Task Mining: The next BIG thing for RPA

With the announcement of Nintex’s purchase of Kryon, it’s clear that Process Discovery and Task Mining are the next logical steps in the evolution of RPA (Robotic Process Automation) and potentially BPA (Business Process Automation).  There is a big difference between Process Discovery (PD) and Task Mining (TM) and it starts with each tool’s approach on how it gets the data.  Our team, customers, and prospects are really excited about these new initiatives. 

 Process Discovery provides a macro view of an enterprise and can show the system(s) where people are struggling and spending most of their time. Process Discovery uses the system logs to determine commonalities amongst many users. While this is great from a top-level perspective, the agility of such an exercise is less pronounced and the follow up initiatives take longer to implement.  

 Task Mining on the other hand, is a micro view of an enterprise. A successful implementation of it consists of a group of 5-7 workers who are performing a somewhat common function. Think of a medium size accounting team within an organization who key vast amounts of data into their accounting system and other supporting applications. Task Mining is installed on their desktops. It then collects a large number of screenshots and learns commonalities and shared processes to determine  the opportunities for automation. Task Mining can discover processes that haven’t been thought of previously but are worthy of automating. In addition, it can help identify new bot potential in areas of focus as a company’s automation journey evolves and matures.   

 I am very bullish on the potential use and adoption of the technology.  There are, however, some limiting factors which relate to scale (not enough computing power) for it to be widely adopted on a mass scale. A starter server for the tech is about 20 GPU’s to handle a small group.  This makes sense on the level of computing it needs to do to accomplish its task. The computing power will be less of a factor as computers can handle more of load, which they will need to do as the adoption of AI increases in the coming years.  

 It’s my thought that the market is underestimating what Task Mining can mean for automation on a widespread level. The ability to discover tasks that can improve work at every level is something far more strategic than most can comprehend.  

-By Peter Camp, Founder & CTO