The past year felt like a real tipping point when it came to the hype and reality offered by the latest innovations in process automation. At the same time, so many new buzzwords flew our way that it was hard to keep track of them all and to sort out what tools and technologies best fit the kinds of business problems that companies need to solve, or opportunities they wanted to tap into.
This was particularly true when it came to the use of the term “intelligence” when applied to any software-driven tool to streamline and simplify a task or process. Towards the beginning of 2018, for example, robotic process automation (RPA) caught everyone’s attention. But when a distinction need to be made between a tool that could record and repeat a process, and a technology that could learn, adapt and repeat a process, intelligent process automation (IPA) became the focus of attention.
At the same time, artificial intelligence (AI) caught fire in 2018, with lots of practical examples of how AI could be applied to analyzing business processes and adapt to optimize a given process without following a detailed set of pre-defined rules.
The application of the term ‘intelligence’ to any technology has become so broad and casual as to make it essentially meaningless. But, the good news is that despite the hype and the associated confusion, there is some truth and some real benefit the various applications of software intelligence to solving business problems.
And, there is a simple way to cut through the hype and confusion. All it requires is a little context and a simple question. Which is to say that when you talk to a salesperson or vendor about their ‘intelligent’ solution, be sure to ask, “What do you mean by that?”
As for the context, here are some things to look for and consider when you assess their response.
The term automation practically defines our modern, industrial age. It was first used in the 1940’s to describe innovations to production lines in manufacturing. Automation, at a fundamental level, is technology by which a process or procedure is performed without human assistance or intervention.
When it comes to software, we more or less know what this looks like based on the desired result/outcome. The problem is that up until recently, software automation for complex processes often meant lots of work, including custom coding, custom integration and inflexible, rules-driven workflows that were subject to breaking down whenever processes (or systems) changed.
That’s where intelligent automation comes in.
Intelligent Automation, from a software standpoint, is different than traditional automation, because it doesn’t depend solely on following a rigid set of business rules. It can adapt by taking input from a human user (or even from an AI tool), update/modify its rules or knowledgebase based on experience, and adapt to new or changing conditions.
In essence, intelligent automation is still rules-based, like traditional software automation, but in a way that is more flexible and less dependent on developers or IT staff members to change. Changes aren’t coded into a rule set—they are configured/modified by user input or by inputs from other intelligent systems.
That translates into much shorter set-up/implementation timeframes upfront and also into fewer requirements for upgrades or manual modifications downstream.
As discussed above, intelligent automation is rules-based. Artificial intelligence (AI), by contrast, is not. AI systems learn, apply logic, knowledge, reasoning, and problem-solving to tackle a defined problem. Once AI has made a recommendation to solve a problem using its less structured, less rules-driven approach, it can apply what it’s learned to other, similar problems.
So, in that sense, intelligent automation and artificial intelligence are similar. Both can learn and adapt. Intelligent automation starts with a base of rules-based algorithms. AI does not.
And many times, for well-defined processes that follow a common set of rules, such as common back office operations like accounts payable or accounts receivable, intelligent automation isn’t just “good enough”—it is the right tool for the job. AI, by contrast, would not only be overkill, it would simply not suit the use case.
In other words, you don’t need an AI tool to learn and analyze how a company receives an invoice, enters relevant data into an ERP system, matches POs/Invoices/Receipts and routes everything for approval. There are common, standard rule sets for that, which can be adapted and modified by intelligent automation systems.
AI is more appropriate for those processes and problems where the process and ruleset is less well defined, and/or where the solution may lie in a vast stream or source of data that has to be analyzed to identify some sort of guiding structure or principle before a solution can be discovered.
Understanding the contrast between intelligent automation and AI can help sort through the rhetoric to identify the right tools to solve process problems within our organization. Often, embracing intelligent automation to solve well-defined problems can lead organization down a path towards tackling more ambiguous, less well defined process problems by adding AI to their toolset.
In that sense, intelligent process automation can serve as a foundation and a gateway to the the promise of artificial intelligence-driven solutions down the road.