Process Automation Fundamentals for Digital Transformation Leaders

Dylan Azulay

Dylan is Senior Analyst of Financial Services at Emerj, conducting research on AI use-cases across banking, insurance, and wealth management.

Process Automation Fundamentals for Digital Transformation Leaders

When it comes to process automation, digital transformation leaders are now navigating the artificial intelligence hype. Although AI can yield some impressive results when it comes to digitizing processes that still involve paper and reducing the time customer service agents spend searching for customer information, leaders are perhaps too excited to jump into AI without knowing the fundamentals of what it entails.

This article intends to prepare digital transformation leaders for taking on process automation projects in the era of AI so that they can stay ahead of the curve as AI starts to become the essential tool for improving efficiencies in white-collar work.

As such, we spoke with Nowell Outlaw, CEO of Vidado, an AI-based document digitization vendor, about what digital transformation leaders need to know in the coming decade about RPA, AI, and their roles.

For more on leveraging AI for process automation, download Vidado’s white paper on the subject.

We begin our discussion with the technology that in large part is dominating the process automation conversation: robotic process automation, or RPA.

Robotic Process Automation Isn’t a Long-Term Solution

Digital transformation leaders are already familiar with RPA. RPA software is in many cases great for automating rote white-collar work, such as manual data entry in a variety of industries, including insurance, banking, and healthcare.

It has applications in customer service, underwriting, and claims processing, among others, and digital transformation leaders may already see it as a useful tool for automating aspects of processes that in many cases they might be outsourcing to reduce costs. Implementing RPA solutions could result in further cost reductions.

That said, RPA isn’t going to fundamentally transform any processes. By definition it replicates the processes that human employees and contractors are already doing within a digital environment. It logs into the same accounts, copies and pastes the same text, and organizes the same documents in the same way an employee would every time, and it doesn’t get any better at it over time.

This may not be enough for digital transformation leaders entering the next decade, and certainly not as artificial intelligence continues to prove its usefulness in the enterprise. Outlaw puts it this way:

In an enterprise company, you’re always driving for strategic improvements across the board where you’re facilitating better underwriting, customer service. With a digital workforce, you’re getting a better expense reduction, but you aren’t absolutely driving new strategic value across the enterprise.

Not only that, but RPA needs to be reprogrammed if the process it’s replicating ever changes, which can hinder further innovation for that process; digital transformation leaders may not want to touch a process once it involves an RPA software because doing so would stop the software from functioning, resulting in downtime for that process. In other words, it might breed more hesitation than leaders at large, slow-moving enterprises already have.

Risk-Averse Culture Holds Back Innovation in the Enterprise

Real transformation requires risk with the hope of great reward, but this contrasts with the culture of innovation at large enterprises, or lack thereof. In our own survey of bankers, for example, they ranked a lack of a culture of innovation as the number one barrier to entry for adopting artificial intelligence at their banks.

This is perhaps understandable; industries like banking, insurance, and financial services have to be careful about the risks they take because any drastic process experimentation can have a significant impact on customers.

For example, a bank might implement an enterprise search application for its wealth managers to better search for customer information across the bank’s CRM, call center logs, and inboxes, and to better search for information about acquisitions, company leadership changes, and other relevant data points across the internet. If it works, this could allow them to make better decisions for their clients’ portfolios.

If it doesn’t, a wealth manager could think they have all the information they need to make a decision for their client, when in fact the software didn’t return crucial information they would have found on their own if they searched for it manually.  As a result, they could cost their client money and open the bank up to a lawsuit.

A Fundamental Framework for AI-Based Process Automation

Despite the risk, however, if digital transformation leaders at enterprises are actually looking to transform business processes in ways that yield 100-, 200-, even 300-percent improvements, according to Outlaw, they will likely need to adopt AI into a process where it fits with long-term goals.

AI is an Investment in Long-Term Goals

Many companies will want to take a slower-paced approach to AI adoption. Instead of rushing in to emulate a competitor’s initiative or use AI for the sake of AI (more on that later), companies should reevaluate how they think about AI altogether.

The truth is that AI is an investment, and a company’s first AI project may yield little ROI in terms of efficiency improvements, cost savings, and revenue. Instead, companies might expect to gain competency with bringing AI projects to life from inception to launch, a valuable skill set they can use well into the future as AI becomes more ubiquitous in their industry.

That said, companies should aim to have their digital transformation and process automation projects relate to their long-term strategies and the goals they want to achieve. The scope should be larger than a small improvement in one aspect of a small process. According to Outlaw:

If you just can automate a very small process and that’s it, there’s little to no larger scale, scope, and size, don’t do it.

Digital transformation leaders may be eager to take on AI projects because they’ve read their competitors’ press releases and heard about AI use-cases in their industry at events. However, efficient AI adoption requires an understanding of the scope of such projects. In high-risk sectors such as financial services, banking, and insurance, AI isn’t “plug and play;” it requires integration with existing software, existing processes, and it can take upwards of a year or more only to find out that AI is only delivering a small ROI if any at all.

