The future is bright for organizations committed to Business Process Management and Automation; the Capture of information from unstructured data sources (like most business documents) is the key that can unlock exponential gains in productivity.
While the latest trends in artificial intelligence, machine learning and robotic process automation (RPA) seem like a combination of marketing hype and science fiction reality, today’s business process automation solutions have a long heritage that dates back to the first business applications in the early 1950s.
The great leap forward that we’re experiencing today doesn’t stem from automating a simple, repetitive process. It doesn’t stem from having the ability to go beyond narrowly-defined use cases to helping to deal with exceptions. The real innovation, when it comes to robotic process automation (RPA), lies in more in the how than the why—specifically, what is required to implement, configure and maintain these solutions to handle routine tasks at a high volume as conditions change.
Historically, the biggest obstacle to automating tasks like accounts payable invoice processes has been that traditional solutions were too inflexible, complex and costly to compete with the adaptibiity and ingenuity of human workers. Unfortunately, all too often the tasks required of human workers were unworthy of the workers themselves—meaning that their skills could be better utilized IF there was an approach to automation that didn’t require tons of effort up front and far too much effort to maintain over time.
The result up until recently has been that employees continue to be over-burdened with repetitive tasks that could be better handled by a machine, if only that machine weren’t so temperamental and fragile when exposed to real life scenarios and dependent upon systems whose integrations were at risk of breaking every time a software update occurred.
Enter the new age of automation. Intelligent Process Automation (IPA) solutions are evolving in a way that promises simpler implementations, less dependence on IT, greater adaptability and a fast return on investment.
One area where IPA has had broad impact and enabled the cost-effective automation of a wide range of back office processes is the capture and extraction of data from semi-structured or unstructured content, including email correspondence and business documents such as vendor invoices and sales orders. The ability to “read” this kind of information, identify the relevant, actionable data and then act upon that information, has been an ongoing challenge that all too frequently has depended upon rigid rules or human intervention—until now.
OCR (Optical Character Recognition) is a well-established technology created to convert text from an image into digital data. In the past, that process didn’t allow for classifying the kind of data that was captured and converted. Over the past few decades, however, data capture platforms leveraged OCR functionality to go beyond converting text images into digital characters, to identifying the KIND of information presented by that text, which could then be parsed to identify data that was actionable or relevant to a process.
Intelligent data capture platforms, for example, are able to identify and extract vendor names, invoice numbers and other details from AP invoice that are either submitted digitally via email, or sent a paper invoices that need to be scanned and digitized.
The challenge has been that the time and effort required to set up those systems, so that they could interpret and extract information from a wide range of document formats was prohibitive. This is largely because the systems followed a rigid set of rules to identify where information might be found on a document, with fairly narrow parameters. Under those conditions, only companies with an extremely large volume of documents, or a fairly uniform set of documents, could justify the investment.
Data capture have continued to evolve, however, solutions themselves have continued to evolve from ‘zone OCR’ applications to more adaptable solutions that rely on multiple parameters and integrate with other data systems to handle any number document formats and types, and also validate the information extracted by cross-referencing ERP or other database records.
Artificial intelligence and machine learning technologies are driving new levels of innovation when it comes to the challenging of getting data out of our business documents and leveraging to automate a process like invoice matching, G/L coding or approval routing.
Today’s systems can identify a document, extract the right information and perform the next step in a process with no custom coding and minimal configuration and quickly off-load the bulk of the work associated with handling an invoice or a sales order. Exceptions to the rules, when they occur, may still depend on a human operator—UNTIL the system has the opportunity to record, process and adapt to the new information and instructions it has received.
As a result, the system ‘learns’ and gets better and better over time, without being depending on a human to program in specific rule sets to be followed.
At Artsyl Technologies, we have focused on solving the major stumbling blocks to achieving the kind of business process automation that have plagued companies for decades. We’re excited to watch this evolution continue to the point where the real innovation taking place occurs among employees who are able to focus more on customer-centric approaches to doing business, than on the tools and technologies they must use to handle basic transactions and routine tasks.
Our hope is that in the future, technology fades into the background, allowing employees at every level within an organization to make the most of their talents and to focus on serving the needs of their customers, partners and vendors.
From where we stand today, that future is just around the corner.