Faced with the need to conduct more than 1,000 swab tests a day at the peak of the Covid-19 pandemic in April 2020, Singapore’s National University Health System (NUHS) sought to ease the administrative burden on medical workers who had to manually register patients and create test records.
The healthcare service provider found the solution in robotic process automation (RPA), which automates repetitive tasks by emulating – through a software script – how people perform those tasks. The technology is usually seen as a low-hanging fruit for organizations that are looking to get started on automation quickly with tangible business gains.
NUHS’s RPA project went live in just six days, enabling it to reduce patient registration time from two minutes to just 30 seconds. Over two months, it conducted 72,000 registrations, with time savings of about 282 man-days.
The Asia-Pacific region, and markets in its constituent countries, have been growth engines for RPA in recent years, accounting for about 17% of the global market for RPA services, according to technology research firm Forrester. Particularly in the aftermath of the pandemic, organizations both large and small have shown interest in automation.
Deploying RPA software isn’t the same as building fully automated processes and platforms from the ground up. With basic RPA, a software robot literally does what a human would do. This includes routine tasks such as data retrieval and entry, button clicks, file uploads and downloads, or invoice processing.
Although this is an important limitation, basic RPA is nevertheless advantageous because it can improve the speed and accuracy of task completion while freeing humans to focus on higher-return work, says Malcolm Ong, IBM’s RPA leader in Asia-Pacific.
Full automation, on the other hand, employs systems, processes and even third-party services that are purpose-built for automation from the outset. For this reason, Ong says, the potential benefit of full automation is much higher – but so is the commitment.
But there is a middle ground. When integrated with other automation software to enhance its base capability, RPA can be used in more situations and becomes a valuable component of an automation strategy that includes technologies such as artificial intelligence (AI), data capture, business rules and workflow.
Ong says: “For example, when RPA is integrated with AI, insights can be acted on by sending instructions directly to bots that complete tasks via other systems, such as an automation platform – with no lag time or human intervention – for improved efficiency, customer and employee experiences.”
As with AI, RPA can also integrate into other solutions and software, including digital process automation (DPA) and business process management (BPM), says Matthew Tan, UiPath’s presales director in Southeast Asia.
“RPA is more lightweight and can be deployed more rapidly,” he says. “It can be integrated into DPA and BPM platforms to fully digitize processes that have automated and manual steps. The combination of DPA/BPM and RPA would provide a seamless handover of tasks from human to robot, from robot to human, from robot to robot, and from human to human.”
RPA software tools must include the following core capabilities as a minimum, according to Gartner:
- Low-code capabilities to build automation scripts.
- Integration with enterprise applications.
- Orchestration and administration including configuration, monitoring and security.
What RPA is good for
Dan Ternes, CTO of Blue Prism in Asia-Pacific, says RPA lends itself very well to a large range of actions, such as customer onboarding, order processing and updating employee information. Crucially, RPA helps to free up employee time to work on higher-value tasks, he adds.
But there are a few types of problem that RPA tools cannot be used to solve, including processes requiring very low latency in the milliseconds, or very high throughput.
Ternes says that as judgment, reasoning and emotional intelligence are uniquely human qualities, processes or decisions that rely on those qualities are not suited for RPA.
“That said, as RPA morphs into intelligent automation and as the accuracy of AI improves, it is starting to be feasible for more of these processes or decisions to be handed off to the digital workforce,” he adds.
Ternes singles out virtual customer service as an area ripe for automation. With technology, deregulation and disintermediation making customers churn a daily headache, an organization needs to be engaging and serve its customers, or risk its competitors stepping in.
“Automation can assist here by combining RPA, chatbots, smart forms and natural language processing to offer responsive customer interactions,” he says. “This enables the organization to deliver a delightful customer experience without the expense and effort of adding more people to satisfy escalating demand.
“At the same time, it gives highly trained and qualified staff the time and space to engage at a more meaningful level, usually with the more important clients on more complex issues.”
