Robotic Process Automation (RPA) may sound like science fiction: Robots take over the annoying steps in work and automate mundane processes. RPA, however, is anything but science fiction and has been around for quite a while in many companies. It has been in place since the early 2000s, especially in the marketing sector, and then found strong acceptance in the automation of software testing in quality assurance. This article describes the path from rigid, rule-based systems to intelligent process automation: From Robotic Process Automation to Cognitive Automation and Digital Assistants to Autonomous Agents.
RPA: Delegating work to the robotic colleague
It is key to understand that robotic process automation is not about “robots” as in physical objects or humanoid machines. Instead, a “robot” is an intelligent software program, that takes over tasks. It is a piece of software that takes care of a staked decision scenario or a process and executes it automatically. A robot in RPA is therefore neither a humanoid robot nor an Artificial Intelligence.
Using “traditional” Robotic Process Automation in practice
RPA is really strong when it comes to repetitive and rule-based processes, where highly structured data is available in digital form. Hence, typical use-cases that can benefit greatly from RPA include, example, checking of requirements when opening a bank account, the verification of incoming payments and open invoices or – as used at Telefónica / O2 since 2010 – a software robot-based automation of certain customer support processes. However, the scopes of these robots are rather limited, which is illustrated by an example from Deloitte for the successful implementation of RPA in purchasing:
In a current customer project at a leading automobile manufacturer, the process of core-data processing is automated by suppliers. […] For example, when considering the validity check of the VAT number, the extent becomes clear. This test is typically carried at each creation of a new supplier. On average, an employee requires approximately four minutes for creating a new core-data set. If this process is automated by a robot, the time required is reduced to only about one minute per process. Manual intervention of the employees is only necessary if the VAT number is not valid. In such cases, the robot automatically notifies the contact person.
Source: Deloitte Robotics Case Study
In this scenario, the robot – a piece of software – is responsible for validating the VAT number. And if the number is not valid, a human has to take over again, because from there on, the process is no longer rule-based: Was the input formatted incorrectly? Was there a “number spinner”? Or why is the VAT number wrong?
It adds up: Automating small, frequent tasks
The automation of routine tasks – such as checking the VAT number – initially sounds like a rather simple application. Nevertheless, the effect should not be underestimated, because, with the corresponding frequency of this task, a huge amount of time can be saved in total.
If an employee does this job 5 times a day and now saves 5 minutes per step, he has effectively saved 5,000 minutes, or 83 hours, on 200 working days per year – that’s a whole 2 working weeks, which can be used towards strategic and value-adding activities! This relationship between “frequency” and “time savings” is, especially in the IT, illustrated with the following graphic:
Intelligent Automation for complex tasks: Cognitive Automation
To be frank, a process that simply validates an ID is not automated intelligently. In our daily work, we usually have to deal with much more complex challenges. We have multiple systems in use, decisions sometimes require tacit knowledge, dealing with partially unstructured data and ambiguity.
A rule-based robot is overwhelmed in the real world. However, with the advances in technology, the fields of application are starting to move into untapped territories waters – away from the safe haven of rule-based “if X, then Y” decisions – towards the application of Machine Learning and other AI approaches: If data is partly unstructured and standardized rules can’t be applied, Cognitive Automation comes into play. Cognitive Automation refers to RPA combined with data science approaches.
Applying Cognitive Automation in practice
Cognitive Automation is strong when it comes to textual content. Technologies such as “Natural Language Processing” make it possible, for example, to automatically classify incoming emails and then process, or route, them accordingly: Is this email a standard question about opening hours? Then it is answered using a standard template; if the email is more likely to be a complaint, it will be addressed to a responsible person who can personally handle the request.
The combination of Machine Learning and case-based delegation to humans cannot always fully automate processes. In return, however, it becomes possible to automatically process parts of the tasks and processes, so that people can take care of the actual value-adding topics. The downside is, that both RPA and Cognitive Automation are a sort of “black box” for the employee and end users, as they can not really interact with – what happens between input and output is not transparent to users. This is where the hitherto highest level of Robotic Automation becomes relevant: Digital Assistants.
Digital Assistants: When robots learn to speak and to chat
Taking “Enterprise Automation Journey” a step further (remember: We started with Robotic Process Automation and Cognitive Automation), we arrive at digital assistants. In contrast to the previous levels, a digital assistant now has a speech and text-based user interface, a so-called “Conversational Interface”. Similar to humans, you can use voice and text to interact naturally using “human” language.
The key difference: The chat is not answered by a human, but by an intelligent software robot. This brings along all the benefits of technology: 24/7 availability, virtually infinitely scalable at rather low costs, and cross-platform- as well as cross-device-access.
Natural Language Understanding opens up new Automation opportunities
So, digital assistants are able to extract information and complex data from conversations, to understand them and process them accordingly. This enables completely new application scenarios, where traditional RPA and Cognitive Automation cannot be of help. For example, entire processes – such as the purchasing process – can be supported by digital assistants to collect required data per process-step from the employees by asking him questions when the information is relevant. Accompanying processes is thus possible with digital assistance systems.
The “trump card” of intelligent automation: Adaptation to the existing system- and IT-landscape
The automation concepts presented in this article have one common denominator: You don’t have to adapt the existing enterprise tool-landscape. RPA and Cognitive Automation are, so to speak, the link between systems and individual process steps – and digital assistants form the bridge between humans and systems. Especially with systems such as our digital AI-assistant Neo, it is possible to connect third-party systems in 18 programming languages. This enables the application of automation concepts even in heterogeneous system landscapes. Existing systems are not replaced by any of these concepts, but ideally used and maintained “on-the-fly” automated.
One tool to rule them all
For employees, this truly is a real relief: Instead of maintaining data in three or four systems, they can use the digital assistant to control and parallelize all systems using a single point of interaction – one interface. Better yet, this interface even speaks the language of the user – and not vice versa. This results in enormous gains of efficiency and effectiveness.
A holistic approach: Enterprise Automation Concept
Which intelligent automation suits my company?
The approaches and technologies presented in this article are not to be considered binary choices – it’s quite the contrary, actually. The technology and approach should be selected depending on the use-case, and then developed in order to maximize the use-case-related value. For example, data can be moved from A to B in many ways-finding the most resource – your task is, to find the most efficient way for your organization that meets security and compliance requirements. The composition of the various technologies in one company can be summarized in an Enterprise Automation Concept and systematized for your own company.
Digitale Assistenten + Cognitive Automation + Robot Process Automation = Holistic Automation-Strategy
By combining all three approaches, synergies can be used optimally. For example, digital assistants are ideal for capturing and structuring data and providing the end user with an interface to interact with (partially) automated processes. Using Cognitive Automation, partially structured data can then be captured, interpreted and processed. This again enables the use of Robot Process Automation in the final step, in order to optimally connect the sub-process-pieces together.
Creating an Enterprise Automation Concept with Neohelden
We are happy to support you and your company on the way to the AI-based automation of processes and tasks. To do so, we build on our digital AI-assistant Neo and integrate existing tools and legacy systems within a proof-of-concept or pilot project. Let us discuss your use cases and automation potentials.
Leave a Reply