December 14, 2017
Rise of the Machines in Insurance
by Steven Kaye
Robotic process automation (RPA) can be a cost-efficient, short-term solution for poor systems integration -- but there are risks.
Robotic process automation (RPA) is gaining attention in the insurance industry, but there is still a great deal of confusion around what it is and what it can do. RPA uses software and algorithms to simulate human actions in existing systems and applications through the user interface layer, rather than via APIs or web services. Automating processes ensures greater consistency and frees up resources.
RPA can be a cost-efficient, short-term solution to the problem of poor systems integration. It allows faster and less error-prone processing without the need to modify code or other resource-intensive approaches.
However, there are risks. Insurers should not rely too heavily on the “quick fix” of RPA to put off modernization and full integration indefinitely. If viewed as a long-term solution, RPA projects may increase environment and system complexity unnecessarily, as well as tie up resources better-used for more sustainable solutions. The underlying issues with legacy technology won’t disappear just because surface integration has been automated, and RPA may end up masking technical debt.
However, insurers should not let the perfect become the enemy of the good. Given the duration of life and annuity or long-tail liability policies, and a lack of appetite for multi-year core systems replacement projects on the part of many insurers, many insurers will operate using multiple core systems for some time. Balancing cost optimization with systems optimization is a familiar trade-off for insurer CIOs. As with most technology investments, the key is making an informed decision rather than accidentally relying on a short-term solution for the next 20 years—a history shared by many of today’s legacy processes.
Providers are promoting more strategic use of RPA for transformational uses, such as governance across infrastructure, as part of a cognitive computing platform seeking processes to automate, or services orchestration. But this is not how most companies are embracing it; most insurers are using RPA with realistic, tactical goals in mind, as so-called “swivel-chair automation,” addressing specific process automation and systems integration use cases.
Strategic transformation is generally reserved for core systems projects rather than “scaffolding” infrastructural projects. Just as some insurers have leveraged business process management solutions or enterprise service buses for strategic transformation, some insurers may do so with RPA. But Novarica sees this as the exception rather than the rule.
The real question that insurers need to answer isn’t tactical vs. strategic but how an RPA solution will be used. RPA solution providers vary in terms of area of focus (front-office, back-office or both), control and governance capabilities, degree of focus on RPA, degree of integration with other technologies such as AI or analytics, industry verticals, partner ecosystems, and pricing. Because RPA solutions sit on top of other systems by design, it is easy for insurers to experiment with multiple vendors.
RPA solution providers often partner with systems integrators to help clients develop strategies and implement RPA solutions. Some solution providers obtain most of their sales through reselling by their systems integration partners. Though RPA is often marketed as a branch of artificial intelligence (AI), at its core it is an application of more traditional technologies, such as screen scraping and rules engines. That being said, the more advanced providers may couple RPA with AI and related technologies, such as computer vision and machine learning, to enable ingestion of audio, images and video, to eliminate data errors stemming from unexpected changes and field-mapping drift and to speed training of RPA “bots.” Several RPA solutions take advantage of chatbot or voice interfaces.
Novarica recommends a five-step approach to RPA, though the same could be used for most emerging technologies with some modification.
- Have a defined project in mind. A project with a limited scope will give carriers the ability to evaluate both the technology and the service provider, as well as the impact on the current workforce, in a relatively short time. As with any technology project, change management is a key component.
- Determine a process for maintaining RPA rules. Like modern insurance core systems, RPA vendors often promise that business users can create, maintain and update rules themselves. The reality is that IT will likely have some role to play in terms of governance and other real-world complexities, and this should not be seen as a failure on insurers’ part. Creating, maintaining and upgrading software bots has implications for IT architecture and security, to name two areas.
- Talk to service providers already in use. At this point, almost all IT service providers have RPA strategy development and implementation offerings. A service provider that an insurer already has a good working relationship with reduces the chances of project failure. Some solution providers have partnerships with RPA solution providers, as well, leveraging their own expertise in a particular area, such as contact center systems or customer service. While there are firms specializing in RPA strategy development and deployment, it seems likely that most will be acquired by larger systems integrator firms seeking to build out their expertise in the near-term.
- Monitor the space for enhancements and new players. Both service and solution providers’ capabilities will change over time, and insurers may be able to negotiate better pricing at a minimum.
- Review the impact of RPA projects, and continue to evaluate strategic options. Insurers should understand baseline metrics for the process or processes they are seeking to automate, to document improvements, as well as any errors. While RPA may postpone the need for a full core systems transformation initiative, it will not eliminate it.
See also: Here Comes Robotic Process Automation
All this isn’t to discourage insurers from experimenting with RPA. As with any technology project, the key is to have a defined use case, proper governance and measurement of the impact.