In the context of process automation, in simple terms, intelligent automation combines artificial intelligence (e.g., natural language processing or NLP, machine learning or ML, and computer vision) with robotic process automation (RPA) and document capture and processing capabilities. This combination can be extended to include business process management suites (BPMS) and process mining tools. In terms of their specific functions, RPA and document processing is the “doing/execution” aspect whereas AI/ML is about “thinking” and “learning”.
The important point is to notice that how such a solution architecture composed of various software products can deliver a greater degree of automation, ultimately leading to solutions to use cases that would otherwise seem too difficult to achieve. Take the case of a good old optical character recognition (OCR) engine, while it helps with digitization of paper-based information assets, the inherent quality issues are hard to ignore (accuracy in best-case scenarios with legacy OCR is about 60%). Talk to an enterprise IT leader about OCR and they will quickly run away mentioning that OCR is of little use, if it were to deliver a strike rate of only 60%, better not do such type of automation.
With the application of AI and ML, and in its reincarnation as intelligent OCR, the same OCR tool is able to deliver a far better accuracy; intelligent document processing (IDP) of course, goes a step further with pre-and post-processing steps. In a true sense, the application of AI and ML has reinvigorated a market segment that was on a gradual decline.
Over the years, the BPMS market has slowed down, with other light-weight approaches to workflow management and management of process-centric applications gaining ground. It is noteworthy that low- / no-code platforms have gained mainstream adoption over the same period. Let us take a combination of RPA, IDP, and AI/ML tools to elucidate the paradigm of intelligent automation. With the combination of these capabilities, users can build and execute intelligent automation workflows, wherein IDP is used for data ingestion and processing, RPA automates a series of tasks (or sub-processes) and AI/ML capabilities improve the strike rate and deliver error resolution. When seen closely, such a holistic approach combines process- and data-driven approaches to automation, as the focus shifts from automating a series of tasks towards achieving end-to-end automation (ideally, straight-through processing or STP).
As RPA enters the mainstream and consumer-grade user experience (UX) with a graphical approach to the development of automation flows gains popularity, the barriers to adoption are significantly lowered, both in terms of skills required and the total cost of ownership of automation software. Intelligent automation expands the scope and applicability to a diverse range of business processes and quite simply, increases the degree of automation that can be achieved when compared to the discrete use of RPA, IDP, and AI/ML capabilities.
Results from a survey conducted by Deloitte* reveal that increased productivity, cost reduction, and improved accuracy are the top three expected benefits from intelligent automation. Executives at surveyed enterprises estimate that adoption of intelligent automation can lead to an average cost reduction of 22% and an increase in revenue of 11 % over the next three-year period. Clearly, intelligent automation has the potential to deliver transformational outcomes, especially with automation of predictions and decisions based on both structured and unstructured data.
The intelligent automation continuum establishes AI/ML as the key component to move beyond task or sub-process level automation and ensure that documents with unstructured data are processed with a greater accuracy and speed as part of an integrated approach to automation. A combination of RPA, IDP, and AI/ML can automate complex business processes to a greater degree and free up humans to work on more strategic and high-priority initiatives. There will always be cases where human judgement is required for exception management; this can be achieved with human-in-the-loop capabilities, similar to case management in a traditional BPM approach.
A good case in point is of a leading bank that improved customer service using intelligent automation. The bank was looking to develop intelligent email analytics for quick processing of 10,000+ email service requests received on a daily basis. It used a combination of AI and RPA tools to decipher customer sentiments, scrutinize the incoming emails, and auto respond to customer emails. With the intelligent automation solution, the bank improved the speed of acknowledging emails from customers to just 5 minutes, thereby improving customer satisfaction by 80%. The intelligent automation solution reduced the time lag in routing the emails to the concerned function by 70%.
A premium housing leasing company based in the US streamlined and automated an accounts payable process using a combination of RPA, IDP, and BPM tools. The BPM tool was used for workflow management. The intelligent automation solution improved accuracy to 99.8%, enhanced productivity, and reduced turnaround time (TAT) by 40%. The housing leasing company was able to scale up processing to 65,000 invoices and 30,000 auto-uploads per month, with 100% process transparency.
*Source: Deloitte Insights, Automation with intelligence, 2019