Despite all the buzz around continuously learning virtual agents or intelligent automation a majority of robotic process automation (RPA) use cases still focus on simply replacing repetitive and routine human activities with robots. According to a recent Forrester study approximately 95% of today’s RPA implementations are limited to dealing with structured data. The next trend could be to have more and more robots deployed to work with unstructured data with self-learning abilities.
When we approach typical RPA opportunities today, we tend to focus on the set of actions or activities (currently performed by humans) which can be replaced with robotic implementations, thereby, freeing up human resource to focus on more “value added” activities. I see the following possible outcomes on the business processes after a successful RPA exercise:
- Improvement of process performance or improvement in turn around time due to faster task execution through automation
- Less rework since human error factor is negated
- Process becomes less costly
Transform your processes
But, in simply replacing human actions with robots we end up not optimizing our business processes. While analyzing the operational procedures to identify use cases for RPA we should take this opportunity to optimize our processes as well. Let’s take a simple example here: a typical RPA use case would be where a bank employee is keying in information across multiple legacy systems while acting on a customer on-boarding request. While identifying this stage of the process as candidate for RPA based automation we tend to forget to ask several questions:
- Do we really need to enter the same information across all these systems still?
- Is capturing all the information still relevant?
- Why do we need to have that quality check process in the next step? Robots would not make mistakes ( like humans ) in entering data across systems under normal circumstances
The above are just some simple example of things that we tend to overlook and simply replicate the as is process steps.
The idea here is to make RPA initiatives transformational by spending some time to optimize the business processes we are dealing with before the actual implementation. This would ensure better outcomes and perhaps better ROI. Poor processes can be automated but it is better to improve them first. Automating a bad process does not make it better. In the short term, the process may become cost efficient by removing human workers, but our end goal should be to make the process more efficient and effective.
RPA and digital transformation
In a lot of cases we look for only quick wins from RPA deployment initiatives by either reducing human interventions or completely replacing those, especially in an enterprise where many software systems are deployed. The other common lookout is to find an alternative to complex and expensive integrations.
While the above establishes a business case for RPA use, it is important to look at such initiatives a bit more strategically. For example, an enterprise should look at where does an RPA fit in, in the overall automation/digitization roadmap, and how such capabilities can complement business process management (BPM) and integration initiatives.
While identifying use cases for robot deployment it is advisable to first optimize the processes and remove activities which seem redundant in today’s context. It is also a good idea to investigate how the process can be made more dynamic and efficient with robot deployment where faster and error free throughput is expected.
Combining RPA with AI
While most of the RPA deployments deal with structured data it is time to look at opportunities to make such automation initiatives more intelligent. Enterprises, in their automation roadmap, should look at augmenting RPA capability and span of usage by complementing the same with AI. Processing unstructured information, determining next set of actions and executing those would make robotics more attractive and usable. For example, use cases where understanding content of documents (unstructured information) and thereby taking a set of actions are applicable, would be good candidates for AI enabled robotics. To make a stronger case consider leveraging machine learning capabilities to continuously train robots for better performance.
RPA and BPM
It needs to be clearly understood that RPA is not meant to replace BPM or BPMS (business process management suite). While BPMS based automaton initiatives usually encompass larger functional and process areas, and are more strategic, long term initiatives, RPA engagements usually happen in pockets and are tactical in nature. Some of the situations where RPA gets precedence over BPM would include:
- The process or sub-process need to be automated for a short period of time or for the duration of a specific activity
- Internal and external systems cannot be linked via intrusive integrations
- Functionalities required may be overkill for BPMS implementation
- For some reason the organization does not have a BPMS initiative rolling yet but need a quick solution to a problem
Connect with me on my Social Media profile.