Pleo is a company card that claims to automate expense reports

There are a plethora of startups attempting to make filing company expenses suck less, such as Belgium’s Xpenditure or U.S.-based Expensify, which focus on capturing expense data and making it easier to file claims. Now a new player Pleo is throwing its wares into the ring with a new company expenses card — both virtual and an actual physical card — that claims to automate expense reports and enable SMEs to take more control over their employees’ expenditures.

Operating in relative stealth mode until now (aside from winning Best Fintech Startup at the Pioneers Festival in Vienna), Pleo consists of the Pleo MasterCard — a prepaid card that can be charged up and handed out to employees, either physically or virtually — but, powered by Pleo’s backend system and app, the card is somewhat intelligent, promising to categorise spending automatically as well as capture any receipts associated with each transaction.

I’m told that, after completing the on-boarding process, including anti-money laundering checks, it only takes a few clicks to begin issuing cards to employees. Each Pleo user can have different spending rights, such as ‘per transaction’ or weekly/monthly limits, with these permissions controlled and monitored via a central Pleo company dashboard. Transactions are updated in real time. The prepaid element also means no waiting to be reimbursed and less waiting for approval, which is bound to be welcomed by employees.

And like similar solutions, the platform is able to export data to a company’s accounting system, which has the potential to drastically reduce admin and issues related to tracking expenditure and accounting compliance.

“The company credit card has not changed in the past 25 years. It takes several weeks to get one and due to its fixed spending limit and delayed data it involves a lot of trust to offer it to an employee. Most smaller businesses therefore only care to give away very few cards,” says Pleo co-founder and CEO Jeppe Rindom, explaining the problem the London and Copenhagen startup is setting out to solve.

“Every business has developed its own work around. Cards are being shared or employees are asked to pay out of pocket. But all of this creates friction, employees are feeling disempowered and there is a lot of added complexity to the administration. Financial managers report that up to 50 per cent of their monthly transactions are unknown,” he adds.

Pleo’s specific solution came after the realisation by Rindom and the company’s other co-founder Niccolo Perra that a ‘card only’ product wouldn’t cut it and nor would yet another expenses app. “No matter how pretty and smart your expense app is, it just cannot tell you who borrowed your card 12 months ago and now the annual subscription renewed,” he says.

To that end, Pleo is currently targeting businesses with up to 100 employees in the UK and Denmark and is opening up its Beta today. Once the free beta ends, the startup plans to charge a fixed priced monthly subscription, for which users get a virtual and plastic card and the Pleo app.

As for how Pleo’s so-called AI tech actually does all that claimed receipt capturing and categorisation automation, Rindom had this to say:

For online purchases, our algorithm will monitor your inbox for incoming receipts and see if we can recognize it and pair it to the transaction data we already have on file. If you are in a small café, we will push a notification to your phone. From there you scan the receipt with your camera and then you’re done.

Categorisations are more complicated than you think. In many cases, we can guess it by knowing the merchant category, but not always. A $150 transaction from a hotel can be a hotel room categorized as ‘travel’ BUT it could also be a dinner you had with 2 colleagues, which should be labelled ‘Meals’.

In those situations, we use machine learning. A lot of different parameters goes into the equation such as ‘time of the day’, ‘amount’, ‘calendar information’, ‘previous behavior of the user’ etc. and then we come up with the best guess.

We also use machine learning to understand what ‘ordinary spend behavior’ looks like and what may be ‘out of the ordinary spend behavior’ or even fraud. The journey we’re on is finally saying goodbye to ‘expense policies’, ‘expense reporting’ and ‘control’.