding a recipe or composing a haiku. Then there’s enterprise AI, which is built for purpose-specific roles inside organisations,” he adds.
Datapay Payroll Assistant falls into the latter category, drawing only on internal information and official databases of legislation and regulations relevant to employment law and remuneration.
“It's not just drawing on someone’s blog post, saying, here's how I think the law should be interpreted,” Kenyon says.
“This is published, authoritative and curated information, and because our system is so good at combining that information, customers get very reliable, concise, plain English and complete answers.”
The assistant’s initial release focuses on providing access to curated, up-to-date payroll legislation for New Zealand and Australia and future functionality will allow users to securely upload and cross-reference their company documents, such as employment agreements.
Given Datacom’s decades-long reputation for reliability, the company has taken a rigorous approach to ensure the AI-powered assistant delivers useful information based on real legislation.
The initial version of the assistant does not connect to actual payroll data, reflecting customer hesitation about exposing sensitive payroll information to AI systems, even with robust protections in place.
The development process began with intensive testing by Datacom’s head of compliance, a payroll veteran with 25 years’ experience. He challenged the system with a battery of both simple and complex questions, scoring its responses for accuracy and completeness. Any gaps identified were addressed through fine-tuning and further testing.
The next phase involved Datacom’s Managed Services payroll team, 40 professionals handling payroll for various clients, who beta-tested the assistant in real-world scenarios. Their feedback was instrumental in refining the system, particularly in making the tool accessible for users whose first language isn’t English or who have challenges like dyslexia.
This diverse testing base ensured the assistant could handle the full spectrum of payroll queries encountered in practice. Datacom also implemented a robust process to keep legislative content current. The system is integrated with government feeds, pulling in updates to relevant laws and guidance documents nightly. Any changes trigger internal reviews to ensure the AI’s knowledge base remains authoritative and up-to-date, a critical factor in maintaining compliance and customer trust.
The first raving fan for Datapay's AI-powered Payroll Assistant was Datacom’s own head of payroll.
“We gave her a demo and she typed in a question about an employee who had just come back from broken periods of unpaid leave,” says Kenyon.
“She gave the Payroll Assistant the dates and conditions and hit ‘go’ and 20 seconds later, it gave her the right answer. Then she asked, ‘Can you write a letter to our employee outlining this?’ A few seconds later, it was done.”
Previously, that task could have taken several days. Kenyon sees the combination of natural language processing and machine learning as a productivity booster that will become central to every payroll department.
“It’s like building a house,” he explains. “You don’t use a hammer and nails anymore, you use a nail gun. But you make sure you train the person so that they don’t nail their foot to the floor.”
The current Payroll Assistant is just the beginning. Datacom’s roadmap includes integrating the assistant with actual payroll data, enabling users to ask questions like, “What has Mike been paid in the last month?” or “What would be the impact of increasing his hourly rate by $5?” This next phase will be “utterly transformative,” Kenyon says, but will only proceed once customers are fully comfortable with the technology’s security and accuracy.
Future plans also include advanced agents for anomaly detection — flagging unusual payroll patterns — and planning tools that can prioritise complex tasks for payroll teams. Perhaps most exciting is the prospect of “explainable payroll,” where employees can receive line-by-line, plain-English explanations of their pay packets, reducing confusion and support calls.
Employee self-service is another priority, with the vision of allowing staff to enter timesheets or leave requests via natural language, further reducing administrative burdens.
Kenyon envisions AI as an enabler of “intelligent automation”, not just automating repetitive tasks, but providing decision support, surfacing insights, and freeing payroll professionals to focus on higher-value activities.
“Payroll people are already productive; it’s the tools that make them less so,” he says.
“By releasing the capacity tied up in manual processes, we allow them to add more value, whether that’s optimising pay, offering financial guidance, or simply reducing stress and errors."
The release of Datapay's Payroll Assistant marks a significant step in making payroll smarter, more transparent, and more human-centric — one carefully engineered answer at a time.
Listen to MinterEllisonRuddWatts' Tech Suite podcast for an in-depth interview with Datapay’s Richard Kenyon on the new Datapay AI Assistant.
Datapay's AI-powered Payroll Assistant helps enterprise payroll teams cut research time and stay on top of compliance requirements and changing legislation.
Payroll is complex. The Payroll Assistant helps you manage complexity by delivering clarity, speed and trusted answers to New Zealand payroll legislative and regulatory related questions, directly inside Datapay.
A live knowledge base combines the Holidays Act, Income Tax Act and your own policies. Our local compliance team updates it as regulations change, so today’s guidance matches tomorrow’s audit needs.