Organizations can embark on process discovery to identify tasks and processes that are suited for automation. This usually involves using a process recorder that is included in most RPA tools to capture the events and steps to complete a process, such as the number of mouse clicks and copy-paste actions needed to receive and process an invoice.
As part of process discovery, UiPath’s Tan says it is also important to assess the time savings that can be derived through automation.
“You wouldn’t want to be automating a process that saves a person 30 minutes a month but would take 100 hours of effort to automate,” he says. “More importantly, process discovery should also assess the overall benefit of automation. Will it help you comply with service level agreements? Will it help you improve customer satisfaction? Will it help you lower the cost of operations?”
Tan also suggests keeping a scoreboard that will allow an organization to measure success and assess returns as the RPA journey progresses. “So essentially, I would advise organizations to pick a process, automate it and measure success, before working to scale their RPA initiatives,” he says.
IBM’s Ong says for initiatives that incorporate RPA as part of more sophisticated automation efforts involving AI, data capture, business rules or workflow management, additional planning and scrutiny of the overall process will be required.
“It helps to first identify the tasks that are most appropriate for automation and lead to improved ROI [return on investment],” he says. “This will help determine whether it is wise to consider an enhanced solution, or continue to perform the task manually, instead of implementing basic RPA.”
In 2019, Gartner coined the term hyperautomation, which comprises a framework and advanced technologies that help organizations deploy different automation capabilities, including RPA, either separately or in tandem and augmented by AI and machine learning (ML).
Recognizing the need for such capabilities, RPA suppliers have been upping the ante by including pre-built ML models designed for specific use cases and workflows. Examples include ML models designed for loan origination, fraud detection and anti-money laundering in banking, or claims processing in insurance.
According to Gartner’s report on RPA software advancements, ML models created and maintained by RPA suppliers offer significant value. Typically, RPA scripts are limited to simple workflows defined by “if then” instructions. More complex workflows that require processing hundreds of different data points would require complicated scripts with “if then” tree structures of thousands of branches.
By adding ML models that can process large volumes of data points within a workflow, says Gartner, RPA can be applied to increasingly complex and more value-adding business processes. Before this advancement, buyers needed to either buy the ML model, often from a system integrator, or build the ML model, requiring expensive programmers and data scientists.
As platforms or systems change around the automation, those same buyers would need to maintain both the RPA script and the ML model. Buyers can minimize these ML development and maintenance challenges when the RPA suppliers offer pre-built, supported ML models as part of their automation solutions.
Lastly, RPA scripts with integrated ML models allow buyers to automate more complex use cases and improve the scalability of their automation programs.
Increasingly, RPA suppliers are also offering tools to help organizations in process discovery and mining that are used to create business process maps by analyzing data from system logs or user desktop activity.
“The automatic creation of process maps is high value to buyers that lack documented insight on how their business processes operate,” says Gartner. “Without these technologies, organizations often rely on business process owners, resulting in an incomplete or inaccurate process map that causes future rework or reliability issues.”
Key questions for your RPA supplier
According to IBM’s Ong, there are eight key questions that an organization should ask an RPA supplier:
- Are you a “pure play” RPA provider or do you consider RPA to be part of a larger automation strategy?
- How extensive and integrated is your automation platform?
- Can you help me find the best integration opportunities and recommend the optimal course of action?
- Do you have a clear roadmap that can show me how to become more automated in the future?
- Do your offerings meet my requirements for security and compliance?
- Do you have the expertise to help me map, prioritize and document my tasks and processes?
- Does your RPA solution offer tools to develop and test bots, manage deployment, and monitor and handle exceptions?
- Do you have a good track record in business optimization and enterprise computing?
These questions are important because not all RPA software solutions are the same and neither are their suppliers. Depending on the organization’s goals, standalone software may be all that is required to begin.
However, Ong notes that there is value in looking for a supplier that can deliver a broader portfolio of software and services to ease transition to an integrated “RPA plus” capability when needed. Organizations can get the advantages of basic RPA while being able to scale beyond pilot projects to broader adoption and utility, he adds.