Digital workforce has evolved as an integral part of modern-day enterprise in the last few years. The Robotic Process Automation (RPA) market has already grown by 63.1% in 2018 to $846mn1 and is estimated to be worth $2.4bn by 20222. Digital workforce mimics human actions on the user interface layer of applications, which makes it more user friendly and secure, as compared to traditional API based automation.
Let’s understand this with the help of an example. Consider a scenario where an HR executive has to add the details of a new employee received through E-Mail into the ERP system. The HR executive performs the following steps to complete the job:
A digital worker would mimic each of these steps with far better speed and accuracy than the HR executive. Furthermore, digital worker keeps going 24*7 without any breaks.
While digital workforce has become an important part of modern-day enterprises considering the advantages and value it has to offer, there are certain ground level challenges, which have to be addressed. For example, it is a very common scenario that an enterprise application is down for maintenance during a certain period. The human worker understands that and tries to complete the job either by accessing a different application or by attempting the job post the maintenance period, whereas a digital worker throws an exception stating that application is not available. In order to ensure the continuity of business-critical processes being executed by digital workers in a fragile application and infrastructure ecosystem, the key challenges have to be identified and handled in run time.
Although the road has been paved for organizations to start their automation journey and reach the destination, there are bumps on the road. These bumps need to be identified and called out loudly so that a strategy to drive through these can be devised. Following are a few challenges that organizations may face in their automation journey:
Digital workers access specific applications in order to complete the assigned job. In continuation with above example, a digital worker needs to access an ERP application to add or update employee records. While it attempts to access the ERP application, the application may be unresponsive due to various reasons cited below:
In each of the above scenarios, the digital worker is unable to update the employee records and thus is not able to complete the assigned job.
Digital workers need login credentials in order to access applications that require sign-in.
As per the standard organization policies, the application passwords or the SSO passwords expire after a certain defined period and have to be reset. Once the credentials are reset as per the organization’s policy, digital workers are not able to access the application with the earlier configured credentials, thus are unable to complete the assigned job.
Similar to human workers, digital workers execute the assigned jobs by working on specific application on a given machine. But unlike human workers, multiple digital workers share the same machine to execute the assigned jobs in order to optimize the usage of the machine.
Like any other software, digital workers too encounter exceptions occasionally. The administrator and the monitoring teams thus access the digital workers’ machine in order to understand why the exception occurred and collect the required information to fix the exception & to make sure that it doesn’t happen again. Since multiple digital workers are working on the same machine, it may happen that while the monitoring team is analyzing the exception encountered by one of the digital workers, they may accidentally access applications being accessed by other digital workers.
Consider a banking scenario where money transfer in a specific currency is involved. Even a small accidental change in currency can cause huge financial irregularities for the organization.
As digital workers are configured to perform actions on specific applications in order to complete the assigned job, they identify the area on the application UI, to act upon, on the basis of various properties such as ID, CSS3 selector, class, coordinates, image etc. In case, these application properties change, digital workers won’t able to identify the right areas to act upon and encounter exceptions.
As an example, consider Mr. Jack ordered a pizza from a website, which has his address details saved as House No. 67, but pretty recently he moved to house no. 520 in the same locality. The delivery boy who has to deliver the parcel to Mr. Jack is attempting to reach him at House No. 67 and is not able to deliver the pizza. In a similar way, if the address of the UI elements on an application UI changes, the digital worker won’t be able to find the right address in run time.
Modern day digital workforce works closely with cognitive services and models to intelligently understand data and make decisions. Cognitive services such as OCR and VISION enhance the capability of digital workers to identify and understand the relevant data whereas Domain specific cognitive models enable the digital workers to make business decisions in run time.
For example, the digital worker has to make a decision regarding whether or not to approve the claim filed by an employee using the submitted travel bills, claims history and other employee related data. It performs the following steps:
But there are times when digital workers are not really confident about their understanding of data they have extracted and the business decisions they have recommended. In such scenarios, digital workers are not able to complete the assigned job.
Digital workers while performing their assigned jobs encounter expected or unexpected business exceptions. Consider an example where invoice details have to be stored in an excel sheet. Digital worker performs the following steps:
Now if the invoice number is incorrect or the details have still not been updated in the Invoice Management System, while attempting to search for the corresponding invoice, the digital worker would get an error regarding invoice’s non-existence.
In such cases, digital workers won’t be able to complete the assigned job unless specific measures are taken to correct the source data.
Just like human workers, digital workers also have limited bandwidth to complete the assigned job. In every organization, there are processes, which are high priority & business critical and then there are others, which are relatively of lower priority and can be completed a little later.
Initially, a specific number of digital workers are assigned for high priority process depending upon the available trends. More often than not, there are times, when there is unexpectedly high workload and existing bandwidth of the digital workers is not sufficient to cater to the same. Although the digital workers complete the assigned job successfully, they miss the defined SLAs.
Organizations may face one or more of the above challenges while in their automation journey. The strategy to manage these challenges has to be in place for a successful automation journey. Although the RPA market is growing rapidly, these challenges can hold back or hamper the success of their automation journey. In a recent study by EY, it was found that 30-50% of initial RPA projects fail 3. It is not as simple as it looks. In my upcoming blogs, we’ll talk a lot more about how to cater to these challenges so that organizations can drive over the bumps without any jerk.