The Rise of Intelligent Enterprise Automation

Editor's note: this is a guest post by Tom Henriksson, a general partner at OpenOcean, a VC firm investing in European data-intensive software startups.

In 1999, as Steven Spielberg was preparing to make the movie “Minority Report”, he assembled a team of 15 technology experts to help him depict the world as it would look in 2054, the year that the movie takes place. The result was an impressive and somewhat dystopic future scape where technology permeates our lives.

It is too soon to say whether the vision of the future depicted in the movie will become reality, but 18 years after the film’s release, artificial intelligence (AI) and what is often called intelligent enterprise automation have had a profound impact in some areas.

Marketing spending across every industry and segment of society is now shaped and driven by artificial intelligence using real-time analysis of massive data sets about consumer habits, data sets culled by sophisticated algorithms from billions of transactions and searches happening every day across the Internet. The efficiency of that marketing technology in terms of automated allocation of resources is breathtaking.

And though every consumer may hate being inundated with offers that cater to their interests (even ones that they might otherwise never express out loud), it is hard not to be impressed. That same technology has also been effectively put to use to sway, and sometimes misinform, public policy.

The revolution that has swept through the marketing industry holds the same promise for other industries and sectors of the economy. Customer relations are another driving force. In a virtual roundtable in early May organized by DataSeries, a global community of data leaders and experts, Philipp Heltewig, cofounder of Cognigy, a company that supplies conversational AI technology to companies, said, “Customers expect instant answers, they expect instant service, no matter what time of day and where they are”. Automation of tasks and services can make it possible to meet such high expectations.

A back-of the-envelope calculation in a 2017 article in Forbes estimated that automation of simple tasks can save a department of a large enterprise more than $4 million a year. The benefits that can be obtained by improving efficiency and employee productivity are evident, as indicated by the results of a 2019 study of large enterprises by Teknowlogy Group, a consulting firm.

Robotic Process Automation is the low-hanging fruit of enterprise automation

Though the future of enterprise automation seems bright in general, the transition remains bumpy because automation covers a broad range of processes. Tasks that are repetitive and non-intelligent are relatively easy to automate, while more complex problems, including what is considered the holy grail of automation, a comprehensive evaluation and continuous reevaluation of all enterprise activities using artificial intelligence, is far more difficult.

The lowest-hanging fruit is robotic process automation (RPA). It is similar and can overlap with intelligent business process management (iBPM) and natural language processing (NLP, which uses technology to automate simple spoken or written commands). Though the suite of alphabet soup acronyms sound impressive and complicated, they are often used to handle automatically such time-intensive, error-prone and mind-numbing repetitive tasks as billing, order tracking and stock management. It is not surprising that RPA is the most developed area of automation.

UIPath, a privately held company headquartered in New York, is the undisputed leader in this field and is the fastest growing technology company in North America, according to Deloitte, the giant consulting firm. Founded 15 years ago in Romania, UiPath now has over 6,000 customers worldwide, nearly 3,000 employees, annual revenue of about $400 million, and reached a valuation of $10,2 billion earlier this week.

At the DataSeries roundtable, Guy Kirkwood, who has two decades of experience in automation and the technology sector and is UIPath’s Chief Evangelist, said that the Covid-19 pandemic is also positively impacting UIPath’s business.

While the use of unattended robots, those that do not need human intervention (a field that is led by Blue Prism, a British company founded in 2001), has not been affected – they are doing what they did before -- the virus, by scattering the workforce, has accelerated the adoption of attended robots that require human intervention and decision-making, and which are UIPath’s specialty.

Companies that already had a large RPA infrastructure have managed the transition during the pandemic relatively smoothly. For example, Ericsson, the giant information and communications company, had most of its 85,000 employees start working from home, but it experienced little or no drop-off in productivity. “Those organizations that have already built capabilities in RPA, they are doubling down on it”, Kirkwood said. “They see automation as a way of getting over the fragility of having their people work from home and are now looking at automating pretty much everything they can”.

Automation will have a profound impact on both work and workers

Despite the potential advantages of intelligent automation, its adoption has lagged in enterprises because of organizational and human issues that hinder faster implementation.

Rakesh Sangani, chief executive and cofounder of Proservartner, a technology consulting firm, said, “Introducing new technology is easy. The difficulty is changing people’s behaviors, changing the way they embrace this type of stuff. I think that is the biggest challenge”.

Large-scale automation will also have a profound impact on workers. The change to a more remote workforce, which has become a necessity during the pandemic, will likely be permanent. In addition, the nature of the relationship between workers and employers will change. Kirkwood said the workforce is currently about 70 percent full-time and 30 percent gig workers. He had assumed that those ratios would invert in about 20 years. Now he believes the time frame for that conversion will be a few years instead of decades.

Such fundamental shifts will almost certainly require changes in public policy as the social contract that has underpinned the employer-employee relationship evolves. Universal income, which had been a relatively fringe idea only a few years ago, is now being discussed by some governments as a possible solution to permanent worker dislocation.

For many companies, the Covid-19 pandemic has added a new element of uncertainty into the formula about how soon and how much to implement AI and automation. According to industry experts and insiders, some enterprises, particularly those that have been slow to automate, have predictably suspended or slowed their adoption of new technology. The financial crisis brought on by the epidemic has also put a damper on new investment.

