[00:00]
Andrew Brown: You know, maybe you hate your current cell phone provider, but you love Verizon when you switch. In that case, you'd be in the top right quadrant. Payroll is the bottom left. You're really unhappy with your current provider and you're really unhappy once you switch to a new one. And it's just like, what are we doing here, guys? This just shows how broken the current ecosystem is. And that's the gap.
[00:31]
Ansel Parikh: Welcome to another episode of The Current, a bi-monthly podcast exploring the intersection of people, finance, and data. I'm Ansel Parikh, co-founder of Finch, the connectivity platform for the employment ecosystem. And today I'm joined by Andrew Brown, the founder and CEO of Check, a leading provider of embedded payroll infrastructure. After selling his previous company, a digital reading subscription service called Oyster, to Google back in 2015, Andrew pivoted to the next big problem he wanted to solve and landed on payroll. Today, Check has powered payroll for more than 35,000 businesses and over one million employees. Andrew, welcome to the show.
Andrew: Yeah, thanks for having me on today, Ansel. Appreciate it.
Ansel: Yeah, so before we dive in, I'd love to let our audience get to know you a little bit more. You have a really interesting background, right? Having founded and successfully sold an app heralded as the Netflix for books, and then a few years later pivoting to payroll infrastructure — what made you realize this was the problem worth solving and dedicating the next decade to?
Andrew: Yeah, I love the surprise in your voice. It's one of the more common questions I get too. Most people are like, why do you want to work on payroll? They're like, I hate doing my taxes. Why would you want to think about that all the time? And to be honest, that's the exact reason. I'm one of those people too. I don't enjoy doing taxes or, frankly, particularly enjoy thinking about payroll. But it's recognizing the fact that we, by virtue of making that easier and simpler for small businesses, are able to have a real impact on the world — that got us excited about it.
The company's mission is to make paying people simple. And for us, that exchange of value when you work for someone and you get paid for it is really the foundation of our capitalist economy. And there's now so many different parties intermediating that. We really said, hey, let's make this just dead simple for people and create infrastructure to make that possible. And that's what we're all about.
Ansel: Yeah, no, it makes a lot of sense. And I like the way you've kind of distilled it into something that's very core to the business of doing work, right? And so the other piece of this is I heard that both of your parents worked in employment law. So I have to imagine you've absorbed a good amount of information from that just as a kid, maybe through osmosis. How much of that early exposure, if any at all, kind of shaped the way you think about the regulatory complexity at the core of what Check does?
Andrew: Yeah, you're exactly right. And I love the term backsolve — I think is very apropos here. I did not expect growing up to end up starting a payroll company. It was an opportunity that we saw much later. But my parents, not only are they both in labor and employment law, they actually met on the moot court team in law school, and very much were the types of, you know, bring it home and talk about it around the dinner table. So a lot of exposure growing up, to say the least.
And I think it gave me just a comfort level in regulated spaces, and in particular in labor, payroll, that sort of thing — where maybe your typical engineer or second-time entrepreneur might have been a little more afraid of it, or not wanting to work with the agencies. As we both know, lawyers only get involved when things have gone horribly wrong.
And so I had seen kind of the worst case scenarios. I saw what a horrible impact that could have on businesses and on employees. And I said, okay, I don't want to deal with those. Let's try and go a few layers up in the stack and actually build software to help businesses avoid all of those downstream problems. My parents warned me against it. They said, do you know all the stuff you're going to end up having to deal with? And I said, maybe, but we'll figure it out. And now, seven years later, here we are.
Ansel: But naturally, things still break. And so given that you all have a front row seat, I'm curious — when you look at this landscape, what are the most common or costly ways that payroll still does break for small businesses today?
[04:35]
Andrew: Yeah, absolutely. It's such a long list. It makes me sad to actually think about. I think the very first one is just data that gets dropped moving between systems. We still see this so frequently, where companies will use one provider for time tracking and a different one for payroll and a third one for benefits — and actually in many cases numerous different providers for benefits — and you end up with just this huge data merging problem.
