In its research report titled Cognitive RPA – The Future of Automation (Jan 2019), NASSCOM states that “enterprises are moving from experimentation stage to early adoption with a focus on adding cognitive components to RPA solutions”. Businesses on a large scale are using RPA products to perform rule-based automation in order to speed up processing time and to minimize error rates. But most of the existing RPA-based initiatives are tactical, focused on driving down costs. This has to change, if businesses really long to embrace true digital transformation.

Finding the right balance between the best technologies available is essential to drive digitalization. In the next couple of years, within the space of artificial intelligence, the highest spend will be on Cognitive and AI-based applications. However, the unprecedented growth of these technologies is creating confusions among enterprises with respect to determining the right technology to invest in. So, how can one decide between investing in RPA and cognitive automation?

To help you make a more informed investment decision, we are demystifying both the concepts through this post.

Making sense of the jargons

Where do you apply Robotic Process Automation (RPA)?

- to mimic recurring, mundane human tasks with more precision by using software robots

- in scenarios that do not require decision-making or human intervention

Where can you apply cognitive automation?

• situations that require human involvement in decision-making such as the presence of voluminous data, which is a challenge for human workforce to make the right decisions

Like RPA, cognitive automation also mimics human behavior, but in ways more complex than the actions and tasks imitated by RPA. An example of this could be how a doctor leverages cognitive automation to diagnose a patient’s conditions and decide the course of treatment.

Why move from RPA to cognitive automation

The growing prominence of cognitive technologies in RPA is manifested by the recent partnerships established between non-conventional players in the ecosystem. HDFC Bank has launched an Accelerator Engagement Program to collaborate with start-up accelerators all over the world for tapping potential Fintech ideas in the areas of RPA, machine learning, analytics and AI. Similarly, Zone Startups India, a global startup accelerator based in Mumbai has aligned with TVS Credit, one of India’s leading NBFCs to launch “FinDhan”. This multi-format startup engagement program allows Zone Startups India to collaborate with Fintech or enterprise tech startups in areas crucial to financial services including RPA.

At present, enterprises that have adopted RPA are using it to automate recurring activities in areas like procurement management, human resources and recruitment, finance and accounting, contact center, web-based activities like form filling, screen scraping etc. and in IT. From a cost reduction tool, RPA has grown to a phenomenal revenue enhancement tool as it can function autonomously 24×7, in real-time.

Using cognitive technologies, RPA software robots could create and manage robots that can handle modifications in the workload. For example, RPA integrated with AI technologies such as ML, NLP, text analytics, and computer vision can support activities that require intelligence such as data classification and validation, sentiment analysis, customer segmentation, fraud detection, document verification etc.

Technologies underneath

RPA uses basic technologies such as screen scraping, macro scripts and workflow automation for performing tasks. In contrast, cognitive automation uses sophisticated technologies such as Natural Language Processing (NLP), text analytics, machine learning, semantic technology and text data mining which enable human workforce to make informed business decisions.

RPA doesn’t require extensive knowledge in coding since it primarily revolves around the configuration and deployment of frameworks. Cognitive automation, unlike RPA, is based on machine learning and requires application of programming skills on a wider scale.

Automation patterns

RPA is a process-oriented and rule-based (“if-then”) technology that facilitates the execution of time-consuming tasks, whereas cognitive automation is a knowledge-based technology in which the machine is trained using models of human conversations and actions to understand how a human talks, behaves or defines rules in a given scenario.

Imagine an organization that processes thousands of unstructured invoices and purchase orders in their database. Here, RPA can work effectively if all the documents follow the same format. Trained software robots can recognize values entered in corresponding fields and support rapid processing.

Now consider the same example in a different scenario where you have multiple invoices with different formats that have different fields of entry, attribute names etc. In this case, the application of cognitive automation can help in building relationships between entities by putting forth questions such as whether the quality was seen before, how it was used previously, in what way it is connected to the one seen earlier etc. By asking these questions, the automation tool can interpret and process data with minimal or no human supervision.

Data processing

RPA requires standardized and structured data to work on, such as spreadsheets or databases, while cognitive automation works on non-standardized, semi-structured or unstructured data such as emails, invoices or images. These kind of data can rely on cognitive automation to establish relationships and find similarities between patterns through constant learning.

When to apply the right technology

Deciding between RPA and cognitive automation depends on the characteristics of your process. If your process involves rule-based, structured, and voluminous data, then RPA is the best choice. If you are handling complex, unstructured data that requires human intervention, then you can opt cognitive automation.

While RPA gives you immediate ROI, it takes some time for cognitive automation to show results as it has to learn human behavior and language to interpret and automate data. If your workflow involves simple tasks and requires human intervention, then it’s better to go for a combination of RPA and cognitive automation.

What lies ahead for businesses

Cognitive technologies can infuse smart capabilities into your existing RPA offerings. This will enable activities that require intelligence. Studies prove that going forward, enterprises will focus on combining RPA with AI technologies such as NLP, ML to create a higher business impact. That’s why RPA platforms like UiPath are investing heavily in areas such as image recognition, machine learning, natural language understanding and generation, process discovery etc.

End-to-end automation brought in by the association of artificial intelligence and robotic process automation is the next big thing in workplace digitalization.