To “Bot” or Not to “Bot”
Origins and Evolutions
Why does Information Technology change our lives so much? Why and how does it really impact our lives? Is it because it is now so pervasive that we have become dependent and cannot perform everyday life and work tasks without it? We are currently entering another technology phase that will further disrupt and transform our relationship with technology.
In the past organisations have worked hard transformed IT systems to meet defined user and process requirements, optimised flow and reduced paper via electronic workflow systems and reduced costs via outsourcing yet still there were levels of efficiency, accuracy, value and scale which were perceived as unobtainable or elusive. Whether outsourced or onshore, the constraining factors were always assumed to be faults and inefficiencies in the process or more ominously, the limitations of human nature and behaviour in the workforce.
The pace of technology change in this area is unprecedented and these previous constraints and assumptions are being fundamentally challenged. Largely this is driven by increasingly levels of process automation and autonomous software component cognition and behaviour. In this article we will consider the key concepts and types of software process automation
Tasks we previously perceived to always have required a human to perform and control such as those in the customer services sector, have become highly automated and tasks are often performed by software "bots" as they are known. This is happening in a subtle and diverse manner but largely without widespread public awareness or detailed understanding. Have a think about the questions below:
- The ecommerce site with the friendly interactive sales agent that helps you decide what to buy. It’s always responsive and seems to know lots about and helps steer you to “best” product.
- Creation of account and contact details when opening new mobile phone, utility or bank accounts. Who actually pressed the buttons to create them? Does it matter?
- Custom composition of email newsletters to ensure that each recipient only gets what they are likely to be interested in. How does the system "know" which were personally relevant?
In most of the above scenarios, it is highly likely that the service was delivered and controlled by automated software processing or a software accelerated and optimised version of existing systems. These synthetic robotic processors or "bots" as they are popularly known, "learnt" or were "trained", to perform these tasks and interact in a way that best met our wishes. Increasingly in process automation scenarios, humans are only “in the loop” to monitor the processes and only step in to handle out of the ordinary customer scenarios or processing exceptions and complaints.
Previously these tasks would have been handled by an outsourced human workforce in low-cost service delivery and customer centres. Using Robotic Processing Automation (RPA), the additional costs efficiencies come from the elimination of human labour forces, rather than its arbitrage. RPA has the potential to revolutionise some aspects of the outsourcing industry. We will cover RPA and its implications for outsourcing in future articles.
The Importance of Software Process Automation
These changes are already happening and the pace of change is accelerating. The changes and the implications are very real for all aspects of society and the potential changes to the nature of work as we understand it are profound. According to the McKinsey Global Institute "Cognitive Automation is going to perform the work of between 110 - 140 million workers globally over the next 10 years".
The media channels are overflowing with apocalyptic stories about the "rise of the bots" and how they are invading the world of work to take over most of our jobs. The reality is somewhat different. Robotic Processing Automation and its more evolved and cerebral cousins Intelligent Automation and Smart Processing Automation, present both opportunities and challenges for us all. Unless we understand what they do and how they work we will not be able to tell the difference or manage them!
Understanding and Achieving Automation
There is no one single "magic bullet" approach or technology which delivers this level of automation. More commonly RPA utilises one or more mechanisms with levels of automation applied at different levels. There are three different general types of processing automation.
- Robotic Processing Automation (RPA) Tools
- Intelligent Automation (IA)
- Smart Processing Automation (SPA)
These different types of automation are generally differentiated by the level of software and machine learning and awareness involved to autonomously adapt task execution.
In the case of Robotic Processing Automation (RPA), limited cognitive autonomy is involved and the systems generally involved the repetition of standardised, structured and procedural tasks. These are optimised for use with using existing systems in an unaltered form, but which are able to execute tasks at very high levels of volume and frequency to a higher level of accuracy. Generally little or no IT transformation of existing systems is involved.
Intelligent Automation (IA) involves the use of extensive machine learning and artificial which has been applied to the domain area in order to produce data backed knowledge stores or domain-specific semantic understanding. This “understanding” is then used to interpret the incoming data source and decide how to continue the execution path. This type of automation is optimised for processing unstructured input sources such as newsfeeds or voice recognition and then transforming these inputs into specialised business or customer outputs which add value. Some IT transformation of existing systems to access domain based rulesets and decision making systems may be involved and often interaction will be at the software API level rather than via a user interface.
Smart Processing Automation (SPA) systems are usually a combination and evolution of RPA and IA in which the artificial intelligence, machine learning and cognitive recognition mechanisms and data stores are harnessed together in such a way that the system automatically “learns” how to handle exceptions, improve the process and optimise task execution. A continuous cycle of feedback, learning and improvement is utilised which generally requires significant processing and storage capability. Sometimes this can be used to further improve structured process execution of existing systems but more often it is used to radically transform task execution through the use of APIs and entirely system to system base communication.
Example Processes Suitable for Software Process Automation
Some typical example areas of uses for software based automation from the financial services sector include the following processes:
- New account creation and verification
- Data validation
- Customer account management
- Financial claims processing
- Report creation
- Form filling
- Record details update e.g. change of address
- Loan application processing
- Standardised Settlement Processing
These processes are highly structured with defined outcomes and rules structures to guide behaviour. They can also be run in parallel without interdependency and at significant scale. Critically in many cases, there is also no perceived loss of service to the customer and the improved levels of accuracy and responsiveness can be seen as a potential market differentiator.
In the next article we will look at Robotic Processing Automation concepts and approaches. In subsequent articles we will look at Intelligent Automation and Smart Processing Automation. In each case we assume no prior knowledge of the topic area.