If we think of the newest trends in IT service automation, or try to follow the recent research, or listen to the tops speakers at conferences and meetups — they all will inevitably point out that automation increasingly relies on Machine Learning and Artificial Intelligence.
Copyright by medium.com
It may sound like the case when these two concepts are used as buzzwords to declare that process automation follows the global trends. It is partially true. In theory, machine learning can enable automated systems to test and monitor themselves, to provision additional resources when necessary to meet timelines, as well as retire those resources when they’re no longer needed, and in this way to enhance IT processes and software delivery.
Artificial Intelligence in turn refers to completely autonomic systems, that can interact with their surroundings at any situation and reach their goals independently.
However, most of organizations are in very early days in terms of actual implementations of such solutions. The idea lying behind the need for AI and related technologies is that many decisions are still the responsibility of the developers in spheres that can be effectively addressed by adequate training of computer systems. For example, it is the developer who decides what needs to be executed, but identifying the best system to execute the processes might be done by software using analytics from within the system.
One of the examples that immediately come to mind is automated testing: test scripts are already widespread, but soon automated testing processes may be more likely to learn on the go, and develop, for example, wider recognition of how new code or code changes will impact production environments.
At the same time, it is always recommended to treat both IT and Robotic Process Automation as a strategy rather than ad-hoc decision. Seen as a continuum, even the initial stage of automation of the simple, repetitive and well-defined tasks will eventually lead to the company’s shift to virtual work.
Such routines are rather easily and cost-efficiently automated with RPA (Robotic Process Automation), where every action is predefined in a manner following “if this then that”-logic. In situations where predictable actions are combined with high volumes, RPA can perform seamlessly without any AI-assistance and may not only save expenses but also increase the quality of operation. At a higher level, the same approach may cover a range of IT Process Automation (ITPA) tasks.
However, when a sequence of actions can’t be well-defined because there exists a variety of possible actions following a situation, such straightforward approach based on predetermined rules however doesn’t work.
Thank you for reading this post, don't forget to subscribe to our AI NAVIGATOR!
In this case, the focus shifts towards smart technologies that can address tasks requiring interpretation. Think, for example, about situations when a simple RPA/ITPA system cannot handle exceptions properly and classify all information that deviates from the predefined rules as errors. In a situation like this, an ML-based virtual worker might provide a better solution, by learning how to classify recurrent exceptions. If delivered as a cloud-service, the worker also provides for a higher flexibility, with more capacity and tools that may be put to use when necessary.
Yet, the utilization of virtual workers should always begin by considering the needs and the choice of technology best suited for the goal. The implementation of smart automation typically starts by recognizing and prioritizing the objectives — the way of thinking inherent to the concept of Intelligent Process Automation.
Intelligent Process Automation, or IPA, refers to a solution, where the technology being used is smart — at least to some degree, such as software robotics, chat bots, image recognition or machine learning. Hence, IPA becomes an umbrella concept for a variety of different technologies that can be utilized together or separately to automate processes.
At its core, IPA combines fundamental process redesign with robotic process automation and machine learning. IPA mimics activities carried out by humans and learns to do them even better. In this case traditional rule-based automation is augmented with decision-making capabilities thanks to advances in deep learning and cognitive technology.
Full Intelligent Process Automation comprises five key technologies:
- Robotic process automation (RPA): a software automation tool that automates routine tasks such as, for example, data extraction. The robot is given a user ID just like a person and can perform rules-based repetitive tasks.
- Smart workflow: a process-management software tool that integrates tasks performed by groups of humans and machines. This helps users initiate and track the status of an end-to-end process in real time; the software takes care of handoffs between different groups, including robots and humans, and provides statistical data on bottlenecks.
- Machine learning/advanced analytics: algorithms that identify patterns in structured data, such as daily performance data, through “supervised” and “unsupervised” learning. In case of supervised learning — algorithms learn from prearranged data sets of inputs and outputs before beginning to make their own predictions based on new inputs. Unsupervised algorithms observe unlabeled data and begin to provide insights on recognized patterns.
- Natural-language generation (NLG): software engines that provide for seamless interactions between humans and technology by following rules to translate observations from data into prose texts. Structured performance data can be processed by such NLG engine to be automatically transformed into internal and external management reports.
- Cognitive agents: technologies that combine ML and NLG to build a completely virtual worker (or “agent”) that would be capable of executing tasks, communicating, learning from data sets, and making decisions based on “emotion detection.”
As it can be seen, IPA promises to enhance efficiency, increase performance, reduce operational risks, and improve response times and customer journey experiences drastically.