Rube Goldberg Machine
It's not a matter of if robotic process automation will upend how we work today, but when PHOTO: Jeff Kubina

A recent Forrester Research report found the robotic process automation (RPA) market — a space sometimes called the robotic digital workplace, among other names — would grow to $2.9 billion by 2021 from a base of $250 million in 2016. 

This is a nice growth spurt by any measure — but it comes with a catch. RPA's current path is not sustainable in its current form, or at least there is not enough demand for what these companies are currently offering to propel the industry to any meaningful growth.

The missing element is AI, specifically cognitive AI enhancement.

Three Vendors Leading the Way

The good news is there are distinct signs the industry is taking baby steps in the right direction. Products are starting to come to market as a result of three partnerships formed or acquisitions made by business process management (BPM) providers of RPA companies over the last year.

Last month Appian debuted Appian RPA with Blue Prism, a low-code application development, BPM and robotic workforce software package bundled together in one offering by a single vendor. "Appian and Blue Prism are leaders in their respective markets, and this alliance brings our customers the best of both worlds," said Appian CEO Matt Calkins. “We've aligned the products and companies well, and implemented successfully together at major client locations."

Prior to that, also in July, IBM and Automation Anywhere announced a partnership that makes available a set of tools that automate and simplify data-intensive tasks within processes managed by IBM’s process management software. 

"Through our collaboration, we aim to help companies free their employees to focus on more meaningful aspects of their jobs," IBM stated in a blog post announcing the arrangement.

The first of these deals came in 2016, when Pegasystems acquired OpenSpan for $52 million. With its year’s head start, Pega has a handful of robust offerings for customers.

Baby Steps Towards a Cognitive RPA 

But these developments are only the precursor to the AI-infused RPA market Forrester envisions. 

"Today RPA is only doing simple things, such as bots typing on a keyboard," Craig Le Clair, principal analyst at Forrester Research told CMSWire. "It is not making decisions and that is where the real value will be." These recent announcements, he said, are important because they show the industry is moving forward. "But the real dimension for RPA is the cognitive one."

"In the next couple of years we will see more analytics supplied to documents and unstructured content — that will be the first phase of RPA’s growing intelligence," he predicted. "Then we will move to smart decision-making, where bots can handle advanced exceptions."

Today anything you want a bot to do must be explicitly programmed into the bot, Le Clair continued. But the future? Imagine a knowledge base that incorporates best processes that an AI-enhanced bot can access. "This bot can then understand what the customer is trying to do, even if it is a complex transaction, and make decisions based on the context and what is in the knowledge base," Le Clair said.

When put that way, one can see how far today’s processes must go when compared to where we are today. The most common current example is in personal productivity and team collaboration suites, such as Gmail and Outlook, both of which now filter inboxes automatically, Alan Lepofsky, Constellation Research vice president and principal analyst, collaboration, told CMSWire.

"Gmail takes it a step further, offering three suggested replies to messages. Microsoft Office and GSuite are both now leveraging AI to help automate common tasks, ranging from designing slides, providing sentiment analysis on spreadsheet values or translating text and audio on the fly,” he said.

$2.9 Billion Worth of Pain Ahead

All of those tasks Lepofsky outlined were once done by humans: humans paid to do a job. In that context, it's easy to see how Forrester is predicting a $2.9 billion market from the current base of $250 million. As it writes in its report: 

"Adding AI to RPA will free it from an exclusive focus on rote tasks. AI will account for an increasing portion of the digital workforce, and in the end, RPA will be a small fraction of the overall AI 'cubicle' market spend."

But $2.9 billion is going to hurt. Forrester estimates the RPA market targets two cubicle job categories: office and administrative; and sales and related. "These cubicle jobs offer a $2.9 billion RPA market opportunity by 2021," the report said, "but only if AI can lift RPA beyond automating rote tasks."

Forrester is not the only one that foresees significant disruption by RPA. The next few generations of RPA will have a huge impact on white collar workers and the way we do business: it will determine who has a job and who doesn’t, Alan Pelz-Sharpe, founder of Deep Analysis told CMSWire:

"The impact will be obvious and affect a range of industries including legal, accounting, financial services, healthcare — anything that is heavy on repetitive data tasks. I would say that about 65 percent of the work people do in those sectors are repetitive tasks, such as reconciling accounts, looking up and checking facts and so on.

"That is the work that the robotic workforce is targeting."

But losing these jobs to RPA will free people up for more creative tasks, right? 

Wrong, Pelz-Sharpe said. "It is a little naive when people say it will free people up to do more interesting work — the reality is it will make a lot of people redundant."

RPA: Not a Matter of If, But When

Cubicle workers can relax, at least for the short-term: the end of their jobs is not imminent.

But while the timeline for an AI-enhanced RPA system is a matter of debate, make no mistake — it is within our lifetimes. Pelz-Sharpe said some of these changes are already underway — "it's happening at some points in the process" — but that the bulk will occur within five to 10 years.

Critics may disagree, he continued, with the argument that these developments are just part of the BPM’s natural progression, and thus the evolution will happen more slowly, but Pelz-Sharpe said RPA is something different. And new.

"The challenge for BPM has been that it has struggled enormously with the integration of the different silos of data and applications," he said.

"It is hard to get those things to talk to each other. But that is the beauty of RPA: it is really easy to integrate data and doesn’t require loads of working building APIs."

"The older systems promised to do this but in reality they were so dependent on consulting to make it work that most people didn't do it. With RPA you can have something up very quickly."

That ease of integration, Pelz-Sharpe said, has been the tipping point.

The Tech Might Be Ready, But Is Your Business?

Le Clair, for his part, is less sure the digital robotic workforce is almost at hand. Some 75 percent of vendors are still in the early stage of a proof of concept or a small pilot, he said, and it will be only over the next year or so that they will begin to look into the operational support needed for a digital workforce.

A bigger challenge than vendor readiness, however, is the change management that will be necessary, according to Pelz-Sharpe. "You can't rip out 40 years of doing something one way and replace it overnight,” he said. "You can’t just shut down an entire legal or accounting firm overnight."

The change management associated with these projects will stretch implementations out for years. "Most companies are not ready at any level — cultural, process or financial," he said.

For further reading: How to Survive When the Robots Come For Your Job