By David A. Steiger, Esq., and Stratton Horres, Esq., Wilson Elser Moskowitz Edelman & Dicker LLP

As with many emerging technologies, Artificial Intelligence (AI)driven tools are sometimes seen by small and medium-sized enterprises as either too expensive or uncertain on a risk/benefit analysis basis to seriously entertain early adaptation. Of course, business history is littered with the corpses of the overly risk averse. Still, smaller enterprises often are saddled with limited budgets to invest in cutting-edge technology.

Is there a cost-efficient way to demonstrate the tremendous value of such technology in a claims or risk management organization that will allow for building management consensus to make appropriate investments in the near term? For many enterprises, the answer to that question is yes, and it is known as robot process automation (RPA).

RPA basics

As Nick Ostdick, a content developer for Trekk Design Group, LLC, recounted in 2016, the three key predecessors of today's RPA are:

  • Screen scraping software (SSS) — Ostdick suggests that SSS was the first technology that created a bridge between current and otherwise incompatible legacy systems. However, he points out that the compatibility between any given software and existing systems and applications can vary, and SSS reliance on websites' underlying HTML code can make it difficult for the average business user to understand. By contrast, RPA software allows visual drag-and-drop features. Also, some sophisticated RPA software makes use of optical character recognition technology to adapt to changing websites without human intervention.
  • Workflow automation and management (WAM) tools — Ostdick highlights that WAM can capture certain key data fields, such as customer contact information, invoice totals and items ordered; translate them into a given database; and then notify an appropriate employee. This eliminates the need for manual data entry and offers increased speed, efficiency and accuracy.
  • Artificial Intelligence (AI) — In Ostdick's analysis, AI refers to the capability of computer systems to perform tasks that normally require human intervention and intelligence, such as financial planning and fraud detection. While AI can be expensive, it does offer increased task accuracy and precision and replaces manual labor.

While each of these advancements provided its own advantages, Ostdick argues that the evolution and deployment of RPA and its ability to combine, refine and reimagine certain aspects of each of these prior technologies is what makes it a game-changer.

In an interview for this article, Mike Bruton, Co-Founder and Chief Product Officer for, a vendor focused on using AI to transform risk management, gives the example of using RPA to gather the data needed to support machine inferencing. Because of the potential to connect AI to an RPA-driven project at a later date, Bruton advises designing RPA with potential future add-ons in mind.

The evolution and deployment of RPA and its ability to combine, refine and reimagine certain aspects of ... prior technologies is what makes it a game-changer.

A 2021 IBM Market Development & Insights1 team report states that RPA can perform many time-consuming back-office tasks, such as filling in forms and moving files at extremely high speed and volume, freeing expensive human resources for more strategic or complex activities.

In addition to the significant cost savings and higher employee morale RPA creates, it improves business agility to drive higher customer satisfaction and helps reduce errors to support compliance efforts.

Anthony Abbattista, Tadd Morganti, Matt Soderberg and Jan Hejtmanek (Abbattista et al.) of Deloitte explain in "A Guide to Robotic Process Automation"2 that RPA software "bots" perform routine business processes by emulating the way people interact with applications through a user interface and following simple rules to make decisions. Software bots generally can execute endto-end processes without human interaction unless they encounter exceptions.

RPA and AI uses

Further, Michalis Zinieris of Deloitte's Technology Consulting Practice in Cyprus advises on his firm's website that RPA is best suited for tasks and processes with repeatable, predictable interaction with IT applications. RPA tools can improve the efficiency of these processes and the effectiveness of services without requiring expensive changes to underlying core systems.

When RPA struggles due to its general inability to make decisions outside of its programming, the learning capabilities of AI can be brought to bear.

IBM's Market Development & Insights team reveals that AI brings together capabilities such as intelligent automation, machine learning, natural language processing, reasoning, hypothesis generation and analysis. This collection of resources can be used to approach tasks that require more complex decision making and analysis, such as natural language processing, recommendation services and online customer support.

Examples of RPA bots and their functions

Michael Lim, an integration executive from IBM, explains in a 2020 article3 that RPA can use unattended and attended bots. Unattended bots automate repetitive tasks without human intervention.

So-called low-code (or even no-code) solutions allow users without extensive programming experience to create software and other systems without having to write significant amounts of code.

Attended bots by contrast can be used by human workers to perform repetitive tasks on demand. While recognizing there are appropriate tasks for unattended bots, Mike Bruton recommends that mission-critical tasks, or functions with regulatory implications or that produce output that impacts financial reporting, use strong governance and quality assurance, which may require use of attended bots to monitor performance.

Simply put, to the extent there are technical glitches or output becomes inconsistent with design, human monitoring can catch it quickly before serious harm is done.

Low-code solutions

So-called low-code (or even no-code) solutions allow users without extensive programming experience to create software and other systems without having to write significant amounts of code. Lim suggests that low-code editors can, among other things, enable users to select from hundreds of prebuilt commands to assemble bot scripts, record user interactions to automatically generate bot scripts and test automations using a local bot agent.

Additionally, this approach can allow for a reduced cost of ownership by running multiple bots on the same virtual host.

Benefits of RPA

In a May 2017 piece, PricewaterhouseCoopers observed that since complex integration is not required with their use, RPA programs can be launched in a matter of days or weeks, resulting in low implementation costs and high return on investment.

