The next frontier: Training computers to read documents like humans

Computers have long struggled to perform this common, everyday task - Is technology finally able to grasp written text?

It has always been one of the most frustrating problems in the world of IT. For humans, reading documents and assimilating the information is an easy task. So, shouldn’t sophisticated computers—which process information far more efficiently—be able to do the same?

Here’s the difficulty: We tend to overlook just how smart humans are. While computers may excel at challenges such as calculating advanced physics equations, humans are equipped to process an entirely different set of complex problems, such as getting out of bed, showering, driving to work, eating lunch and so on. Any of these activities would leave even the most powerful of computers floundering.

Similarly, written communication is an area where computers struggle, while humans take it in their stride. Case in point, chances are you’ve found this article simple to follow so far. In contrast, there’s not a computer in the world that has the ability to do what you are doing right at this moment: Reading and comprehending this material.

Such innovations still reside firmly in the realm of science fiction. However, Pitney Bowes can get you closer. We have made great strides in this sector and are able to deliver an e-invoice processing solution that features a natural language processing engine that can “read” documents we’ve programmed it to understand. This approach is far more advanced than many of the other options available today.

For instance, many competing technologies depend upon information, such as a VAT number, to always be positioned within the same distance of a unique portion of text. In a perfect world, that’s fine. However, if the content on the page shifts slightly, the solution’s ability to locate this information could fail, and often does.

Many clients come to Pitney Bowes after they’ve tried these solutions and experienced frustrating issues. They found themselves spending excessive amounts of time and effort correcting the software’s mistakes, only to be told they have “yet another special case” that requires additional—and costly—configuration. 

That is not how we do things at Pitney Bowes.

Our digital e-invoice processing capabilities offer 100% accuracy, guaranteed. No data input, no time wasted, no costly mistakes. Unlike the competition, you have the confidence of knowing our solutions are offered as a fully supported service, rather than a software license. We configure the solution to the unique requirements of your business, then test it rigorously during installation to ensure everything is working to spec. Furthermore, if any issues should arise, we’re on call to correct the problems, quickly and efficiently.

Digital invoicing offers a variety of valuable benefits, including more control, more transparency and more time to focus on delivering value to your customers. By switching to our solution, you can remove manual data entry almost entirely and maximise your inbound processing, creating savings of up to 80%* within Accounts Payable. All without any significant impact on your supplier. The only thing they will notice is an accurate and timely payment schedule. If they’ve made any errors themselves, these will be flagged earlier and help them get paid. Everybody wins.

100% accuracy guaranteed. It bears repeating. Because when it comes to invoicing, there is no room for errors. Pitney Bowes can stand behind this statistic because we are confident that we have built the right solution and that it will work for your business.

In summary, while computers may never be able to curl up and enjoy a good book, we have crafted solutions that enable them to easily share what they are reading. Which, in turn, can help you write a more profitable future.