Three Stages of IPA Evolution

Three Stages of IPA Evolution

From Robotic Automation to Intelligent Automation and Beyond

From Robotic Automation to Intelligent Automation and Beyond

When it comes to business process automation, it’s amazing how rapidly things seem to have evolved in the last 18 months alone. If you look back to 2018, robotic process automation (RPA) was an emerging new technology approach that was garnering lots of attention, but it’s application wasn’t widespread. Fast forward to today, and not only had RPA been applied across companies of all sizes and industries, but across an amazingly diverse set of use cases. What’s more, the terminology and technology has rapidly evolved as well. Robotic Process Automation (RPA) has given way to Intelligent Process Automation (IPA), allowing companies to go beyond automating repetitive keystrokes and mouse clicks to automating processes that require some decision-making and process logic to automate a process.

What does the next evolution of process automation look like? Let’s take a quick look at where we’ve come from, where we are today and what the future holds for intelligent process automation.

The Evolution of Intelligent Applications

Intelligent applications have evolved from Automators to Augmenters, and are continuing down a path towards becoming Disruptors.

Automators are applications that automate predictably repetitive processes, handling relatively low value but high value work, so that process owners can apply themselves to higher-value work like planning, analyzing and communicating to further improve processes, strengthen relationships and drive the business towards meaningful business goals.

This is the value delivered by RPA solutions. The kind of automation they deliver were previously delivered by solutions as common and simple as macros; or by more complex rules-based workflow automation tools. The thing that set RPA apart is that it doesn’t require a lot of coding and customization to build and deploy RPA solutions, compared to other approaches.

While the level of intelligence is low, the ease of implementation and a certain level of independence from the system being automated (because you’re essentially automating keystrokes, not operations within any one system), is what gives RPA such broad appeal.

From Robotic Automation to Intelligent Automation and Beyond

The distinction from an automation standpoint, is not so much what’s being automated, but how, and at what cost. Essentially, this means that RPA can be applied to lower volume tasks than you’d typically use a rules-based workflow automation platform for. Lower cost and complexity for automation translates into a broader range of applications, where you don’t need high volumes of work to justify the cost and effort.

From an AI or process intelligence perspective, there’s not a lot going on here. Opening a document, automating keystrokes and entering data from that document into another system, has broad applicability, but without a lot of intelligence, it can only handle some very straight-forward tasks. Still. RPA provides a springboard for more intelligent process approaches, and it’s why the world is moving on so quickly from RPA to IPA.

But, imagine combining the simplicity and broad applicability of RPA with other innovative new technology approaches like natural language processing, machine learning and other deep learning techniques. That’s where the next step in the evolution of intelligent processes gets really interesting — especially when you design them to work alongside human operators and allow the system to adapt their algorithms based on the input of a process ownerr.

Augmenters: the next step in the evolution of automation

While RPA is good at automating processes that rely on a simple set of rules, like other rules-based approaches, it struggles to address the exceptions and corner cases inherent to the majority of business processes. Sometimes it’s enough to automate the majority of cases and leave the thornier stuff to human process owners. But if RPA didn’t evolve any further than that, it would be little more than an easier way to automate the kinds of tasks typically relegated to rules-based workflow tools. Low cost and ease of deployment would be its only differentiator. Good enough to lower the bar for what can be automated cost effectively, but not enough to truly empower process owners and transform the nature of their work.

This is where machine learning and natural language processing come in. But rather than relying on some sophisticated, general AI tool to figure out what to do when a process exception occurs, intelligent process automation tools function as ‘augmenters’ for the work that human process owners are good at.

In the words, who needs a general AI platform, when you have a human brain and a process expert to quickly identify and course-set when a bot encounters something it hasn’t encountered before? If the bot can learn and adapt its algorithm with a little input once, then it can handle a similar kind of exception on its own the next time.

In this way, intelligent process automation solutions can do more to automate a broader range of complex tasks in a way that is easy to implement and maintain, delivering more and more value over time as it continues to learn and adapt from human input.

Costs and complexity of implementation remain relatively low; value and applicability grow exponentially. In that context, IPA is able to add value to many of the high volume, higher complexity tasks that have been automated in the past, but with greater flexibility and precision. Accounts payable and sales order automation are two prime examples of applications that have become a gateway for an even broader set of process automation that leverages intelligent process automation.

In those scenarios, an IPA vendor invoice or sales order application can sort through documents to identify invoices or orders, and can extract the right transaction data based on a set of common rules. Where it can’t find the right information, a human operator can pitch in and find the right data within a document by executing a few mouse clicks, one time. The system records and analyzes those mouse clicks, so it knows where to look the next time. That’s just one example, among many within a transactional workflow. Maybe the problem is related to routing and approvals. If the rules don’t allow for the system to proceed for some reason, it alerts a process owner, they analyze the solve for the exception, and the system adapts its rule set on the fly.

In this way, human operators see more and more of their workload automated, without a dependence on IT or development. More work gets done, more tasks get automated and over time, more and more processes get automated to.

IPA solutions augment the capabilities of the process owner. Process owners augment the capabilities of the IPA application. Win-win.

Disruptors: creating new solutions enabled by AI

You can probably imagine where this evolution may be heading next. The ultimate vision for any automation tool has been to go beyond eliminating routine tasks to truly innovating and transforming work. Today, IPA solutions allow for this by freeing up process owners to spend more time on analysis, planning, forecasting and other value-added work.

But what if AI tools could be applied to processes in a way that allows them to analyze massive amounts of data in ways that human process owners simply can’t, offering up deeper analyses and going beyond a single process to look at how to streamline and refine multiple, inter-dependent processes?

Truly intelligent virtual assistants. Truly safe, reliable and predictable self-driving cars. Systems that can talk to us and navigate in the real world beyond what we can imagine today.

This is the ‘sci fi’ part of the equation that hasn’t had as much practical, real world applicability yet, as we can see from how ineffective (and dangerous) self-driving car technology can be when used improperly. Really, from that perspective, we see that today’s self-driving car tech really has not evolved outside of the ‘augmenter’ stage.

But we can see where it’s heading. The only problem with the notion of truly disruptive intelligent processes today is that it can lead to a misunderstanding and mis-valuation of the power of how much is truly possible and practical when it comes to AI and intelligent processes today.

The more companies embrace what is possible now, however, the more we’ll see intelligent process applications demonstrate more power and greater possibilities.

Take action today!

Are you ready to evolve your processes and explore what’s possible? Contact Artsyl Technologies today.


Optimize Your
Data Capture Processes
with Artsyl docAlpha