Expense reports aren’t the most exciting thing in the universe. But when you reach the scale of say, Amazon, and have tens of thousands of employees, making sure everything comes through smoothly and the right things are being billed to the company could mean the difference in millions of dollars to your bottom line.
And when Anant Kale walks into the office of a big company and asks for a batch of expense reports to do that, it’s when he’s trying to show that company that it’s something that can be done with technology. Kale runs a startup called AppZen, which sells into large businesses and looks to automatically audit expense reports in order to encourage employees to ensure they are only expensing the right things to the company and not going out of bounds — by a little or a lot.
“If you look at the back office, automation has been around for decades,” AppZen CEO Kale said. “SAP and everyone else is making billions selling their software. That software has solved major problems, and it has digitized things — it has moved people from client-server to cloud, where we are focused on is the workflows which are human in nature. A transaction comes to you, you have to look at it, use expertise…
To do that, AppZen said it has raised $13 million in a series A financing round led by Redpoint Ventures with seed investor Resolute Ventures also participating. Redpoint Ventures’ Alex Bard — this is actually his first investment at the firm — will be joining the company’s board. To date, AppZen has raised more than $17.5 million in financing.
Each expense goes through AppZen through the variety of platforms it works with — and in reality, you probably will never see it in action, or even see whether it’s actually working in your system. AppZen looks through historical behavior to see if someone is, say, upgrading a seat on a flight when it’s not entirely justified (say it’s a short flight instead of something intercontinental), and then flags that expense. AppZen then figures out if it’s worth a nudge (like an email), a slap on the wrist (like telling the employee to pay), or worth escalating higher if it’s a continuing issue.
“My biggest question was, how was Appzen (a seed stage business) able to win the trust of some of the world’s largest companies who are also notoriously difficult to win over and could we continue to scale that magic?” Bard said. “What I later found out, is that these wins were due to a combination of Anant’s hustle and, very importantly, the fact that they were able to show real value by doing pilots — the product is so compelling that it has been able to immediately shows customers a significant cost savings they could be realizing and an improvement in employee experience.”
The whole idea here is to create this environment of continuous reporting, where each expense gets a decent glance before it gets logged permanently. And, often times, it’s not the end of the world if something falls out of bounds of a completely rigid model — grabbing an extra pack of gum while at an airport when on a work trip is probably something that falls on the harmless side of the spectrum, for example.
Each employee has a histogram which helps determine whether or not they are expensing things properly. The goal, Kale says, is to try to get employees to become more compliant over time if they are not entirely already. And, often times, it might not even be intentional on the employee end. In the end, AppZen’s goal is to try to proactively educate and help encourage employees to use expenses properly instead of outright judging — though, of course, that only goes so far.
“What we are tracking is your behavior changing,” Kale said. “Is that [score] dropping month over month, so we know you aren’t gonna do this again and again. The employee knows hey I’m gonna be caught, and then the expenses don’t happen. That’s how we save money — not by being the police [in the system].”
AppZen isn’t alone in attempts to throw machine learning in this space — even if the company is able to collect a big moat of data from all these pilots and company reports. There are plenty of smaller startups like Pleo looking to find a smarter way to do expense reports that can keep a close eye on employee behavior and make sure everything runs smoothly. And the larger organizations which have been throwing human eyeballs at this problem may eventually realize they have a wealth of data that they can start throwing resources at to build something similar.
The next step, Kale says, probably moves in the direction of invoicing which uses some similar principles. A tool like AppZen could potentially help spot duplicates, fake vendors, or even deals between employees and vendors. Of course, you start where you can and build from there, but building a set of analytical tools that can fit into an array of new products is what helps turn a lot of startups into multi-billion dollar companies.
“When we hear of a problem, that drives everything for us,” Kale said. ” That’s the right way, how can we solve this. Based on that we figure out, hey, this requires some kind of training, in many cases, we have to look at hundreds of documents. It’s always based on what are we trying to do in terms of specific compliance arguments we can solve for the customer. We think if you can solve it for this customer, you can solve it for other companies.”