What is Intelligent Automation?

Posted on:

Transformative technologies have a problem. They’re hard to explain. Try telling a movie mogul from 1983 that within a generation people will not only watch their films on tiny slabs of glass they can carry in their pocket, but will shoot and edit big-screen quality footage on them too. You’d have got blank stares.

Intelligent Automation (IA) has a similar challenge. We’re not just talking about a new application to manage stock control or invoice processing – that’s an old paradigm for software. Nor is it just some kind of add-on based around artificial intelligence tech such as machine learning. And it’s not robotic process automation (we’ll come to why in a moment).

Even an accurate one-line definition of what IA can do doesn’t capture its full importance – any more than “you can watch and shoot video” adequately explains a smartphone.

So rather than aim for a catch-all definition, we assembled a group of internal SMEs to discuss and debate what they consider to be the key attributes of Intelligent Automation – elements that have to be in place to deliver the transformative effects we’ve seen it have on all kinds of organisations and processes.

Spoiler alert: in almost every case, the most important element of the definition explains the “intelligent” part of Intelligent Automation. Once you get your head around that, understanding the full impact of IA is much easier.

“Intelligent Automation is the ability to interpret and execute digitally automated tasks based on undefined inputs.”

This is a cold, raw explanation of what IA does. The most important word in this definition is “undefined” – it’s what makes the automation “intelligent”. A spreadsheet automatically sums a column of figures – but throw a text string in there, and it loses its mind. IA can parse and manipulate data that has not been categorised. It adapts to perform its process.

But there’s one other attribute of IA: it should be able to do so without fundamentally changing the underlying software that runs the process. Most organisations rely on systems that would be incredibly costly or complex to replace, and making significant changes to them is also risky and expensive. IA should deliver its benefits in harmony with these mission-critical systems.

“Intelligent Automation is the bundle of technologies and capabilities that addresses a task holistically and delivers it not just automatically, but as effectively as possible.”

That idea of a technology stack is now pretty well understood, even outside more technical communities. We instinctively know that an app or web service, for example, is made up of several different technologies – and IA is no different.

The reason this is important is that word “holistically.” It’s relatively easy to script a robotic process automation (RPA) solution to take a simple task – transposing figures from emailed documents into an accounts payable system, say. But IA looks at the wider context of the task: how does it work as part of a larger process? How are its outputs used? How might it be improved within that wider context? So, RPA might be part of IA, but IA is so much more than RPA.

“Intelligent Automation’s roots are in process automation, not just robotics.”

The use of artificial intelligence tools such as character recognition and machine learning clearly support the “intelligent” part of IA. But it’s also important to acknowledge that the application of IA is all about taking whole processes to a new level. It seeks to understand underlying business logic and the requirements not just of the process itself, but also how it fits into a broader set of inputs and outputs, so it can adapt around unstructured inputs and execute under variable circumstances.

This is partly what separates Intelligent Automation from discrete AI tools: IA acts and delivers change; AI simply manipulates data, albeit in very sophisticated ways. Think of it this way: we ought to be able to deploy Intelligent Automation across a spectrum of decision-making, triggering a range of actions where the user feels it’s appropriate for them to be automated – even through an interlinked set of processes. A machine learning system on its own cannot.

“Intelligent Automation must preserve the user experiences around the task: IA delivers seamless, dynamic and humanlike interfaces, especially for business users.”

The early days of computing radically altered the way people thought about interactions with organisations. Suddenly, it became important for them to present regimented data that machines could easily understand.

Over time, they’ve got better at parsing fuzzier inputs – so HR can scan CVs, for example, and allocate the information contained in them into standardised forms during a hiring process. IA should apply the same principles to much more complex tasks and processes. It should enable users – without technical expertise – to interact with a process simply.

“Intelligent Automation uses machine learning and data analysis to explain business logic in a way that allows an organisation to improve processes and deliver better insights.”

Importantly, then, IA should allow users to interrogate the process, and especially its outputs, in a way that allows for innovation and new insights. That means whatever tools are applied, users can understand how they are affecting the process, and they can calibrate them to refine processes and outputs, without having to understand the ‘black box’ intelligence delivering the result. You shouldn’t need a data science degree to be an IA user. But you should be able to use it to transform your organisation.

"So, does that mean we’ve nailed it down?"

Not entirely. Because the other thing that’s clear is that even though we have a range of tools, approaches and applications for Intelligent Automation, it’s developing all the time. The use of fast-evolving artificial intelligence technologies, for example, will broaden the scope of any Intelligent Automation implementation. And the emerging ubiquity of object recognition and tracking will see automated decision making in increasingly complex non-digital environments.

But it is becoming very clear that emerging technologies layered upon the solid foundations of those organisations that have embraced IA, will accrue the biggest value at pace.

New call-to-action


4 minute read