Intelligent Automation – Steps of progression
Often when people first hear about Robotic Process Automation, or RPA for short, they picture robots doing the job humans currently do. Not only that, in their mind the robot might even do the job the person currently holds. That is, understandably, an unpleasant thought. Nobody wants to be replaced, not by another human being, and certainly not by a robot. Yet it is extremely unlikely a person would be completely replaced by a robot when we’re talking about RPA.
The robots are trained to handle certain situations and react to them in certain, predetermined ways. Any sort of creative thinking, or novel situation, will make them too confused to function. This puts pressure on the design and creation of the robot, but should also ease the fears of the people who fear robots are here to take over their jobs. In fact, people tend to open up to RPA after gaining personal experiencing in a project that utilizes the technology or just by seeing in practice how software robots work.
In 2015 a U.S. based research company Horses for Sources created a model of continuum between the simplest data centre automation and fully automated processes. They aptly named it “The Horses for Sources Intelligent Automation Continuum”. The model visualises the progression from the most rudimentary automation – simple scripts and scheduling – to full automation with true artificial intelligence. Depicted between the two extremes stand various stages of progression.
Image with permission: HfS Research, The Services Research CompanyTM, Intelligent Automation Continuum.
Intelligent Automation Continuum
Far left – Data centre automation
As said, this is the simplest form of automation. The automation is done using scripts, schedules or some other fairly simple means. No matter what exact method is used, the process is based on some sort of trigger – a certain time or an event (such as an email being received) – and the data it handles is very rigidly structured. The process is always done the same way without variation. There are minimum number of branching behaviour and decision points.
To the immediate right to data centre automation is RPA. While RPA solutions mimic humans in the use of keyboard and mouse, they are still very simple, handle structured data, and are triggered by a scheduler or another event.
Autonomic platforms are already much more complex than RPA or data centre automation. An autonomic platform makes decisions on its own, whereas in RPA the decisions are hardwired to the system. These systems check and optimise their status in order to adapt to changing conditions – so they can make better decisions in the future.
For example, say a specific service goes offline. Unless the designer of the solution had prepared for this, an RPA solution would send error messages incessantly unless it was shut down or the service came back online. An autonomic platform would realise something was amiss and maybe keep trying the service to see if it came back online, but it would not send alerts to anyone after the first alert.
Cognitive computing is again one step closer to how a human brain works. These are systems, that, according to IBM, are capable of learning at scale, reason with purpose, and interact naturally with humans. They are able to learn as information changes, and react to changing goals. They can often interact with both humans and other devices and thus gather information efficiently. They are often designed to handle ambiguity by asking questions and prompting for more information. They can also “recall” previous interactions to help resolve future cases. They can also identify context specific information, such as meaning, syntax, goals, etc.
It is telling that while products in the other categories have names similar to any other software products, such as Cicero, Blue Prism, and ignio, products falling under the cognitive computing category are likelier to have humanlike names, such as Accenture and IPSoft’s Amelia and IBM’s Watson.
True artificial intelligence
This means completely autonomic system which can interact with the world in any sort of situation and accomplish its goals. In other words, working like a human mind would. While computers have become more and more capable, it has been easier to say which is NOT in the domain of artificial intelligence, than defining what is within it. For example, image recognition used to be considered an example of artificial intelligence, but nowadays it is seen as routine technology.
In this end of the continuum the process is completely based on rules, instead of rigid triggers. The system can also deal with unstructured data just like a human could.
Note how there are no example solutions here yet.
Business Process Management and information storage
Overarching everything is business process management (BPM) and the way information is stored. Business process management is essentially the optimisation of business processes. This is important, since regardless whether a process is to be automated, it should be executed efficiently. Automation is just one of the tools used in BPM.
It is also significant how and where the data, the process uses, is saved. Is it saved in a database with a clear structure? Is it essentially scanned forms? Is it on the same server the system is running on? Is it in the cloud? These are all questions which affect the automated system.
So as exciting as the thought of handing off all my work to a robot is (we all dream of a break once in a while, right?) it seems it’s not possible to hand off ALL work tasks anytime soon. But just think, is there anything you’d like to pass on? Reporting expenses? Writing reports based on some data you get from a system? For most people there’s something in their workday or week they could do without. Surprisingly often it could be handed off for a robot to do.
Face tomorrow’s challenges with digital workforce by your side! Contact us to unravel the automation potential hiding in your organisation.
Article: Emma Luukka – RPA Solutions Consultant, Digital Workforce