And the net result is frequently something's wrong on a given payroll. And maybe to make matters even worse, going back and fixing that after the fact is just like pulling teeth: difficult and complex. And so it's cases like that where business owners, I think, feel really stuck.
Businesses are looking to pay kind of as little as they can to have this need just taken off of their plate, while at the same time, when something goes wrong, wanting to have A+, white glove, immediate, expert service. And those two things are in real tension with each other, right? And I think that's honestly a lot of what we're trying to do in the ecosystem is saying, hey, we can actually break that tension. We can build better tools, better systems that enable you to, one, get more things all in one place — you don't have that data merging problem. And on top of that, when things do go wrong, they can be fixed very easily.
Ansel: Yeah, no, I like how you're talking about that data interoperability. You've kind of built this ecosystem where you have a range of different partners that you even work with to help some of that challenging process of translating information, but also again making that seamless. And so I'm curious — how have you approached this partnership ecosystem? Are there things that you've looked for in partners that have helped you feel like, hey, we're actually solving this main problem kind of from the core versus a Band-Aid?
Andrew: Yeah, I think Check… it's always easier to go build the future than it is to predict it, right? The very classic quote. And I think we're a perfect example of that in our little slice of the ecosystem, where for us what we saw was companies building software in kind of the workforce management ecosystem that just screamed that it needed to have payroll included.
Homebase was our very first partner, and I think they're a fantastic example. They have this export payroll button in their application, and it was just so clear that it needed to be a run payroll button. And it's really that simple. It's how you take this one button, turn it from export payroll to run payroll. And that really basic core idea has now spanned all this work and $100+ million invested in all the things that we've done over the past seven years.
And so I guess the point there being: when we go and we look at who we're partnering with, the folks who are building on top of us, it's much less of sort of a business school style top-down exercise — like, let's pick this industry and this other one and this one's broken. No, it's much more like, who are really passionate entrepreneurs and executives that really know their space, that are building amazing companies that have a real need here? There's no guesswork really about, do your customers want payroll? They absolutely do. It's also not that hard to figure out if your platform is well positioned to offer it. If you're doing nothing with the employees, if you literally have no concept of the employee roster, you probably shouldn't be talking to us. But if you have the employees, and maybe you're doing some form of time tracking, or just wherever you sit in the flow you happen to be relatively close to that payroll transaction — well, why aren't you providing it? I can almost guarantee your customers are not happy with their current solution. And that's where Check comes in.
And that's been kind of the beauty, I think, of the whole business. And frankly the most fun part of my role is that every day I get to go talk to interesting new entrepreneurs in new spaces, building really cool new businesses, and help make their dreams possible.
[08:41]
Ansel: You started back in 2019. Are there any verticals that have been maybe surprising for you, that you wouldn't have thought about when you first started, that you're like, this is actually a really exciting one that we're solving and partnering?
Andrew: I think I'll give you maybe two examples of spaces that I never would have predicted that have ended up being really powerful for us. And they're on the complete opposite ends of the spectrum.
The first is the construction space. It's something that I knew very little about when we started the company, and was not a core focus for us initially. But what we've come to understand is that construction is one of the most complex types of work in this country. It's heavily hourly, it's heavily unionized. And there are all sorts of different earnings. People are often working in multiple locations. They can be working across different state lines, different and even local tax lines. And it's just a really tough industry, and one that is often run on still kind of — mainframes is the wrong word — like desktop computers sitting in closets somewhere, and they're trying to move off a 30-year-old product. And we've seen not only one, but probably at this point half a dozen or so different partners who've been built on top of us that are seeing really, really strong success in the space. And I think it just speaks to how deep the pain point is for those platforms today, and therefore how hungry they are for a new solution.
