CIO Insider

CIOInsider India Magazine


The Unreasonable Effectiveness Of BOTs

Pradeep Rajendran, Technology Leader, Federal Reserve Bank of St. Louis,,

A proven strategic leader with a blend of leadership and technical skills. At Federal Reserve Bank of St. Louis, he is spearheading technology strategy, roadmap and implementation while managing budget and IT procurement for the division.

The golden age of automation is here! We are rapidly moving towards Hyperautomation where bots combine current and emerging technologies to deliver greater efficiencies. Let’s dive into some background to understand the forces that are increasing the effectiveness of bots and driving Hyperautomation.

Evolution of Automation bots
Sure, the term ‘workflow automation’ has been around for a long time, and the ability to interact with legacy solutions with approaches like screen scrapping and Optical Character Recognition (OCR) has been around for a while too. However, these methods on their own were not scalable. They were brittle and required programmers to write scripts for scraping and automation.

To fill this gap entered Robotic Process Automation (RPA) that allows us to automate using “low-code” drag and drop mechanisms. RPA allows us to connect legacy systems that lack APIs to modern systems and move structured data between them better, faster and more consistently than humans.

RPA is only ideal for very short-lived or repetitive tasks that do not require decision-making or human intervention. RPA is used more for clerical processes that have rigid structures rather than complex decision driven processes. Despite this limitation, it offers a great increase in productivity, reduces error rates and increases employee morale by removing the need to do rote and repetitive tasks.

Machine Learning and Natural Language Processing
While RPAs were being slowly adopted by the enterprises, the Artificial Intelligence (AI) revolution was accelerating. With increased processing power and Deep Learning, Natural Language Processing (NLP) went from awkward to robust leading to an explosion in Chatbots and Conversational Platforms.

Conversational Platforms, armed with backend information retrieval capabilities are being deployed widely. Customers use natural language with ease to interact with these chatbots and smart devices like Alexa and Google Home. While this has

opened the door to a new and drastically improved user experience, most chatbots are still being used to provide information rather than triggering complex processes.

While process automation, NLP, Machine Learning models, etc. were becoming more sophisticated, it still wasn’t easy to use them. Organizations were required to buy powerful hardware, build custom machine learning models, and hire data scientists and technologists to leverage all these advances. All of that changed when cloud providers like AWS, Azure, Google, etc started competing to provide these capabilities as Platforms as a Service (PaaS) or in some cases even as a Software as a Service (SaaS).

Democratization, leading to citizen access and no-code/lowcode models, has thrown the doors wide open. Suddenly even small organizations could afford to implement automation. The focus shifted from technology to business opportunities

Democratization, RPA, artificial intelligence, and NLP combined into the perfect storm to give us cognitive automation and HyperAutomation

Cognitive Automation and Hyperautomation
As organizations mature, they find RPA restrictive. To get more efficiency, they need to automate complex decision making, understand and process natural language, and use machine learning to handle complex scenarios and unstructured data.

Democratization, RPA, Artificial Intelligence, and NLP combined into the perfect storm to give us Cognitive automation and Hyperautomation. Complex decision making is now possible with AI and ML. The bots can now ingest and make sense of unstructured data by using NLP and ML models. Data that are in handwriting format or non-text formats can be leveraged using Intelligent Character Recognition (ICR) data, image processing, etc. Organizations can now build a system based on a loose set of rules that can make decisions, learn and evolve to serve the complex needs of today’s organizations.

On the user interaction side, we can have more self-service automation as they use interact with chatbots and virtual assistants that use NLP and can kick off cognitive automation based on the customer input.

Example Use case
All of this is a bit abstract, so let’s look at a very simple use case of customer support. A customer can interact with a chatbot that identifies the customer and provides some initial information. However, based on the customer’s input, the chatbot recognizes that further action is required. Now the chatbot kicks off the automated ticket and assignment process. The process classifies the ticket, assigns priority and sends it to the right team. This whole flow did not involve a human other than the customer. On the other hand, it uses NLP, text analytics, RPA, AI-based decision making, and Cognitive Process Automation.

(Un) Reasonable effectiveness of bots
Advances in Artificial Intelligence and NLP have provided the ability to interact with customers in natural language, process unstructured data, learn from data and make decisions. Having understood the background and how so many technologies and breakthroughs are being combined, I am sure you will find that it is not at all unreasonable that bots are now remarkably effective. We can automate more and faster than ever before ushering in the age of Hyperautomation.

The most exciting part of this story is that this is just the beginning of the automation revolution. As companies mature, they will embrace multi-experience, IoT and other approaches to get more data and to enable more complex automation.

Current Issue
Doing Business The Intelligent Way