The rise of robotic process automation (RPA) in insurance has coincided with many similar emerging technologies, such as AI and chatbots, but interest in RPA is more about fixing legacy integration issues than embracing forward-looking tech. It serves as a cost-effective, short-term solution for complex infrastructure issues – replacing manual integration points with a scriptable, automatable solution – and in the insurance industry that may be more important than the most advanced AI available.
Novarica does have concerns about insurers relying too heavily on RPA as a permanent fix, and insurers face the risk of investing too heavily in RPA if they treat it as a long-term solution rather than a temporary fix. RPA won’t make problems from siloed legacy systems go away, and CIOs considering its use should make reasonable plans for true integration and modernization in the future. At some point insurers need to tear off the Band-Aid and consider infrastructure optimization over cost. That being said, it’s also true that even as insurers modernize there will always be new legacy systems in the environment, and it’s likely RPA (or RPA-like technology) will have a place in future architectures as well.
There are two types of RPA vendors: those that offer RPA solutions (software providers), and vendors that resell solutions as well as offer assistance in developing RPA strategies and implementing RPA solutions (service providers). While many vendors are shifting the message about RPA from a tactical process automation technology to more strategic transformational use cases, this is not how RPA technology is currently being used in the insurance industry. Rather, most use cases are tactical in nature, intended to solve specific problems of integration between systems. In a particularly interesting use case, Zurich – in collaboration with Capgemini – recently used Blue Prism’s RPA solution to automate the issuance of standardized international insurance policies.
By and large, RPA is used for sub-processes in billing, claims, finance, HR, and policy issuance, among others, the goal of which is to automate data handoffs from one system to another rather than requiring humans to enter the information manually. This cuts down on the inevitable inaccuracies of manual data entry, and also frees up human resources to perform more complex tasks. One drawback is that the bots have to be reprogrammed anytime forms or processes change, or if there’s an exception that hasn’t been taken into account – a more strategic approach to RPA would involve developing an enterprise-wide platform, centralized governance, and integration with AI-related technologies to not only mechanically execute processes, but to learn, analyze, and make decisions.
For more on this, see my recent brief,
This blog entry has been reprinted with permission
