Software robots are entering many areas of the enterprise, ranging from marketing, sales, accounting, human resources, onboarding and procurement in the most data-intensive industries including financial services, insurance, energy and transportation and logistics. In fact, by 2021 Forrester Research estimates there will be more than four million robots doing office and administrative work as well as sales and related tasks, with the market reaching $2.9 billion. The capitalization of software robot developers, such as UiPath, Automation Anywhere and Blue Prism, increased tenfold over the past year, while the demand for Workfusion's robotic process automation (RPA) solutions spiraled up 850 percent in the same period. Google has also reported a tenfold growth in search queries about RPA related topics in recent years.
RPA in Action
Software robots may reside on the server or users’ desktop, serving to launch a program, move the mouse around the screen or check cells in a document. Software robots mimic human actions via enterprise RPA. They perform routine and repetitive tasks like logging into the system, reprinting data from one system to another, generating a report or adding meetings and other events to the calendar.
One of the main advantages of RPA is the relative ease of design. It does not require a tight integration of systems, which is especially helpful to enterprise organizations with many systems, some of which are older legacy systems. The solution can be scaled up to support a wide variety of automation use cases across different business groups in the company, because of RPA's flexibility and ability to support automating all kinds of repetitive manual work. And the payback period is fast, according to Ernst & Young (pdf), from six to nine months.
One example of RPA’s benefits is found within Siemens medical unit. Siemens Healthcare (pdf) recently implemented an RPA system that collects genetic data of clients for diagnosis of diseases. The solution automatically set parameters for analysis using more than 90 different settings. The robot extracted the necessary data in 15 mouse clicks, after which the results were entered into Excel to compile various reports.
Another example is ICICI, one of India's largest banks, which has installed 750 software robots to process financial transactions and has doubled the number of transactions to two million per day in two years. RPA is used for interbank operations, currency exchange and retail lending. And in 2017, Ernst & Young introduced 700 robots that search for information in the knowledge base on personnel issues, collect data from resumes, and remind employees about meetings and hotel reservations. The company estimates it will save more than two million hours on routine actions within a year and a half.
Make Your RPA Project a Successful One
Despite companies' interest in RPA, the Ernst & Young report cited earlier states almost 30 percent of projects end in failure. Most of the problems are in scaling the solution, managing and controlling the robots. Here are a few factors to consider for successful implementation of RPA projects.
Keep RPA Projects Focussed
Do not try to use software robots everywhere and all at once. Robots work well when there is a well-defined business process in place. The software operates strictly according to the instructions, therefore, it is not recommended to use RPA in processes with more exceptions than rules. Additionally, it is important to understand what RPA can and cannot do. Know when to continue to use humans to manage a task or to extend RPA to incorporate other solutions that complement it and better address the business problems — it’s akin to shovels being used for garden-beds and excavators are used for building houses, know when to use the right tool.
There are many complementary solutions to RPA, so it’s important to assess the specific business problem you are wanting to fix. If it’s automating and digitizing unstructured content, then seek an intelligent capture solution. If you’re needing to get a handle on incoming customer communications such as emails, customers asking for product status, a payment update or invoice status, you may want to incorporate a solution that can understand text within a communication and make appropriate decisions based on what is being asked. If you’re wanting to automate the inputting of structured data into an ERP system, then that’s an ideal scenario for building a robot. RPA is best for targeted automation.
Related Article: Busting 8 Robotic Process Automation Myths
Don't Feed RPA Unstructured Data
Using RPA exclusively to process unstructured information will not produce the desired results. Standalone software robots cannot extract data from contracts, applications, invoices or pictures of IDs. When combined with software that can learn and classify a wide variation of documents, unstructured content can be turned into structured information that is then fed back into the robotic process.
Here's an example of how this works in a bank: If a robot must open an account for a legal entity, it will first connect optical data recognition to digitize information from the set into eight to 10 documents. Then using knowledge and learning that it has built up about the layout and text contained in the different documents, the software intelligently extracts the organization's name, address, CEO's full name, share capital, shareholder composition, powers, etc. After that, it hands off the structured data to the robot which can then upload the data to the banking system, where scoring is triggered. As a result, an employee receives a completed customer card and quickly, and sometimes automatically, makes a decision to open an account.
This is an ideal example of technology enhancing an employee’s work process. RPA speeds the processes not only in banks, but also in insurance, logistics, healthcare, manufacturing and other spheres.
RPA Solutions Require Oversight and Training
Robots need to be monitored. If your RPA solution is kicking out 25 percent of the data, then you need to know in order to add or change a business rule and check it regularly — which leads to fewer touches by humans and faster processing. A next level of automation is emerging that allows for more intelligent automation. It revolves around training robots to learn from the vast amounts of unstructured content and data that exist within an organization and is critical to processes.
If RPA robots today represent the general assisted robot with basic skills, the next level of robots (aka digital workers) will be highly skilled, capable of identifying documents and extracting data, and even understanding the context of data in a document. This next level of automation is less about robots processing based on fixed rules and more about using the capabilities of machine learning and natural language processing (NLP) in very specific ways to further expand what a digital robot can automate.
Related Article: How to Allay Employee Fears of Robotic Process Automation
We're Only Getting Started With Intelligent Automation
When RPA expands beyond just automating and connecting with applications, it expands into intelligent automation where you will need to be able to not only monitor robots, but also be aware of how processes are performing and how often humans need to address exceptions. This is where process intelligence comes into play. For example, if you are automating a document process, such as onboarding a new customer, it’s a more complex process and other metrics are needed such as the quality of data, how often a human gets involved, the length of the process, and discovering other manual processes that are good candidates for RPA and complementary solutions. The use of automation and the value an enterprise can achieve will be looked at and measured on how quickly organizations can address a wide array of areas across the business.
The robot represents the new digital worker that will be called upon to handle a wide variety of simple and complex tasks today and in the future. We are in the early stages of this intelligent automation journey, but it is clear we have entered a new era of productivity which started with RPA, but clearly is not where it ends.