I think the other end of the spectrum, more recently, has been AI-native providers. So these are companies like Warp and Central, which was recently acquired by Mercury. And these are not just companies that in the last six or twelve months said, hey, Anthropic, OpenAI, they seem like they're doing well, let's go do AI. No, they were companies that, again, I think like Check, three years ago, maybe four years ago, said, hey, we have a problem. It's way too hard as an entrepreneur to run my business. I want to figure out how you can make this as seamless as I can — maybe do it from directly within Slack, have the platform just handle everything for me. And they built on top of our APIs and built their own products out that make that just radically easier and kind of more straightforward and conversational and, frankly, like fading into the background magical. And that's another use case where we're seeing really strong traction and really take off. And again, these sort of AI-native ones serving YC businesses and local construction companies could not be more different types of businesses — and two of the most exciting ones being built on top of us right now.
Ansel: Yeah, I love that juxtaposition. But the need is really strong, right? In both.
But I do want to switch gears a little bit to something that's pretty exciting, something that everybody loves talking about, and that's the tax challenge. Because payroll, right, it's fundamentally a tax business — and I think you've even said that before. And so you're dealing with federal rules, fifty states, tons of local jurisdictions. I've read that it's like, I don't know, 24,000 tax codes that change every year or something like that. And so I'm sure it's kind of a nightmare for anyone outside the ecosystem. For those who become experts, I'm curious — how did you approach this patchwork? What makes it so difficult for a small business owner, or even a mid-sized platform, to get this right on their own?
[12:04]
Andrew: Yeah. I think the way to think about the size of the problem is that when you narrow it to sort of payroll taxes in the US specifically, you're talking about on the order of about 10,000 different taxes across all the different federal, state, and local jurisdictions. On top of those taxes, you're talking about on the order of a thousand or so different individual agencies that you have to actually integrate with and work with in some way.
And for each of those agencies, they're going to have some number of independent forms that you're going to be responsible for filling out and paying and processing. And not only is it complex, but it requires very high precision. No one wants sort of a rough estimate of their payroll. The IRS will fine you on a daily basis for every day that your payment is late for one of your federal tax payments. And so there are real monetary sticks behind getting these things right, which I think is so much of why this industry exists.
And it's funny these days, because obviously AI is amazing. I somewhat frequently get the question, can I just say, hey, Claude, build my payroll system, and do it that way? And I think what people miss is, no matter how good AI is, the reality is this information in many, many cases — I think surprisingly, at least to me — is actually not publicly available on the internet. And yes, there are tax codes that are written in law, but then the way those are actually interpreted, the way they actually get expressed by the agencies — we end up spending more time than you would believe both emailing with and on the phone with these local agencies asking them to clarify things.
Frequently they will in practice do things that are the exact opposite of what is written on their form, or the exact opposite of what they actually tell you you're supposed to be doing. And then when you call to try and fix it, sometimes you get through in a minute and you get Susie and she's amazing and really wants to help you out. And other times you're on hold for two hours and then they talk to you for ten minutes, can't help you, and hang up on you, and you start all over again. And you never really know which end of that spectrum you're going to get when you call.
We've believed from the beginning that a platform like Check needed to exist. We look back over the last 75 years and there is more and more software getting built every year, but we saw payroll actually being quite stagnant. And we were just like, that doesn't need to be the case. We can refactor out all this complexity, provide it as a service, provide it as an infrastructure platform to these entrepreneurs, and let them go tackle their own industries, their own businesses. And I think honestly the further on we get, the more I appreciate just how right that thesis was. The space is so brutal when it comes to the long tail of complexity and issues, and that's why just so few people end up trying to do it on their own. And we have not solved the problem yet. There's still a long way for us to go to really make this simple and straightforward for all the business owners out there.
Ansel: I think in the last year or so I read that you all moved over $1 billion in payroll volume, right? Which is astronomical — but also every single cent counts, because that goes to someone that's worked hard for that money. And so getting that right, getting that right at scale, is incredibly important. And most, if any, company is not really built to do that, because it's not a core competency. And so I'm curious — I think when you talk about… yes, please.
[15:42]
Andrew: Can I give you a fun example there actually that I think folks will enjoy? So yeah, we'll move several tens of billions of dollars in payroll this year for north of a million people. So the business is operating at significant scale. Meanwhile, a couple of weeks ago I was dealing with a tax notice — which is a document folks get in the mail if something is wrong with their payroll.
