More than 50 years ago, General Motors installed the first robotic arm in an assembly line in Ewing Township, N.J. Since then, robotic automation has evolved, and how!
Inspired by science fiction author Isaac Asimov’s law of robotics about doing no harm, the inventor of the first robotic arm, George C. Devol, applied his work to one of the more hazardous jobs in an automobile manufacturing plant. Thus began the life of robotic automation in saving humankind from hazardous labour.
Claiming that this is the beginning of RPA is being too literal. Robotics in the digital age, even in its early days, had nothing to do with humanoid robots or arms. However, it has always been about automating tedious, repetitive and high-volume tasks to free the human to do more cognitive functions.
If you asked an accountant, their first experience with automation would have been with macros in spreadsheets. A customer support executive will talk of interactive voice response systems (IVRS) as their first tryst with robotics. A librarian may have been excited by optical character recognition (OCR) and our very own neighbourhood graduate might have found god in screen scraping technology. Search engines use bots – lines of code without a physical form that open websites, read them and tag them. Most people on social media would be able to identify a bot (short for robot) that can post messages across platforms since the early 2000s. The agile development community had also, by then, begun to build robotic automation to perform repetitive test cases.
All of these, in one way or another, have been predecessors of what we call robotic process automation (RPA) today. Leslie Willcocks, Professor of Technology, Work and Globalization at the London School of Economics’ department of management, loosely defines RPA as “a type of software that mimics the activity of a human being in carrying out a task within a process. It can do repetitive stuff more quickly, accurately, and tirelessly than humans, freeing them to do other tasks requiring human strengths such as emotional intelligence, reasoning, judgment, and interaction with the customer”.
The first wave of RPA was in business process management (BPM) — automating simple, repeatable back-office processes such as invoicing, email updates, data entry, data migration between applications etc. For instance, receipt of invoices over email was automated to create corresponding entries in the finance management system. Business users were also delighted because they could automate their task through a visual interface — just click, drag and drop. And all this could be done with almost no change to the underlying technology!
Even as this was incredibly useful as a support mechanism for human work, most RPA systems only performed what’s called structured data interaction (SDT) – they processed information that was in a specific form. This necessitated organisations to already have automation-ready processes running in a workflow engine. When unstructured data was fed into an RPA tool, it failed to perform because it lacked cognitive ability.
This wave of robotic process automation certainly had its wins. It improved productivity and significantly reduced errors. However, it only automated partially, was often deployed on individual computers that made scaling up a change and, most importantly, required upgrades whenever a change was made to the data format it accessed.
These three problems were adequately addressed in the next wave of RPA evolution — automation of more complex workflows led to the end-to-end automation of entire processes instead of singular tasks as before. This paved the way for a virtual workforce complementing employees, instead of a virtual assistant helping one employee. The rise of cloud technologies also helped by allowing RPAs to be implemented on the cloud for scale and high availability. Enterprise-grade automation solutions were able to scale automatically, had contextual awareness, could perform advance workflows, at negligible incremental cost.
This era also saw the use of machine learning technologies in RPA solutions. They advanced to process and derive insights from structured data, even if it is from varied sources and is unpredictable. Mining available data, RPA systems could make simple decisions such as dynamically calculating insurance premiums for customers, based on their past and contextual information. Yet, the unstructured and unpredictable data that is vastly available — user information on social platforms, user behaviour online etc. — proved to be difficult to mine for RPA solutions.
That’s not the case anymore. Artificial intelligence, speech recognition and natural language processing technologies are boosting RPA into the realm of cognitive automation — where a bot can think and act as humans can. Here, the RPA can process unstructured data that it may not have been explicitly programmed (‘trained’) to process. Not just that, it can also discover relationships between data points, identify correlations and perform decision-making intelligently.
Such advanced technology, one would imagine, would take years to build and months to deliver. You’d be surprised! IDC spending guide predicts that worldwide spending on cognitive and artificial intelligence systems will grow to $19.1 billion in 2018. In fact, a properly-built cognitive RPA — such as JiffyRPA — is often pre-trained to perform specific business processes and cope with minor, organization-specific variations. It can hit the ground running in just a few weeks, without even the slightest hand from IT teams and data scientists.
Intelligent RPA is not the future. It is the present and it’s available now.