It has already become a silent but effective force within many organisations, working behind the scenes across all industries to enhance inefficiencies and drive results.
“RPA has proven its efficiency at business processes involving administrative work," explains Sean McCarthy, a Managing Director in Macquarie Capital's Telecommunications, Media and Technology (TMT) business. “It has already transformed the process-driven parts of various sectors such as health, logistics, insurance and finance."
McCarthy points out that because much of this reform has been in streamlining customer-facing services, many people may have experienced the impact of RPA without even realising it.
“When you call to apply for a credit card, you are no longer usually going straight to a person in a call center - you're answering a series of automated responses and the way you answer them will take you down a different path," he notes. “That's RPA."
“Another example is in accounting software where RPA sits on top of the workflow and prompts you to do things like upload receipts, then pulls together all the data you need to reconcile accounts."
In fact, RPA has become so ubiquitous in this field that AI provider UiPath estimates that as much as 100 per cent of all travel and expense report processing can be automated. More broadly, McKinsey estimates the total automation potential of the finance and insurance sectors at 43 per cent.
The limits of automation
And yet, despite both the obvious benefits and high uptake, to date, RPA has also come with its own set of limitations.
"RPA works by applying logic-based rules to known data," says Tej Shah, Managing Director at Macquarie Capital TMT. "In other words, it replicates specific tasks. If it's given data that falls outside of this scope, the bot fails and human intervention is needed."
"This is time-consuming, costly and complex, especially if you're looking to scale RPA across a whole enterprise."
Shah says there is real potential for AI to work together with other technologies such as machine learning and natural language processing (NLP) to overcome this challenge.
"Together these technologies could monitor what humans are doing and immediately learn from that, so that the parameters of what they can deal with begins to grow, almost exponentially."