Robotic Process Automation involves automating business processes using software robots or tools. RPA has many similarities to GUI testing tools in software but has better features to handle data.
The RPA revolution is already transforming enterprises in the areas of medicine, pharma, insurance, manufacturing, and many others by automating repetitive tasks in any process and greatly reducing human intervention.
RPA is a relatively simpler tools- based approach, which mimics human actions. It can be effectively augmented by using Artificial Intelligence in what is called Cognitive Automation.
RPA when combined with AI technologies like natural language processing, speech recognition, image processing, can make systems do judgment and perception based tasks. This is called cognitive automation. In such cases, it can mimic human mental capacities like cognition, evaluation, prediction and so on.
Simply defined, RPA uses software robots to carry out repetitive human tasks, which do not require decision making or using judgment. Some techniques it uses are screen scraping, scripts, and workflow automation. It does not require coding and is just a proper configuration of some of the software frameworks, using if-then logic and in a process. It works with structured data only. RPA can rapidly automate business processes without disrupting existing systems. For example, forwarding service desk tickets to an appropriate category based on some text in it. RPA reduces human effort and complements it.
However, RPA requires all data to be in a structured format. Some of the well-known RPA tools are UiPath, Blue Prism, and Automation Anywhere.
Cognitive automation is much more complex than mere RPA. It can minimize or even eliminate human effort in a process.
Using AI techniques, RPA can go beyond mere automation of mechanical tasks and copy human behavior by looking at vast quantities of data and making sense of it. It is meant to process unstructured data, unlike RPA.
The machine programs observe several human conversations, patterns, and behavior to make decisions and learn from them. While RPA accepts structured data into a system and processes it, only Cognitive automation can take it further by finding out correlations in the data, patterns in it and learn from those to help make judgments and predictions.
Therefore, Cognitive automation can be said to be AI applied to automating business processes over and above RPA.
Most data processing flows take the path: data – judgment – action based on an algorithm and can use RPA. With cognitive automation, you can add some predictive capability to large quantities of unstructured data. Unstructured data cannot be processed by algorithms alone and cannot yield information easily. Some techniques for unstructured data are natural language processing, images need vision-based techniques, and audio needs speech recognition techniques to yield usable data. It creates metadata for this unstructured data like tags, indices, etc. Robots may be able to read anything seen on a screen and make sense of it. Emails, bank statements can be read using keywords and processed by a robot.
This information can now yield trends and patterns that can be used to make predictions. These lead to actions that are not based on hard rules but more on observation of past patterns and decisions made on them, for example, when to write off a loan. So, while RPA is data intensive, cognitive automation is information intensive.
Though human intervention cannot be eliminated at this stage, at least the best possible option can be suggested by the system. Based on this actual action is triggered.
If such a system is looking at business documents in particular business flow, it will try to find similarities between documents in various areas of the business: say 1. An invoice, 2. PO numbers 3. Shipping addresses, and try to build a correlation of these.
If X was seen before, what action was taken? How is it connected to some other areas? How well is it connected?
The more data you add, the more the systems learn, that too in an unsupervised way.
To sum up, Cognitive Automation add intelligence to data by simulating a human thought process, whereas RPA will improve the accuracy of a repetitive business process leading to better customer satisfaction. It is like using a human hand versus using a human mind.
Cognitive Automation is radically transforming many industries leading to reduced costs and increased speed.
Some pre-trained cognitive automation solutions for specific industry business processes are already available and can be deployed with minimum effort.
Current applications of Cognitive automation are in many sectors. Some vendors provide IQ bots that you can use without having the need for data scientists. Any process that needs humans to organize large amounts of unstructured data before processing can begin, can use this kind of ready-made solution.
The bot helps in the end to end data processing without the need of an expert by smartly applying a combination of the following technologies: machine learning, NLP, fuzzy logic and so on.
Financial Services have been early adopters and will prove to be the biggest market for cognitive automation. “Amelia” is a multi-lingual software robot for this industry. Here cognitive automation can help in investment decisions.
In the Insurance industry, it is deployed for fraud checking and policy renewal claims processing and so on. These systems can work 24 by 7 unlike humans and lead to huge cost savings. France’s AXA group is using this for their underwriting tasks.
In the IT industry, it is most widely deployed to monitor application health and optimize testing.
The Healthcare industry has a lot of documents to manage, like patient records, reports, medical images, and this field is ripe for applying cognitive automation. Maintaining paper files and transferring them to digital format is a suitable task. Based on patterns in past data, such a system can help maintain proper inventory of medical supplies and predict patient turnover and admissions, leading to better management. Cognitive automation can help with the regulatory requirements in the drug development and delivery Pharma sector.