In many cases, this is the case even when attempting to make small improvements, and so it’s important for those improvements to amount to a larger goal when scaled out to the entire process. According to Outlaw:

If [a digital transformation leader says] ‘I’m going to fix how we do underwriting at a massive scale,’ great, but [they’re] not going to replace [their] entire underwriting process day one. What [they’re] going to do is try and take the tool and address certain segments of the underwriting process, learn from that, and then try to scale it out. Learn, expand, and then push it out big time.

Example: Insurance Underwriting

For example, a digital transformation leader at an insurance carrier may want to improve efficiencies in and automate the underwriting process. This would be a worthy end goal, but underwriting involves many different tasks. The insurance carrier may want to focus on using AI to digitize paper and scanned applications that come in before they use it to automate and personalize ratemaking per customer and more accurately assess a customer’s risk. This is for two reasons:

  1. Document digitization applications often have a more easily foreseeable ROI. Digital transformation leaders could measure the difference between automatic digitization and manual data entry in terms of time spent relatively easily.
  2. Subject-matter experts could more easily tell whether or not the application is working by comparing the paper or scanned document to its digitized counterpart. Thus, they can determine if the digitized document has any missing or misplaced information. This ability to check the software’s output with relative ease makes it less risky to rely on it.

In contrast, it’s unclear by how much a predictive analytics application for underwriting might increase revenue or reduce the risk the insurance carrier takes on before piloting the software, and even then, calculating the ROI will still be difficult.

Such an application also suffers from the “black box” problem of machine learning more than a document digitization application; underwriters don’t necessarily need to know how an AI-based document digitization software digitized a paper or scanned form, but underwriters absolutely need to know how an AI-based underwriting application came to the decision it did about whether to onboard an insurance applicant or not. The consequences of rejecting an applicant based on a faulty output can be dire, putting the insurance carrier at risk of noncompliance with the law.

Start With Small Projects to Build AI Competency

Despite this, according to Vidado:

Because backoffice service workflows don’t contribute directly to revenue, they lack visibility and prioritization, and most innovation happens only for new business workflows. The back-office simply isn’t a priority.

As such, it may be more difficult for digital transformation leaders to get budget and approval for overhauling the less exciting manual data entry that enterprises are currently relegating to outsourced labor than it might be for applications that have the potential to directly increase revenue, including the aforementioned predictive analytics application.

But Outlaw doesn’t advise companies jump into the largest, most experimental projects first, even if they have the potential to directly generate revenue and reduce risk:

…If its related to AI… in a lot of ways, companies have been burned hard by, instead of doing ‘crawl, walk, run,’ they say they’re just going to run…those massive projects don’t always work. With a new technology, you probably need to apply a ‘crawl, walk, run’ kind of approach so the technology comes in, you spend that initial period of time learning from it, understanding it, and then you can start to scale it out.

The Danger of Falling For AI Hype

Digital transformation leaders that are too eager to use AI for AI’s sake risk wasting their company’s time and money with what we call toy applications. Companies adopt AI toy applications to put out a press release that allows them to say they’re doing AI; in actuality, they’re not taking their AI initiative seriously.

These initiatives don’t align with any long-term business goal or strategy for data dominance, and, as a result, they’re unlikely to make it out of the pilot phase. Leaders that steer their teams into using AI without fully understanding how much of an investment it is to adopt it will find out the hard way how difficult it is, and in the process, they might upend processes that are working for little or no ROI and without any AI competency to show for it in the end.

In addition, the AI hype that results in toy applications can also steer leaders to jumping into AI projects without evaluating the risk and working with AI vendors without evidence of success with past clients. According to Outlaw:

It’s important to evaluate risk vs. reward, and when you’re aiming for the fence, make sure that what you’re being sold, demoed, can back it up with real world evidence for you. The worst case, and what we’re starting to see, is a lot of hype and a lot of false promise, and the truth is, at the end of the day, the improvement that was promised, the technology wasn’t able to deliver it. 

So measuring the yardstick of not only ‘Gosh, this looks really good,’ but also evaluating ‘What risk are we taking on as an organization in doing this?’ and then factoring that with ‘Does the vendor solution have evidence they can do this? Can they back it up with real concrete customer references?’ so that you can de-risk the process and you can make sure you’re going to get that 100 – 200 ‘X’ return on process automation improvements. 

The biggest lesson for digital transformation leaders: start with a small project with little risk to gain competency with AI, but make sure that project aligns with a long-term goal of significantly improving a major process at your company to make it worthwhile. Then, take what you learn from that small project and scale it up to other aspects of the major process.


This article was sponsored by Vidado and was written, edited and published in alignment with our transparent Emerj sponsored content guidelines. Learn more about reaching our AI-focused executive audience on our Emerj advertising page.

Header Image Credit: Enterprise Talk

Stay Ahead of the AI Curve

Discover the critical AI trends and applications that separate winners from losers in the future of business.

Sign up for the 'AI Advantage' newsletter: