Businesses are constantly searching for ways to drive revenue, cut costs and increase efficiencies. There are many options that companies can leverage to achieve these goals but there is one in particular that is getting a tremendous amount of attention - RPA. Robotic Process Automation (RPA) is the new darling of IT and business leaders alike, but why?
Over the last few
Yet, there are still significant misunderstandings around what RPA is and how it can be used to improve existing business processes. This guide will explain the origins of RPA, where it is today and what the future holds.
RPA is software that automates repetitive business processes. It can interact directly with multiple applications, perform a wide array of operations, and process data much as a human can but with greater efficiency. This advance in technology is leading employees to switch from more manual, tedious tasks to do work that is more productive to the company, allowing them to improve their bottom line.
Robotic process automation software comes in the form of bots, which learn by “watching” people perform their daily tasks. Once trained, the bots are either set up on a server to work around the clock or are placed on employees’ desktops to assist with specific daily tasks as needed.
Bots provide a distinct advantage to humans in that they are free of the many physical and practical limitations that impair human workers. These fundamental differences manifest in several aspects of the workflow and
Reliability – Bots do not need sleep, take breaks or get distracted. They can perform the same tasks at an unflagging pace for as long as needed without suffering any fatigue or having lower quality output.
Speed – Computers can process information and execute tasks at a much faster and more consistent pace than humans, potentially processing hundreds or thousands of pages of data in just minutes. In addition, bots can scale to the size of the project at hand and, given enough CPU power, can deliver a target turnaround time.
Consistency/Accuracy – Bots maintain a level of consistency unmatched by humans. They achieve the same results every time and eliminate human error, allowing businesses to process information more accurately than ever before.
Cost – For the work they do, bots cost a fraction of the price of a regular employee. And once trained, bots can be multiplied or deactivated at little to no extra cost, allowing organizations to scale appropriately.
RPA technology may seem new, but its origins go back to the early 90’s where businesses tried to integrate with older platforms and mainframes without robust APIs (application program interface). They approached these challenges by building and implementing screen scraping in addition to workflow automation tools. RPA combines both concepts and improves upon each of them in several ways.
Screen scraping – Screen scrapers are data entry programs that take data from one application and copy it into another recursively until they have iterated through all the available data – automating so-called “swivel chair work.” Among other things, such programs have commonly been used for collecting data from an outdated, or “legacy,” application and transferring it to a more modern one.
Workflow automation – The concept of workflow automation has existed since the Industrial Revolution. However, this term was first extended to include computer-based tools starting in the 90’s. The earliest digital workflow automation tools combined several tasks into the click of a single button. For instance, macros serve as shortcuts to execute a series of tasks that might otherwise have required a lot more time and effort. Other workflow tools enable automatic switching between multiple applications, even without API functionality.
On a basic level, RPA software combines the functions of both workflow automation and screen scraping tools. However, RPA is more robust, flexible and scalable than any of the preceding technologies without adding additional complexity.
Simple – RPA software can usually be installed and set up without any coding. It also requires minimal IT involvement because it mimics human tasks and, therefore, requires minimal, if any, coding. This makes the setup procedure for an RPA bot largely the same regardless of how you plan to use it, making the whole process relatively
Robust – RPA bots can interact directly with both the user interface (UI) and the API, providing a distinct advantage over older automation systems. Unlike older technology, which typically depended on each element being in the same place every time, RPA bots recognize buttons and fields within an application even in a different location.
Flexible (“Attended” vs “Unattended”) – Because it combines both screen scraping and workflow automation, RPA can be used in either
Scalable – RPA bots can be installed in small batches, by the tens of thousands, or in whatever quantity is needed to get the job done. They can easily be deployed or deactivated, to help organizations impacted by seasonality without adding additional cost or effort. Many RPA vendors are now allowing organizations to move some or all their processing into the cloud so that they’re not even limited by the number of servers they own.
RPA is currently having a profound impact on many industries by improving efficiency, reducing costs, boosting revenues and increasing customer satisfaction rates. To date, the most affected industries and departments have been those with large amounts of simple, repetitive tasks. Below are just a few examples of how robotic process automation is used for business.
Online Retail – RPA bots can automatically create accounts, accept payments, process orders package and even ship items. The whole order fulfillment process, which once took weeks, now takes just hours. There are also automated systems to handle returns, complaints and inventory management, in addition to several other critical, complicated and costly processes, which were previously done manually.
|Case Study - Amazon, by using a combination of RPA and physical, automated robots, increased its inventory capacity by 50 percent, and ship packages with an average human handling time of just one minute. Encouraged by these successes, they are rapidly expanding their distribution centers and continuing to invest in new automation technologies.|
Finance – Several facets of the finance industry have been automated to varying degrees. One of the earliest and most visible forms of automation was the ATM, or “automated teller machine.” More recently, banks, mortgage lenders and accounting firms have all introduced web portals and have begun automating time-consuming tasks, such as redacting personal information, filling spreadsheets, doing calculations and filing various forms.
|Case Study – One financial institution recently switched from processing customer requests completely manually, to using an RPA system and processing times dropped from 20 minutes per request to two minutes. The institution estimated an ROI within two months, and approximately $2 million in annual savings.|
Compliance – Laws and regulations are, by definition, completely rule based and incredibly complicated, making them an ideal candidate for automation. RPA bots have taken a significant amount of compliance work, that would otherwise demand the attention of teams of lawyers and/or accountants to stay updated on regulatory changes, while going through all the relevant information quickly and more accurately.
|Case Study - One large auto manufacturer automated their compliance work, saving over 11 percent annually on compliance-related expenses. Federal regulations across the country are estimated to cost business a total of over $2 trillion each year|
Human Resources – RPA systems are currently used to standardize and automate payroll. Bots can find, access and organize employee information across multiple departments and locations. This eliminates unnecessary data entry and human error while saving valuable man hours through faster processing. Many of the same applications have been used in customer service to a similar effect.
|Case Study - A large outsourcing firm recently used an RPA system to automate their HR department. They increased processing speeds tenfold, improving employee satisfaction. They also drastically reduced their exposure by automatically revoking all access rights immediately after an employee leaves the firm.|
Healthcare – Many healthcare organizations are beginning to implement RPA systems in their back-office. For example, some organizations are using bots for scheduling appointments, processing test results, handling billing and other similar tasks. This allows doctors, nurses and other healthcare professionals to eliminate much of the paperwork from their routine and spend more time with each patient.
|Case Study – There was a study done on 41 Texas hospitals. It found that hospitals with automated clinical information systems showed significantly reduced costs, 16 percent fewer complications, and a 15 percent lower mortality rate, on average.|
Lately, the best and brightest in the RPA field are infusing their automation systems with various AI-based technologies, creating what has been aptly termed “IPA,” or Intelligent Process Automation. This has vastly extended the capabilities of their bots, giving bots similar sensory, processing and communication capabilities to those of humans.
Computer Vision – Computer vision and perception is the science of allowing machines to “see.” More precisely, it allows computers to understand the contents of an image or video clip by identifying shapes and characters by looking at them in context. For instance, a computer vision algorithm might use text recognition technology to identify a letter but then see that it’s in a column of only numbers and, with that in mind, try to reevaluate the mismatched character. Computer vision algorithms analyze every part of the image, from color contrasts to colinear structures. These vision methods can be used to identify and understand objects, tables, graphs
Machine Learning – Machine learning gives computers the ability to “think.” With machine learning, a computer can analyze data and find patterns to learn from. Machine learning, for example, might learn traffic patterns by analyzing traffic history, and then use those patterns to predict future traffic. A clever algorithm might even notice abnormalities in traffic and figure out if there’s an accident, a holiday, or a sporting event causing delays, based on more specific patterns around certain areas.
Natural Language Processing (NLP) – Natural language processing is made up of two main parts: understanding (NLU) and generation (NLG). NLU algorithms translate human language into logical statements that a machine can understand and use. NLG allows machines to write or “speak” in natural-sounding human language by translating in the other direction, from the machine’s computational output to normal words and sentences. These processes enable machines and people to communicate with one another naturally and directly.
Validation Protocols – Validation protocols are algorithms which enable machines to validate their own data output. Using computer vision and machine learning, machines can check how closely their actual results align
IPA, with all its technological advances, can do a whole host of things that regular RPA cannot. Even in processes that have already been automated, IPA technology is often more effective than typical RPA systems. Here are a few reasons why:
Accuracy – Using machine learning and computer vision, IPA bots can recognize common patterns and structures and use them to more accurately process data. These considerations vastly improve the accuracy of OCR, data capture, and other methods of classification and recognition.
Efficiency – Machine learning can also be used to recognize patterns within a machine’s own processes. By analyzing them, a machine can eliminate redundancies while anticipating common errors, iteratively making itself faster and more precise.
Adaptability – With computer vision, bots no longer must be dependent on the consistency of each application. If the interface updates or changes, computer vision algorithms can recognize these changes and recalibrate, then find the buttons, fields and functions that may have been moved or altered.
Compatibility – Using computer vision and natural language processing, IPA bots can process input data that standard RPA bots would not be able to handle because they can understand and respond to written text, speech and complicated images.
Complexity – AI-based RPA systems can handle more complex tasks that require a far more nuanced understanding of what they are doing such as sorting or classifying documents and images even when they aren’t in the same order.
Decision Making – Bots can make real and complex business decisions using predictive analytics to anticipate the outcomes of several actions. Computer vision even enables machines to make these decisions with only partial information, by using context and common patterns to fill in the gaps.
Sequential Decision Making – Validation protocols allow a machine to evaluate its own results and decisions. Depending on the machine’s confidence in those decisions, it can continue building on them to make additional assumptions. An effective validation protocol enables longer and more complex decisions to be made as it gains greater confidence.
Front-End Processes – AI-based advancements like machine learning and computer vision allow bots to be increasingly independent from humans. This fact, coupled with natural language processing, opens the door to a whole new range of tasks that bots will be able to handle such as reading and writing contracts, responding directly to customer inquiries and even having live conversations.
Many automation companies have begun incorporating some form of machine learning or other AI-based technology into their systems. This technology, however, is still in its relatively early stages, and it will take time before these features can be fully realized.
For many industries, RPA is just the first step on the path towards automating business processes. Experts predict that in short order, artificial intelligence will do more to help RPA disrupt businesses than it has on its own.
According to Gartner, the combination of RPA and AI will make the average employee three times more productive than they are currently. This is because there are many high-volume tasks and processes that have yet to be automated but have significant room for improvement.
Automotive – The automotive industry in recent years is one example of how automation is not just a part of the production process, also becoming part of the product itself. Using computer vision and machine learning algorithms, cars (and all types of vehicles) have begun driving themselves. Soon, fully automated cars, trucks, ships and, eventually, entire fleets of vehicles will be commonplace or even standard.
Insurance – The insurance industry requires several types of evaluations. From the outset, valuation and risk must be assessed as part of the underwriting process. When there is a claim, the insured item must again be appraised before the company can process it. Although some of this process is automated already, the more perceptive, less defined parts of the underwriting and appraisal processes still require a person to get involved. With computer vision and machine learning technology, machines will soon be able to independently underwrite insurance policies, evaluate claims and authorize payment. This will make the entire process faster, cheaper and more precise.
Medicine and Pharmacology – Doctors have long used technology as an aid, employing it to capture detailed and specialized images of cells and molecules. Intelligent automation systems, like IBM Watson for example, are already analyzing test results, detecting and predicting various medical conditions, and even advising treatment. AI is being used in radiology, oncology, cardiology and several other specialties to diagnose diseases and conditions earlier and with greater accuracy. These processes are closely monitored by medical professionals but, as the technology continues to improve, will soon take place without involving a doctor at all, and perhaps eventually perform entire examinations or even surgical operations.
Research and Development – Uses for automation will even extend to the discovery of new drugs and the creation of new materials and products. Organizations will use AI-based algorithms to compare and predict the interactions of various chemical combinations on different bodily functions or ailments. These algorithms could replace many of the necessary testing procedures by creating their own combinations and scenarios. As these algorithms become more effective, smart machines will continue to take on more of the research and development processes.
Virtual Assistants – There has already been a significant uptick in personal virtual assistants on phones, computers and even stand-alone devices that have more in common with humans than ever before. A big reason for this is that AI is making personal assistants smarter, more capable and more communicative to close the gap with humans. Using similar NLP and machine learning technology, IPA systems will soon be running drive-throughs, call centers and even media outlets.
Online Retail – Recent advancements in IPA are about to open new doors for retailers. With computer vision and AI technology, machines can extract more information from product images and better understand sizing and style, allowing online retailers to customize their customers’ shopping experience. Combined with machine learning, they will be used to help shoppers quickly and easily find the products that will fit their style and body type, with more descriptive search parameters and a deeper understanding of each item. Eventually, retailers will be able to predict customers' needs to help shorten the buying cycle while increasing customer satisfaction.
It’s impossible to meaningfully predict the potential scope of intelligent or cognitive automation. However, in addition to the specific processes and industries directly affected by automation, people in leadership roles should anticipate several broader, less obvious changes. With priorities, public opinion, the workforce and even the very nature of the marketplace shifting, business leaders will need to embrace automation or risk letting their business fall behind.
Technology in Public Opinion – People are putting increasingly more trust and confidence in technology as they become more comfortable with it. In the public’s view, using RPA, IPA or other forms of automation used to be a cause for concern in many industries, often associated with cutting corners and lower quality. However, automation has now become a benchmark of modernity and forward-thinking.
Customer Priorities – Automation is lowering costs and raising the quality of products and services across the board. This enables and forces companies to devote more resources towards distinguishing themselves from the competition.
Workforce Balance – As more tasks are being automated, organizations are increasingly in need of managerial skills, rather than technical skills. More technical employees will need to learn how to become more strategic project managers to become more employable and ensure that automation is generating an ROI for the business.
Changing Marketplace – Due to automation, services and information are becoming faster, more accurate, more accessible and more transparent. Customers are now able to compare insurance policies, mortgage rates, or other services increasingly easily and instantaneously, putting providers in more direct, almost auction-like competition.
The first wave of the RPA revolution has already yielded powerful results. With the advancements in computer vision technology and AI, the future of RPA is extremely bright. Leading innovators are leveraging these technologies to enhance their current systems so that they are smarter and more automated. As a result, businesses can operate more efficiently and have more control over their bottom line.