I recently attended a conference with the theme “Disruptive Creativity,” and a recurring topic of discussion was around robotics process automation (RPA). While this technology purports to replicate a human interacting with one or more computer systems, these aren’t the robots we know from science fiction movies (R2D2, iRobot, Terminator). The applications range from expensive cognitive computing systems like IBM’s Watson, to relatively simple and inexpensive to deploy software systems that automate a process such as matching and posting incoming invoices.
In this blog, I will discuss the applications at the low end, because I think this area will receive massive attention and growth in the next few years. These applications are relatively easy to build with current technology, promise a clear ROI, and offer exponential process improvements. All of these could lead to a rapid uptake. The implications for the workforce are massive; but that is a discussion for another post.
RPA is suitable for processes that are repetitive, deal with huge volumes and/or periodic fluctuations in volume (e.g. during month-end) and have rules that are well defined and mature processes. RPA has been proven to outperform humans, as it is not prone to fatigue, can work 24 hours per day, and can be scaled (switched on) at peak periods and eliminate errors from transcription of data.
So what does robotic process automation do? RPA in relation to business processes is software that is configured to complete work previously done by people. This includes tasks like transferring data from one or multiple input sources like e-mails or spreadsheets to systems of record. RPA does not assist people in doing their job more efficiently; it replaces some or all of the work previously performed by a person.
One example is the application of a pre-calculated credit to a customer’s account. The process performed by a person looks something like this: The person runs a report in SAP to extract the payment record of a specific customer for the last 6 months. The data from the report is copied into a spreadsheet to analyze the risk, and calculate a credit rating. The credit limit is then applied to the customer’s account in SAP. In a Telefonica O2 case study, this process was completely automated. The robot logged into the system of record, extracted data, estimated the calculation, and recorded the results in the customer record. This configuration of the robot took approximately 2 weeks.
An important premise of RPA is that instead of re-engineering the process or modifying business rules in the systems, the automation software interacts with the application layer of the existing systems. No business rules are changed, and no software changes in the existing applications are required. Telefonica has deployed 160 robots automating 15 core processes, resulting in volumes of up to 500 000 transactions processed per month. They’ve also reported 3-Year ROI of between 650-800%.
Those are massive benefits from an early adopter of the technology, and will likely be felt by many organizations as these tools become more readily available. In my next post, I will take a closer look at the prerequisites for such tools.
Questions or comments about this article?
Tweet @Winshuttle to continue the conversation!