If your job involves looking into digitization opportunities and automation of business processes, it’s not far reaching for you to come across awareness for robotic process automation (RPA) and cognitive automation. RPA is not new; it has been around for many years in the form of screen scraping technology and macro.
There are many bombastic definitions and descriptions for RPA (robotics) and cognitive automation. Often, marketers even refer to RPA and cognitive automation, simply interchangeably with the A.I. (artificial intelligence) umbrella. Perhaps, the easiest way to understand these 2 types of automation, is by looking at its resemblance with human.
- RPA (robotics) is to mimic human actions
- Cognitive automation is to mimic human thinking
Robotic Process Automation (RPA)
Look at the robotic arms in assembly lines, such as automotive industry. A robot doesn’t have to “think”, but to repeatedly perform the programmed mechanical tasks. It repeats what have been programmed into it.
Similarly, in the software context, RPA is about mimicking human actions in an automated process.
Examples:
- Automatically extract information (using OCR) from a payroll account statement submitted as part of loan facility application, then, calculate 3-month average income, and compare that with the applicant’s declared earnings.
- Automatically repeat a sequence of clicks in a standalone software to generate cost analysis over a material costing optimization proposal.
From the above 2 examples, it’s easy to observe that the biggest benefit of RPA is savings in time and cost on repetitive tasks otherwise performed by human. Aside from that, quality consistency is another key advantage.
If one promotes RPA to you, as if it was the mother of all automations and the only new form of workflow automation, be wary!
Cognitive Automation
Cognitive computing tackles problems by thinking, reasoning and remembering – mimicking the way humans “think”.
In cognitive computing, a system uses the following capabilities to provide suggestions or predict outcomes to help a human decides.
- Natural language processing (NLP)
- Machine learning
- Cognitive analytics
- Sensing
Examples of cognitive automation use case:
- When a customer’s name is flagged from AML screening database during a e-KYC (electronic know-your-customer) process, cognitive system can automatically interpret the type of risk involved, then send out automated email message that prompts customer for additional information and documents. The enhanced due diligence activity can be automated through chatbot that involves natural language processing. With machine learning, risk analytics is a predictive outcome from historical data and patterns, to help human make quicker decision from more concise recommendations.
- Machine learning programs can identify anomalous actions for near real-time fraud detection. In banks, machine learning can establish patterns based on the historical behavior of account owners. When uncharacteristic transactions occur, an alert is generated indicating the possibility of fraud.
Cognitive computing is not a machine learning method; but cognitive systems often make use of a variety of machine-learning techniques.
Cognitive automation is not meant at making decision on behalf of human. But, interpreting information the way human thinks, and constantly learn, to provide possible outcomes in assisting decision making. However, do note that, bad assumption leads to bad conclusion – no matter how concise a computer is in the process of thinking.
We hope this post achieves its objective at sharing some insights into the recent development in business process automation. Should you have more thoughts and experience to share with us and our readers, feel free your comments.
