When people think about Robotic Process Automation (RPA), they believe that it is a technology that we may use or come across in the future.
I attended a very informative session on RPA conducted by Nasscom GIC Forum at Hyderabad this week which destroyed this myth. The main speaker was Ankur Kothari, Co-founder of Silicon valley based Automation Anywhere which is a 13 year old company and a leader in RPA solutions. There was a panel discussion too –Sapna Grover of Pega Systems and Subba Perepa of JP Morgan Chase were among the panelists.
Most of the white-collar day to day tasks involve three kinds of activities - Do, Think (Cognitive) and Analyze. Routine daily tasks in the ‘Do’ category contribute to over 90% of regular and repetitive work. RPA is best suited for automation of such routine and repetitive tasks. They may be in the Finance and Accounts, Banking and Financial services, HR-Payroll, Supply Chain and similar functions. They may require some cognitive (thinking) work.
There is a perception that RPA technology is new and therefore can be risky if put to immediate use. In fact it is over a decade old and is mature. We may not be aware, but Bots have been extensively used in the last 6-7 years in commonly used Apps like locating entities like restaurants, petrol pumps etc. nearby.
Business users hold the key to development of RPA solutions – they need to explain the work processes in detail. Implementation team has to guard against the tendency of business users to highlight complex exceptions and ask for these to be automated first. Exceptions do not normally exceed 10% of the work that we do and can be taken up later. ROI is less on automating these, and these should be given low priority.
Manually performed tasks use inputs from a variety of systems – ranging from mainframe legacy applications to latest Apps. RPA mostly takes inputs from the GUI layer and hence can be implemented quickly in a simple way. Complexity of the back end system is of little concern. Manual steps are replicated on screen when Bots run.
When RPA is implemented, Digital Workers (DW) take over part of the work which was being done manually. A DW can be taken as a Bot or set of bots that perform work equivalent of 6 man-hours per day. A Digital Worker does not get tired, is not affected by moods, does not take leave and can perform repetitive tasks without errors.
The cost of human time and effort is more significant in RPA implementation than the technological cost. RPA projects for a single process or part of it can be completed in time frames of 3-4 weeks. It is therefore best to start with a Proof of Concept in one area and develop say maximum 50 Bots with a project duration objective of three months or less. Some of these Bots may be available readymade.
Once this POC succeeds, the organization can set up a Center of Excellence that helps various functions in implementing RPA. In the following 6-9 months, the COE can help develop and deploy up to 300 Bots. This
can then be extended organization-wide and one can see around 1000 Bots in use in the next couple of years.
In Analytics, the current trend is towards real time analysis compared to periodic analysis (weekly, monthly, annual) normally done in business organizations. For example, you would like to see statistics related to Cricket while a match is in progress, not a week later. Implementing real time analysis in business functions changes the perspective completely and can have significant impact on faster decision making.
If your business involves high volume of transactions with sizeable manual effort in tasks that are based on these transactions, it is time that you start adopting RPA. Address the low hanging fruit first - simple processes with high volume.