We take a look at a practical approach to identifying RPA and AI use cases for P&C insurance core processes.
Sometimes the right tool for a job is the tool you have, at least according to my father. That’s what he said when I would complain it was hard to rely solely on a pair of pliers when we inevitably needed to fix something on our family farm in rural Nebraska.
While I understand the sentiment that there is a tool for every job, not everyone can afford to buy a tool every time something comes up – so the best tool to have is one that you can use for many jobs.
Being raised on a farm infused some practicality in me. We often had to get things done with the tools we had. Usually, that consisted of a hammer and that pair of pliers I complained about. I often look back on that time in my life and draw from the forced “make do with what you have” situations it presented.
The advent of Robotic Process Automation has me reflecting on the challenges I faced during that time in my life. A few colleagues and I went through an exercise where we identified RPA use cases across the insurance value chain. In this article, we’ll cover how to identify the best use case for Robotic Process Automation in insurance core processes. You may have read our white paper, Taking a Business-Driven Approach to Continuous Improvement for Insurance Core Systems and Processes. We will fill in some of the process gaps that may linger from a continuous improvement approach to your core systems.
Why Everyone Needs a Robot
So, why is it even called “robotic” process automation? It’s not an actual robot, at least not in the way we traditionally think of a robot, right? But, when you consider the underlying concept, it may be closer than it first seems.
Let’s dive into that farm experience I mentioned. Every fall, we would build a temporary electric fence around our cornfields after harvest. The fence allowed our cattle to graze on those fields, and they could eat the corn left on the ground. I hated building fences. It was hard. I would have to drive a fence post into the ground every 20 feet or so, usually in the freezing rain.
I would constantly hammer my hands and was not very precise with my distances between posts. I would frequently daydream about having my own personal robot to do the job for me. Now that would be some robotic process automation!
Is that fantasy that far off from what we know as RPA today? Those are completely different things, right? Maybe not. Let’s consider for a moment why I wanted a robot to build my fence.
It was repetitive, required endurance and it needed to be completed quickly and during the fall and winter, when I had a lot of other things going on. But why did I dream about a robot? Why not some fence building system that was purpose-built for that task?
When put in this context, modern RPA solutions really aren’t all that different. We yearn for a convenient and practical way to have an intelligent robotic assistant perform tasks in the way we would.
Let’s apply this to a present-day example. As a consultant in the insurance industry, I have had the opportunity to see RPA in action and have developed an understanding of where we can apply it. Amazingly, the same concepts from my robot daydreams on the farm still apply.
Effective RPA Core Systems Solutions
I am going to focus the remainder of this discussion on P&C insurance core systems and how we can apply these concepts to identify and create effective RPA solutions.
We’ve started to see a few insurance use-cases show up repeatedly with RPA, such as the Certificate of Insurance process. These typically start with the highly manual, repetitive tasks that either create bottlenecks or take away from the valuable time of key personnel.
While I think this is a great start, there are more examples with the potential to mature over time, such as at the front end of both the claim and quoting processes within the insurance value chain.
If you’ve ever tried to configure claim assignment in a popular core claim system like Guidewire’s ClaimCenter or Duck Creek Claims, you know it’s not an easy task. Mainly because you don’t have all the information you need within the system to automate it, and it is difficult to validate if a claim went to the right place.
Claim adjusters rarely feel they are fairly dealt out assignments, and it causes stress for all involved. Companies often directly tie efficiency metrics around closing claims and aging claims to performance measurement. To make matters worse, high volume and additional needs elsewhere exacerbate assignment challenges. System configuration solutions tend to be a little too abstract and don’t relate as directly to existing processes and assignment rationale.
RPA may be an effective solution. Claim assignment is an intuitive action in that there are typically a set of steps a manager takes to determine to start a claim. They check current workload, look at the attributes of the claim to determine which adjusters have the ability or certifications to work it, and they likely want to determine adjuster availability.
Robotic Process Automation can get you started with simple interactions and increase in complexity to do tasks such as analyze personality matches. The goal is to automate the checks that bog down your claim manager and create bottlenecks. You can get initial assignments that the manager can tweak by checking the simple stuff with an RPA bot.
Supercharging RPA With AI and Machine Learning
So, how can you add in the sophistication and grow it from there? The use of artificial intelligence (AI) or machine learning (ML) can really give your robotic process automation an intelligence that takes it to the next level. And, it can often surpass the analysis a human may apply to the situation, often due to time constraints or implications not evident to even the most astute people.
Applying an AI or ML algorithm is somewhat complex, but very achievable with today’s modern toolsets. It is a common and logical extension to many RPA scenarios.
By doing this, determinations or recommendations for assignments revolving around claim complexity or personality matching can happen quickly. This step reduces escalations and prevents a claim from bouncing around to several adjusters before official assignment occurs.
This improvement allows employees to present their claim managers with data points and assignments, saving everyone time.
People often make assignments more complicated for the wrong reasons, causing most organizations struggle with delivering a claim to the right adjuster quickly. RPA combined with AI or ML can be a great way to deliver meaningful improvement in this area.
Like claim assignment, the triage, assignment and initial evaluation of submissions can be an equally painful exercise for Underwriting Managers. Many carriers get more submissions than they can address, and they develop bottlenecks in the triage and assignment process for submissions.
Initial activities, like checking the submission for completeness and requesting additional information from the agent can eat up valuable time for underwriters, managers or even underwriting assistants.
By applying RPA to this triage of submissions, you can start simply. Check for completeness, verify availability and possibly even include assistants (the human ones) appropriately. Once in place, AI or ML algorithms that focus on priority and potential authority needed can have significant impacts on where to focus underwriters. Underwriting managers can focus more on directing decisions vs. submission assignments.
Further sophistication can improve the involvement of marketing or risk control appropriately. Like the claim assignment example, these steps can be intuitive when considered in the context of the existing process, but unmanageable if you try to abstract it as part of your policy administration system configuration.
Both the claim and submission assignment examples above highlight the versatility of an RPA solution. From a process perspective, they have a lot in common, but if you tried to implement a purpose-built technical solution, each would look very different. You would add complexity and invest a lot of capital in a targeted and rigid solution.
An RPA solution is a versatile approach that closely aligns with steps and processes already in place or at least envisioned and desired. You can get reuse out of similar process and actions, even if they are with different systems and teams.
The adoption of Robotic Process Automation (and subsequently, AI and Machine Learning) is becoming more and more prevalent for carriers. Simple, repeatable use cases are a great place to start. Be somewhat literal in your interpretation of the term “robot” when you consider where to apply it.
Core systems should be a key focus area during these considerations. They are wrought with enhancement requests that sit in IT backlogs, untouched. It’s okay to daydream a little and ask yourself “what would I do if I had a robot? What would I have it do for me?” You might be surprised how achievable that daydream can be.