Other parts of this series:
Roboticizing a process is typically—and appropriately—not a laborious effort. These projects usually are measured in weeks, which means that training a robot takes about the same amount of time needed to train a human worker. Complex tasks, however, might necessitate a longer training period.
But how aggressively should a financial services institution (FSI) or insurer implement robotic process automation (RPA) throughout the organization? Given the efficiencies we have seen RPA deliver, which I noted in my last post, many management teams are eager to scale up the business case for robotization. But as Accenture explains in two reports, Robotic Process Automation and Insurance Robotic Process Automation , management ought to move more slowly.
Taking a circumspect approach has nothing to do with an organization’s information technology (IT) fortitude. The best RPA software should be sufficiently flexible to meet the most robust IT standards for operational integrity, security and supportability. The key design point of the best RPA software is that it does not change any underlying systems, which typically would be a complex and expensive undertaking. RPA software gathers data and integrates processes at an abstracted level, using a variety of techniques and interfaces that ensure underlying systems are not affected. Indeed, RPA requires no IT. Business users show a robot what to do so it can perform work that otherwise would require people.
Still, FSIs and insurers should not scale up RPA enterprise-wide in one-fell swoop, because not all processes can be automated with equal results. So the question about RPA isn’t, “Can we?” but “Should we?”
Other industries that have successfully rolled out RPA programs understand this. In those industries, management took a strategic, holistic approach to building their RPA capabilities. Instead of rushing to scale up the business case for RPA, they typically followed a more structured process-by-process approach. They studied the tasks that humans were currently performing organization-wide and analyzed process characteristics to identify which tasks were the most repetitive, high-volume and rules-based. Those that are fall in the scope of automation. Those that are judgement-based don’t.
This approach also prevents disconnected robotics capabilities from springing up. Those programs’ managers have identified and answered key strategic questions regarding program governance, sponsorship and alignment with business and IT change. That is important, because as RPA programs gain traction, they inevitably will affect existing capabilities within the business. By ensuring a common approach across business and IT, RPA implementations avoid clashes with other change programs—planned or underway.
Typically, companies should configure between one and 10 processes initially. Then a rolling program of processes can be introduced once an operational agility framework has been established. The average time to establish the framework is between four and 12 weeks from project initiation
How easy is it to train and manage a robot’s activity? A robot is trained through a procedure flow chart, which is managed and audited to document the procedure. Management information (MI) is gathered automatically as the robot operates, which allows developing and fine-tuning a process in light of real data. Modern robots systems also include failover and recovery capabilities so that if changes or downstream failures occur, a “smart” response can be trained into the overall system. Before a robot is fully committed, it is monitored through all its procedural steps with MI and monitoring tools to validate the procedure. The robot’s initial go-live speed can be set on trickle to ensure that it is acting according to requirements. Once that is verified, the process can be accelerated to mass automation speed.