“Companies have been trying to automate literally every angle of a company’s processes but the automation of accounts payable has been left to last,” says Amit.
It is an example of robotics process automation but applied to a specific function.
RPA: the key players, and what’s unique about them
This all begs the question why, why now, why automate accounts payable, at all?
For Amit, resources, or rather lack of them, is a key point. “The accounts payable function within a company will not likely get a lot of resources,” he says, and that in turn means “it doesn’t get so, much attention, it is one of the last areas to enjoy the benefits of automation.”
So what does that mean? What part of accounts payable can be automated? Amit explained, it can be applied to:
* capturing supply information,
* vetting it,
* executing payment,
* communicating with suppliers,
* capturing invoices,
* matching them to purchase orders,
* updating ledgers,
* making decisions about what ledger code should be applied to what invoice line,
* and choosing the right approvers automatically.
But then automation can also help with compliance. Indeed, it is any process driven task, such as following regulatory procedures, that is often the low hanging fruit of any form of process automation, whether it involves a software robot or not.
“Eventually accounts payable should be ubiquitous and it will be just part of the toolbox of every company,” says Amit.
How it works
“When we get the supplier information, we run it through fuzzy logic and other machine learning processes to vet them against black lists and other means to decide whether or not the supplier is legitimate. We execute the payment and if there is any issue with the payment, we apply logic and decision making in advising the supplier how to correct whatever error that happened in the process.
Five ways OCR tech can improve workflow efficiency
OCR and logic
Translating information from hard copy is more tricky. That’s where optical character recognition, or OCR enters the story. “But OCR can only take you so far, he says, “we use a managed service to augment whatever the OCR can do with people, and that will get you to a higher percentage capture of the invoice.”
Amit gives another example:
“When you capture an invoice, you need to decide, for example, whether say a laptop is a marketing expense or an IT expense or an engineering expense. Logic and machine learning logic can be applied to help determine what the right account should be applied to a specific invoice. Machine learning can also help determine who the approver is, and we do that by learning through past behaviour from the customer.”
Chen Amit is the co-founder and CEO at Tipalti, which automates payables operations. Its customers range from the likes of Amazon and Twitter, Nikon and other giants to small companies and all the range in between. Today, they process around $6.5 million in annual run rate
