In this second instalment of our series, The Future Tense of Work, we focus on Integrated Automation. This topic lies at the intersection of people (employees) and technology dimensions in the model below. For more detail on this model, you can re-cap on our introductory article.
Automation can be achieved using several different technologies; bots (i.e., attended & unattended RPA), APIs, events, smart workflows, case management and AI (vision, speech, language, decision).
Often it can be confusing as to which automation technology should be applied to which context. Marketing hype created by automation tool vendors only amplifies such confusion by creating the perception that “one size fits all.” The nature of an automation problem can be determined by looking at 5 key parameters i.e.-
The matrix below uses the above parameters to group the automation landscape into 9 possible prioritised segments, associated with different automation technologies.
The different automation technologies highlighted in the model are:
RPA is mainly focused on automating mundane, low-value, repetitive tasks carried out by office workers in their day-to-day jobs. RPA acts as an employee productivity tool freeing up employees to concentrate on higher-value tasks. RPA is implemented by building automated scripts, which operate predominantly via the existing user interfaces of enterprise and desktop applications in a non-intrusive way. This is applicable for older systems that would not necessarily expose APIs for integrating data, or when automating smaller, simpler desktop tasks where integration via APIs can prove to be more complex and time-consuming.
This refers to the RPA mode where a bot is triggered on an agent’s desktop to execute alongside their day-to-day manual tasks. Attended RPA is primarily applied to automate simple repeatable tasks such as inputting, retrieving, or moving data between one or more desktop applications.
Such automation is best suited to employees working in front-office functions such as contact centres, where automated repetitive data processing tasks can significantly reduce transaction handling times and boost productivity.
This refers to the RPA mode where an automated process containing rule-based tasks is triggered to run on a server machine independently without any human intervention. A bot can be scheduled to run a batch of transactions in this mode and can handle most exceptions itself; or send a notification to a human, where it is unable to resolve an exception by itself. Unattended RPA is better suited to back-office functions that are standardised and transactional in nature, for e.g., invoice processing. The main benefits here are to improve operational efficiency and increase operational throughput.
This refers to business activities carried out in the form of cases, where a case can be defined as a unit of work that does not have a fixed boundary to be pinned down as a specific business process workflow. The activities carried out in a case may not happen in a fixed sequence. The resolution path a case may take can be unpredictable and may require multiple sub-processes or individuals to be involved. A typical example of case management is incident management.
Unlike cases, these refer to business processes that can be represented as rule-based workflows with well-defined activities and outcomes. Such processes have a higher occurrence than cases and can be orchestrated for execution purposes. Given that the level of predictability is higher, orchestration allows many of the steps to be executed as automated and to bring in human workers where needed.
These allow building application & data integrations between multiple systems, devices, and data sources to enable straight through processing typically via API-based connectors. Integrations of this kind often require data transformations to support multiple formats and communication protocols such as SOAP, REST, JSON, XML, etc. These are either processed in a batch mode, or as request-response in real-time, or as asynchronous messages processed via an enterprise service bus using queues or topics.
This refers to the integration of data as a continuous stream of events processed in real-time, which is typically needed in large distributed systems with very high transactional volumes triggered via a number of digital channels and devices. Here scalability is the crucial requirement for example, in digital payment processing, or in digital e-commerce business models such as Uber or Netflix, or in social media platforms such as LinkedIn.
This refers to artificially intelligent cognitive capabilities that augment human workers in order to develop knowledge for analysing and processing highly complex unstructured data that relates to speech, language, vision (handwritten text, chat conversations, images, videos), and decision making. The key technologies involved here are Natural Language Processing, Conversational AI, Computer Vision, Machine Learning models and Big Data Analytics.
Integrated Automation is all about combining different automation technologies into a unified approach for solving different problems related to business process automation within an organisation. Definition may vary; from being a less predictable high-level case with a low-frequency occurrence to a business process with better-defined workflow activities and a higher rate of occurrence, or specific low-value rule-based high-volume tasks. These different kinds of automation problems may be related to each other as depicted in the figure below.
Integrated Automation is all about augmenting humans with automation technologies. Depending upon the nature of the automation problem, which may fall in one of the 9 segments depicted above, a combination of automation technologies may apply.
For instance, a case with higher transactional occurrence but inherently low predictability may be solved by augmenting human workers with case management and predictive machine learning (ML) model (AI) technologies.
A business process with higher transactional occurrence than a case, and with medium predictability and programmability, can be solved by augmenting human workers with smart workflow and attended RPA technologies. Similarly, a task with high transactional occurrence, with even higher predictability and programmability, and with high levels of unstructured data processing needs, can be solved by augmenting human workers with a combination of unattended RPA, API level integrations, and AI technologies (computer vision, speech recognition, natural language processing).
When analysing the automation problem space based on the 5 parameters mentioned at the start of this article, the diagram below depicts how different automation technologies may overlap when looking to solve varying automation problems. To achieve strategic outcomes, it is imperative for businesses to take an integrated approach to automation and to invest in automation technologies wisely.
Contact us for more information on how we can help you with your integrated automation needs.