It was a Michigan, I believe, state unemployment insurance tax for 13 cents. And this customer's understandably like, I mean, they're not that worried, it's 13 cents. But they're kind of like, why am I getting this notice in the mail? What did you do wrong? Why is this 13 cents off? And the reason is because Michigan expects you to round the tax amounts to the nearest dollar. But if you just go read their form and all the instructions, it does not say that anywhere.
My understanding is that the law actually does not say that, but the way their system that actually processes it was coded at some point in the past expects that. And if you don't do it, you therefore get a notice. And that's just one example of one tax with one agency with something that ends up going wrong — and you multiply that by all of the ones that we were discussing a few minutes ago. That's the nature of the beast with what we do.
Ansel: Yeah. And I can imagine if you haven't built these types of things yourself and you get that type of question from a customer, you pretty much shrug and go, I don't know, but can we just pay the 13 cents? Right. And I think there's a value in knowing why it works that way. And also then again, building for the infinite edge cases is really where the moat also becomes really powerful, and where your network starts to really compound, right? Because then anyone new that's building on top of you gets to benefit from all that knowledge and insight that you get with the massive scale you've created.
Andrew: Hundred percent.
Ansel: But I do want to talk about — again, we've talked a lot about the abstractions, all the fun stuff that goes behind the scenes — but you do talk about Check's path to payroll, this structured process to help platforms not just build a payroll product, but actually market and sell it and support it. And I think this is something that's really interesting, because it's different than most other businesses, right? So can you walk us through: what does that journey look like from a platform that's going from zero to running live pay runs?
Andrew: Yeah, absolutely. And I think just to frame this, I've used the word infrastructure a lot, and I do that very intentionally, because I think that's the right way to think about Check — not to think of us basically as an embedded provider. And the distinction that I'm drawing there is there are other folks where you can go and get sort of an iframe that is just like their own payroll product that they also distribute directly, and you can put it within your web app. That's totally fine. That might be a good solution for your business. But it's not going to let you do something new and innovative and different. That's what Check is for.
And so I think that what that means for us is when partners come to us, each one of them is ultimately going to have a product that looks and feels and acts quite different from each other. And that means we have to really help them understand and work in partnership and in conjunction with them to take the infrastructure that we know how to provide, as well as all of the industry knowledge and best practices that we've developed by working with approaching 100 platforms at this point, in order to help them actually get to market and be really successful. At the end of the day, it's their business. They have to build it. We can't do it for them, but we can help you along the way.
And so, to boil that down to the details — what does that mean? There are really three parts that we work with partners on as they approach going to market. One is the actual product build. You've got to make sure that your product works, meets the needs of your customers, does what you would expect it to. The second is really the support and service surrounding payroll. When you get that first question from your very first customer about, hey, why is my payroll calculated this way, or how do I onboard to the service, or anything like that — do you know how to handle it? And are you able to confidently be there and be representing yourself well as their payroll company? And then the final piece is go to market. And selling payroll is not the same as selling their existing core product most of the time. And often folks need and want advice and help with, how should we approach this, what are common objections, how should we position in the market, those sorts of things.
But the net of that is you end up with kind of this crawl, walk, run of getting initially live — your very first customer, making sure that you're able to run payroll for them, onboard them, answer their questions — expand that into a beta, and then ultimately a GA. And you're usually, through that whole phase, developing one go-to-market expert within your team who knows it well, who can then begin to scale it across the team and go from there once you have that initial momentum.
[20:42]
Ansel: It's pretty cool because you're co-building something together, right? And you're building something that if and when it's successful, both parties are actually succeeding. And actually the employers themselves, right? Because ultimately they get to have everything in one place in a native experience that isn't cookie cutter, but rather built for their specific business. And I think that's a pretty exciting future.
And so when you look at the future ahead — I think you've described a vision where payroll kind of just fades into the background. Business owner can click a button, run payroll, and have employees paid instantly. No more running payroll days in advance, old ACH rails, things like that. How close are we to that reality? And what still needs to change in the infrastructure to get us there?
