This week, we examine business process automation, where recent advances in machine learning technology are enabling enterprises to automate tasks for unstructured data.
On its list of the top 10 strategic technology trends for 2020, Gartner included a number of familiar heavyweights such as blockchain, IoT, and AI-powered cybersecurity. However, top billing went to a new contender, with Gartner hailing hyperautomation as the number one trend—quite a bold statement, especially considering that hyperautomation is a brand-new concept. So what is hyperautomation, exactly? And how does it differ from plain-old business process automation? Let’s see what the experts have to say.
Gartner defines hyperautomation as “an effective combination of complementary sets of tools that can integrate functional and process silos to automate and augment business processes.” In an earlier paper published in July 2019, Gartner described it as “a foundational tenet that allows organizations to cope with the rapid pace and scale of digital business.” Both descriptions offer a fair amount of latitude, so perhaps in order to determine what hyperautomation is we should start by determining what it is not. On this point, Gartner is clear.
To date, the most widespread manifestation of business process automation has been Robotic Process Automation (RPA), a class of rule-based software effective at automating highly structured workflows (e.g. a fixed sequence of clicks that achieves a given purpose within an application). Yet Gartner’s deep dive on hyperautomation is careful to designate it as a movement beyond RPA. The title is unambiguous: “Move Beyond RPA to Deliver Hyperautomation.”
So while RPA may sit at the center of business process automation, hyperautomation is really about what comes after RPA. In short, hyperautomation demands robotic helpers that are much less, well, robotic.
In order to understand what moving “Beyond RPA” entails, let’s start by understanding RPA.
Broadly speaking, RPA tools work by “peeking over the shoulder” of human users, memorizing the steps of a particular workflow within a particular UI and then replicating those steps in a hands-free manner. This approach is effective for automating unambiguous, step-by-step workflows that occur within a consistent environment (like a payroll app) and deal with highly structured data (e.g. numbers and discrete bits of info like employee records).
Conversely, RPA struggles when faced with workflows involving unstructured data (e.g. images, videos, PDFs), inconsistent environments (e.g. document types with varying layouts), and complex multiple-choice decisions (e.g. image classification or image quality assessment). In this case, automation is no longer as simple as laying out a sequence of steps for the software robot to follow. Gartner classifies these more complex workflows as those that go “beyond scripted procedures.” Going off-script—successfully adapting to change and complexity—is a quintessentially human capability, one that RPA tools will probably never have.
Unscripted or not, a process such as image classification still gets plenty tedious once you’ve done it a few thousand times, and a growing number of enterprises now find themselves buried under an avalanche of data too complex for RPA tools to curate. In order to automate the processing of richer unstructured data and reap the tremendous efficiency gains that come from doing so, enterprises must look beyond RPA. In particular, Gartner cites machine learning as a cornerstone of hyperautomation moving forward. As the name suggests, machine learning entails truly learning from humans, rather than simply copying them.
Next week, in order to scope out the present and future of hyperautomation, we’ll take a closer look at this distinction between copying and learning (hint: the latter’s got a lot to do with feedback).