But some companies that have already begun to automate and are therefore more comfortable with the technology are stepping up their plans, using the widespread disruption caused by the pandemic as a justification and an opportunity to reinvent the organization and processes of their companies.

Edward Challis, cofounder of Re:Infer, a company that uses AI to analyze human communications, particularly conversations, said that the decision to accelerate automation by some of his company’s clients had been stunning. “I feel like we’ve had five years of progress in the course of a month”, he said.

Indeed, new market research released in early May by IDC and Everest Group found that from 40 to 50 percent of companies plan to increase their investment in automation because of Covid-19.

Top management must lead the transformation to hyper-automation

RPA and related task-oriented software can improve efficiency, but to go beyond that takes more sophisticated programming and there the barriers to more automation remain formidable, partly because of their relative complexity.

RPA firms are now investing in complementary technologies which include Process mining (automatically tracking how an organization works, how communication flows, how problems are solved, etc., and dividing that information into usable datasets),process orchestration (using analytics and AI to help design better ways of working based on the process mining data), analytics and document understanding which help increase the scope and scale of automation, as well as enable end to end process automation.

These complementary technologies are so new that few companies in the field stand out.

The best known perhaps is Celonis, a German-based company founded in 2011 that specializes in process mining. An emerging leader in this area is Enate, which started offering service orchestration solutions in 2017. Kit Cox, the company’s cofounder, said that enterprises have so far been slow to adapt to process orchestration. “There aren’t really a lot of people doing things beyond task level or meta task level at the moment”, he said.

One company that is taking the plunge is TMF, a professional services firm based in Amsterdam. It is working with Enate to rethink and automate its organization, which includes 7,000 employees spread around the world.

Process orchestration requires a commitment from across an organization as the ultimate goal is to get different parts of a company, software robots and humans, to work more seamlessly together just as they might in an orchestra. The automation process is more than just automation, however, which is why it is so much more difficult. “Introducing automation technology should change a manager’s goals immediately”, said Heltewig of Cognigy, “but it usually lags”.

The top of the automation pyramid is what Gartner, the technology consultant, has dubbed hyper-automation. Essentially, it allows the AI software to take the lead in problem-solving and rethinking an organization’s processes from top to bottom by using and analyzing the data and information gathered from throughout the company and from other automation processes.

While task-oriented automation, such as RPA, is tactical – it addresses a specific problem – hyper-automation is strategic – it assesses how to meet the goals and vision of the entire enterprise. For that reason, it also requires the cooperation and impetus of the heads of companies because it must be driven from the top down, as opposed to the bottom up.

Some chief executives understand this and realize they must be more involved. Kirkwood recounted a recent meeting with the head of one of the world’s largest companies, which is in talks to hire UIPath.

The executive told him, “technology adoption is my job”. The backing of the stakeholders in a company is vital – the employees working with the vendors and the software and the support of management. The corporate culture of an enterprise, as much as the quality of the technology, becomes a wildcard in determining the success of any automation.

Anything that can be intelligently automated, will be

It seems that intelligent automation is at an inflection point. Those in the industry are, for the most part, optimistic about the future, near, medium and long-term. The momentum for automation was already building prior to Covid-19, but the sudden and widespread impact of the pandemic has accelerated the transition, in some cases by laying bare the need for change.

Dominik Dellerman, the cofounder of Vencortex, an AI-based decision-making platform, said that he has been fielding a lot of calls from mid-size companies that had moved cautiously toward implementing large-scale automation in the past. “Companies are saying that we need to invest in automation for the future because the next time something happens, we want to be prepared for that”, he said.

Another way to look at it, Kirkwood said, is “anything that can be automated not only will be but should be”.

Still, just because enterprises may now see the need for wide-scale automation, it does not mean that it can or will happen overnight. Challis of Re:Infer said that to implement the scope and type of process automation that enterprises want will require a big investment also in hardware. “I think we are scratching the surface of automation”, Challis said. “You can’t run machine learning on a desktop”.

Industry insiders are united in one vision of intelligent automation: It will become so seamlessly integrated into how organizations work, and so easy to use, that it will no longer be remarkable. People will stop paying attention to it because it will just be there, working in the background, and providing a myriad of services and solutions.

What will this automation look like ?

One model is the melding of technology and human decision-making depicted in “Minority Report”. A more impressive AI is something like Jarvis, the hyper knowledgeable virtual assistant that Tony Stark relies on in the Iron Man comics.

While those examples are science-fiction, and therefore might seem far fetched, there are already real-world AI programs that reveal a glimpse of what might be possible in the near future. One model is AlphaZero, the program developed by Google to play board games like go, shogi and chess. It used a neural network to teach itself those games in a matter of hours, becoming so adept that it easily beat the human and computer world champions.

The games themselves were mystifying to human experts as AlphaZero used schemes that seemed to violate the basic principles of best play that people had long assumed to be true. The fact that people could not understand how the program “thought” was unimportant, the results spoke for themselves. Some people also found AlphaZero’s play to be “beautiful”, almost like a Caravaggio painting or a Mozart sonata. The same might one day be true for how intelligent automation solves organizational problems.

“It will be like what I describe as genie summoning”, Enate's Cox said. “I don’t need to know how it works”.