AI-Native Global Payroll Platform Delivered Rapidly with Mendix

“The number one challenge is lack of scalability,” said Emmanuel Remy, chief revenue officer and co-founder of datascalehr.

“As payroll providers grow larger and larger, they struggle to protect their operating margins. As they add more countries and individuals, there is greater complexity in their activities, and the only solution is to put more boots on the ground to make up for the lack of software that can support their needs.”

In 2022, three industry experts set out to fill this gap in the payroll software market and started datascalehr. Their AI-enabled software streamlines multi-country payroll processes, allowing providers to shift their focus from technical implementation to improved customer experience.

To get their software to market quickly and remain both flexible and scalable, the datascalehr team developed their core product offering with the Mendix low-code platform in just 8 months.

A Platform That Works

Jerome Gouvernel, CEO and co-founder at datascalehr, saw the challenges of the industry firsthand during his 15-year stretch at one of the leading global payroll providers. Gouvernel was at the forefront of a team that was the first to even offer multi-country payroll for customers.

Multi-country payroll is incredibly expensive to establish and maintain. Payroll is handled differently in every country, so supporting hundreds of countries means differing deductions, social insurances, or taxes – all made more convoluted as geopolitical relationships become more strained.

At that same point in his career, Gouvernel was also charged with leading product strategy for an innovation lab with the same global payroll provider. “We discovered that the missing ingredient was a technical platform that was good enough to experiment, but also good enough to be production-grade,” he said.

Traditional ERP platforms were unable to meet these requirements, so Gouvernel’s team began experimenting with low-code. They evaluated vendors such as OutSystems and Salesforce, ultimately selecting Mendix.

Fast, Flexible, and Scalable

Fast forward to today and these parallel paths of software innovation and multi-country payroll converged for Gouvernel. “We realized that, even 20 years later, there are still unaddressed problems with the way that multi-country payroll is carried out by even the largest providers,” he said.

datascalehr was founded to create the world’s first pre-trained AI that “frees the payroll outsourcing industry from the shackles of code maintenance, data integration, and reconciliations.” The foundation for their software was one that had been pressure-tested by Gouvernel and his team all those years ago: Mendix.

Gouvernel and his co-founders, Nicolas Delord and Emmanuel Remy, saw that Mendix would provide the:

• Flexibility required to prototype and respond to customer needs as a startup.
• Reliability to scale their solution without performance issues.
• Compliance and security certifications – such as SOC2 and ISO – that were required of any credible player in the payroll industry.

“We have full trust in the platform. It’s certified and it does a much better job than we would be able to do independently in terms of hosting and uptime and failover and data security. Traditionally, a company like ours would just have no chance dealing with large enterprises if we didn’t have the Mendix Cloud to fall back on,” Zawadzki said.

datascalehr’s CTO and co-founder, Nicolas Delord, cites this completeness as critical to bring their flagship solution to market quickly with the required AI capabilities. “I was able to focus on the execution, not the technical tasks like maintaining frameworks or worrying about deployment,” said Delord.

Data Integration Made Easy

The datascalehr team estimates that payroll providers typically spend up to 75 percent of their research and development budgets just establishing the right connectivity to their client’s data and payroll partner network.

“Connecting is only half the battle,” Gouvernel explained. “You have to bring the payroll results back up the pipe and compare them back to the original instructions that were given. But by that point, the data from the payroll looks completely different from the original data because it has been transformed multiple times.”

With datascalehr’s software, establishing these connections can be done with minimal effort. Whenever the user — typically a payroll or HR team member — initiates a request, an AI model guides them through the steps from setting up to mapping and building connectors.

The team created the first iteration of their offering — the eponymous ‘datascalehr’ — in less than a year. The solution offers features such as:

• Seamless integration to any payroll system within minutes
• Efficient reconciliation of legacy files or systems
• Real-time tracking and validation of field changes
• Continuous monitoring of period-to-period variations
• Data synchronization across organizations’ ecosystems
• Secure email links for downloading data

“Building a system and UI that understands this mapping gets very intricate,” said Zawadzki. “Bringing AI into this process allows that knowledge to be reused for the next mapping. That’s what makes our platform unique, self-learning, sustainable, and agile.”

The Power of Generative AI and Low-Code

datascalehr’s AI journey began with machine learning. “We figured out how to use Machine Learning and started enriching it even more. We used all the standard ML algorithms, ones that are all open source. Then we applied them to our specific domain,” Gouvernel said.

Oftentimes people want AI to problem-solve out of the box. But AI engines are probabilistic; you can’t always use the answers they provide. “The key is to take the output and transform it into a deterministic outcome,” Gouvernel said.

Datascalehr’s AI model tackles larger problems by addressing smaller ones. It leverages patterns derived from inputs such as country, payroll system, and employee data. ML handles approximately 85 percent of the workload. Experts verify the ML’s work, and the system adapts its future responses accordingly.

The platform’s success hinges on how intuitive it is for subject matter experts to use. “datascalehr was originally ML, and now it’s a mix of ML and large language models (LLMs). The processes are so simple that any end user who understands a little bit about payroll can configure it themselves. Which means you don’t need to allocate software developers to do it,” said Gouvernel.

With Mendix’s support, datascalehr can swiftly transform information from an indeterminate state to usable.

“Now business users don’t have to go to a development team when a payroll result file contains a new paycode. We leverage the knowledge model to make the adjustment. The user can review and confirm the decision, and that’s how you build the knowledge model and solve the scalability issue,” said Delord.

Rapid Experimentation and Innovation

What datascalehr provides their customers is what Mendix provides for datascalehr: the ability to focus less on technical execution and more on solving problems for their customers. “I have time to focus on an idea and solve a business use-case, which makes more space for creativity and putting our customers first,” said Delord.

Introducing new features — like a reporting prototype and general ledger reporting — in response to customer feedback has been quick and efficient.

Often moving faster means sacrificing on quality. However, Mendix provides guardrails such as test automation and pre-built connectors to ensure governed development.

“I was pleasantly surprised that a test automation tool like Menditect just exists in the Mendix Marketplace,” Zawadzki said. The Mendix Marketplace offers datascalehr several other game-changing connectors to services like OpenAI and AWS Textract.

These tools become invaluable for datascalehr’s customers — who are standing up their multi-country payroll implementations an estimated 10x faster. “Ultimately, our customers don’t need us to maintain their product over time, because our goal is to create a sustainable model for them,” said Remy.

“Mendix’s ability to exemplify abstract concepts speeds up understanding by a factor of ten. When you talk about collaboration, what other platform could you do this on? Real collaboration is with your users, not between your developers,” Gouvernel said.

In the era of artificial intelligence and machine learning, Mendix is enabling businesses like datascalehr to deliver enterprise-level solutions in a rapid and cost-effective way. “At the end of the day, none of this would have been possible without Mendix,” Remy concluded.