PwC also noted that savings and efficiencies are enormous, including:

  • Up to 80% cost reduction
  • Rapid results with no spend on customizing existing or new systems
  • ROI in approximately one to two years
  • Reduced errors, saving on the cost of fixing errors
  • No interface development
  • Agile capability to respond to ever-changing business processes
  • Improved process efficiency (approximately 30% reduction in average handling time)
  • High-quality output, less rework.

PwC estimated in an October 2017 article that 45% of work activities can be automated, which would save $2 trillion in global workforce costs.

Abbattista et al. argue, however, that the core benefits of RPA go beyond cost reduction to include greater employee morale — enabling teams to focus on higher-value tasks that truly require human judgement; more time to dedicate to innovation and customer experience; and detailed data capture.

Unlike humans, who may skip a process step or are inconsistent in the way they process a transaction, a software bot performs the task in a standard manner, free of any variation, thus ensuring high levels of accuracy.

Various authorities highlight a wide variety of industries and functions that can use RPA to change the way they do business, from financial services, health care and manufacturing to customer service, IT operations, compliance and claims processing.

RPA is a baby-step toward AI

IBM Market Development & Insights posits that through RPA, organizations are capitalizing on AI to streamline many repetitive, formerly manual back-office tasks. RPA solutions, including those offering low-code or no-code authoring tools, enable line-ofbusiness users to solve emerging business challenges in real time.

They add that although both general automation and RPA solutions are delivering quantifiable business benefits, RPA adopters say they're getting even greater value from their solutions. That said, Mike Bruton asserts that as soon as you introduce any need for judgment into a process, to complete it successfully you will need to use an AI solution.

RPA automation strategy

Abbattista et al. recommend a five-step process to business leaders implementing RPA that looks beyond initial deployment and defines how automation will grow within the organization:

  • Assess for automation opportunities
  • Build your business case
  • Determine the optimal operating model
  • Identify your automation partner(s)
  • Plan the automation road map.

Bruton generally concurs with this approach, but adds that, in his view, priority should be given to identifying an automation vendor to get their expertise to the table as soon as possible to deal with the other elements listed.

RPA + AI = Extreme automation

IBM's Michael Lim defines extreme automation as the application of AI and RPA combined with foundational automation technologies — such as workflow, decisions and content management — to digitize and automate virtually any type of work at scale.

High-level business needs for automation haven't changed, whether it's to speed time to value, do more with less, or create resilient customer and employee experiences. What's changing are the tools to deliver on the promise of extreme automation.

Lim notes that there's a new urgency to move from simple-task automation to scaled or even extreme automations that can handle higher-profile customer-facing and revenue-producing processes.

According to Lim, 50% of companies with RPA initiatives have less than 10 bots in production. He notes that figuring out how to scale beyond those pilots or small projects remains a challenge, but this is where AI-driven RPA tools come in.

Instead of using automation and AI to "duct-tape" legacy operating models together or speed up a limited number of simple tasks in bulky processes, Lim highlights a combination of RPA and AI software that enables users to develop intelligent, straight-through processing.

Bruton comments that before implementing a technology solution, however, it is important to understand what already exists in the marketplace. Existing products may provide value-added solutions without substantial customization or configuration. If a core product that addresses the business need already exists, you can save substantial development time and expense.

A peek at RPA and AI's future

According to Nick Ostdick, industry analysts expect the combination of RPA solutions with more intelligent technologies will have great potential for widespread adoption across all industries.

Machine learning and cognitive computing, for example, are able to deal with unforeseen errors and exceptions in a business process, learning and adapting based on previous actions and experiences. Unlike traditional automation, they are able to apply judgement and creativity to their work, which essentially will allow companies to automate enhanced visibility, transparency, communication and collaboration across their value chain.

Unlike humans, who may skip a process step or are inconsistent in the way they process a transaction, a software bot performs the task in a standard manner, free of any variation, thus ensuring high levels of accuracy.

Ostdick concludes that the days of cognitive automation are on the horizon. Software bots are already able to automate simple, repetitive processes, and through the combination of RPA with these intelligent platforms, they soon will be able to improve their own performance and make complex decisions with little intervention or additional programming. This has the potential to make companies more agile and responsive, which is crucial in today's increasingly global and complex marketplaces.

The hard truth: Invest time and talent

However, despite all the excitement, Lim cautions that there is no silver-bullet RPA tool that will solve the challenge of scaling automations or realizing ROI. Lim advises using integrated tools and resources to create end-to-end automations across business and IT operations. And it takes time and talent to be able to do that.

PwC notes that RPA also can introduce risks if appropriate controls are not in place and monitored. For instance, because RPA action is consistent, any error becomes a systemic and widespread issue across that business process and data set.

Or, if there is a business process change but the robot has not been modified to reflect that change, it may fail to perform or may introduce inaccuracy. Another potential risk arises if someone gains unauthorized access to a bot, then alters it or uses it to conduct unauthorized processing.


With carefully planned implementation, RPA may be a cost-effective way to incorporate advanced data management tools in your operations ahead of risk adverse or slower moving competitors.





Originally published by Thomson Reuters Westlaw Today.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.