.png)
A conversation with Matt MacInnis on AI, global payroll, and why unified systems are starting to matter more than ever
When Matt MacInnis joined Rippling, it was not part of some long-term plan to return to HR tech.
In fact, it was the opposite.
“I swore I would never work in HR tech again,” he said. “Here I am back in HR tech.”
What brought him back was not the category itself, but the chance to work with Rippling founder and CEO Parker Conrad and to help build something fundamentally different.
That difference is now showing up in a meaningful way as AI begins to reshape how software works.
A lot of HCM vendors have been talking about AI for years. Most of it has been incremental.
Matt did not mince words on that.
“I don’t think what they have is anything other than glorified document summarization and the occasional simple data look up,” he said.
Rippling took a different approach. Instead of rushing something to market, the team held back.
“We’ve been kind of pulling our punch… we’ve been waiting to talk about AI because we want to make sure we have something that’s actually really useful and impactful.”
The result is a product that behaves less like a feature and more like a new interface to the system.
Matt described how it shows up in practice.
“I use it every day, like literally every day,” he said. “Almost in every meeting where someone make a claim… I just type into Rippling AI… and boom, I immediately get the answer.”
That shift moves the HRIS platform from something you navigate to something you converse with.
And in most cases, something that does the work for you.
The reason Rippling can do this comes down to how the system was built.
Matt broke it down into three pieces.
“Number one, you got to have access to all of the data in your organization,” he said. “Number two, you got to wrap it all in permissions… and number three, you got to put metadata on top of that information so that the agent can understand what it’s looking at.”
Each of those sounds straightforward. None of them are.
Most HCM systems were not designed this way. They were assembled over time, often through acquisitions, with different data models and user experiences stitched together.
Matt described those systems bluntly.
“They’re sort of Mr. Potato heads… if you’ve ever had to log into the system more than once… you know you’ve found a seam in the system.”
That fragmentation becomes a limitation in an AI-driven world. Without clean data, consistent structure, and a strong permission layer, the agent cannot safely or reliably produce information.
Rippling, by contrast, has been built on a unified model from the beginning. That architecture goes from notable to critical in an AI-first world.
One of the more interesting parts of the conversation was how AI can move beyond conversations and into workflows.
Matt pointed to a workflow like managing payroll changes across multiple countries, which previously took manual data entry in dozens of different payroll run screens.
“The rippling agent… you give it a CSV… off it goes,” he said. “It just does it.”
That type of workflow is difficult to replicate in systems that are not deeply integrated.
“In a system that’s not integrated the way ours is integrated, there’s just no hope of that happening,” he said.
Rippling made a bold bet in their early years to invest heavily on global payroll and workforce management.
That decision was not driven by short-term demand.
“We always played a long game,” Matt said. “We don’t need a quick duct tape solution.”
Instead of stitching together third-party tools, the team built native infrastructure in key markets and connected everything through a central system.
“It was very clear that companies were going to want to manage all countries in one system,” he said.
That approach is now paying off in a different way as AI enters the picture.
When all of that data and workflow lives in one system, the agent can operate across it seamlessly. If it does not, the experience breaks down quickly.
“You need to migrate to a unified system,” Matt said. “You’re just going to be left in the dust if you’re still dealing with this stuff in a year.”
Delivering a global experience goes beyond just getting the payroll calculations and tax withholdings right.
Matt pointed out that building for global teams requires more than copy-and-paste translations.
“You have to use the lingua franca of any given country in any given cultural context,” he said.
For Rippling, that means hiring local product and design teams, understanding regional expectations, and avoiding the subtle signals that make software feel out of place.
“If they are an employee of the company… it ought to feel like they’re a first-class member of your team.”
That includes how they appear in the system, how workflows apply to them, and how the product communicates with them.
We closed the conversation on the broader question that is dominating the market right now.
Is AI going to disrupt SaaS?
Matt’s answer was more measured than the headlines.
“The market is usually right in the long run,” he said. But he also framed the current moment as uncertainty rather than conclusion.
“It really reflects a retrenchment… a defensive posture… we don’t know what’s going to happen here.”
At the same time, he pushed back on the idea that existing software companies will simply be replaced.
“They’re going to fight tooth and nail,” he said.
And for companies that operate in more complex environments, the position may be stronger than it appears.
“If you move money, if you’re in a regulated business… these sorts of things… are very defensible positions.”
His view is that both sides of the ecosystem will matter.
Model providers will continue advancing the technology, while application-layer companies apply it to real workflows, real data, and real customer problems.
Rippling’s approach has been consistent since Conrad first coined the phrase the “compound startup.”
Build a unified system. Invest in infrastructure. Play the long game.
For years, that may have felt like a slower, more difficult path compared to assembling a platform through integrations or acquisitions.
Now it looks more like a prerequisite.
AI is raising the bar for what software needs to do. Systems that are fragmented or loosely connected will struggle to keep up.
Systems that are unified, permissioned, and structured have a different opportunity.
As Matt put it, “what a time to be alive.”
For companies that built with this moment in mind, it may also be the right time to pull ahead.