Andrew: Yeah. I think it's sort of the classic, the future's already here, it's just not evenly distributed kind of a thing. If you are running on certain Check partners, I think you're pretty close to this. You are probably approving payroll 8 o'clock the night before, on Thursday before you're paying your team Friday morning. So it's not quite instant, but it's pretty close. You may not even be running it — it may be running in the background for you, kind of just getting automatically approved, highlighting any exceptions that may be occurring. A lot of it can just happen for you.
In terms of what really still has to be built though, I think an awful lot of it is what we were talking about earlier. It's like, what happens when something goes wrong? Because things inevitably will go wrong. That's just the nature of the world. And for the most part, payroll products today are quite rigid. They really aren't able to deal with those edge cases very well. And I think that's a huge amount of where we're investing as an infrastructure provider, where I see our partners building on top of us — is taking those cases that frankly historically have been support tickets using some often just kind of mind-numbing back-end tool to go handle things, and instead making those first-class data objects available in the API that become frankly just a single API call to solve. And something that not only is it possible with infrastructure, but as a result of having it being done via API, now MCPs, CLIs, various forms of agents can go call those as well. I think that's going to totally transform the space.
And one of my favorite things I've seen over the last couple of years was a survey of a lot of different industries that showed: how happy are you with your current provider in the industry, and then how happy are you once you switch to a new one? So you know, maybe you hate your current cell phone provider, but you love Verizon when you switch. In that case, you'd be in the top right quadrant. Payroll is the bottom left. You're really unhappy with your current provider and you're really unhappy once you switch to a new one. And it's just like, what are we doing here, guys? This just shows how broken the current ecosystem is. And that's the gap.
It's how do we move it from that bottom left to — if not the top right, because I really don't want people to love their payroll company. I want them to be indifferent and barely remember who they are. I think that's what we're aiming for. So if we can move it kind of halfway up that curve, I think that's the goal. And in order to do that, we're going to have to be able to handle those unhappy path cases really, really well, not just the ones that are easy.
Ansel: Yeah. I think you perfectly summed up the challenge with infrastructure as a whole, right? People are only really hitting you up when things are going wrong, and having a way to build for those inevitable failures, those edge cases, is the thing that's really the valuable point, right? And so given all the things that are changing in this ecosystem, I always have one final question to ask every one of our guests: how do you think all these shifts will change what your job looks like in the next five years?
[24:29]
Andrew: Yeah. To me, the single thing that is most changing all of our jobs — both mine as the CEO of Check, but also ours as a company — is just the incredible capabilities that we now have available via all sorts of foundation models and tools and agents that are being built on top of them. And I think, as I see that, we're in the early days of seeing that shift today. And I think in five years, when I look at it, to me, so much of Check's job will be actually less of helping people through how do they build on top of our APIs, and much more like agents will be the most common user of our APIs and other interfaces that we provide to those agents. And it'll be cases where, again, like our support team — much of that will be handled by these agents who are able to answer the questions.
And so Check then becomes, yes, a core deterministic layer that handles all the complexity we've talked about. But actually above and beyond that, we become a data company in a huge way, where the quality of the evals that we're able to do and the specificity of those, and the way we're able to handle the data for our industry to provide high quality agents and experiences on top, is the core of what we do.
And I think that has huge implications for my job. I think I'll always be doing sales — that's just a part of being in this seat and being in kind of an enterprise, partnerships-driven type of business. But I think it's going to be a lot more time thinking about data and evals. And we'll always have the core deterministic part, but a little bit less so, or at least more built out on top of that.
Ansel: Yeah, well, getting back closer to your engineering roots a little bit if you're talking about evals and data infrastructure. So it all comes in a circle, doesn't it?
Andrew: It does, it really does.
Ansel: Awesome. Well Andrew, I really appreciate you taking the time. This has been a fantastic deep dive into the world of payroll infrastructure, and super excited for the things that people build on top of Check next.
Andrew: Thanks, Ansel. Appreciate it.
