In Conversation with: omni:us and MS Amlin
In early 2023 MS Amlin Marine NV selected omni:us to automate the insurer’s corporate marine insurance claims process. Today, MS Amlin Marine NV is using omni:us to entirely automate the claims process for low-complexity cases and assign a claims handler automatically for more complex claims. The time taken to triage complex claims and payout simple claims has been much reduced.
What are your roles at MS Amlin Marine NV and omni:us?
Daniella: I have been working for MS Amlin for almost 14 years, starting as a claims handler and progressing to Head of Claims. I lead a team of just over 100 claims handlers, processing marine claims.
Daniel: I am the Chief Revenue Officer at omni:us, responsible for making the most of our partnerships with insurers and technology companies. I also work closely with business development and sales teams. I have been working in the Insurtech space for the past three years.
What was the driver behind MS Amlin Marine NV seeking to automate its claims?
Daniella: I believe that the 80:20 rule applies to claims; around 80% of claims are low-value, high-volume claims and should be automated as claims handlers bring limited additional value to these cases. I was therefore keen to find a solution that would automate 80% of the first notification of loss (FNOL) process. This also frees up time for the claims handlers to spend more time on complex claims where they add the most value.
Additionally, one of MS Amlin’s values is continuous improvement. Our claims practice in the Netherlands is seen as a soundboard for trialling new approaches and testing the viability of my aspiration for 80% claims automation.
Why did MS Amlin Marine NV select omni:us as its partner?
Daniella: MS Amlin Marine NV offers nine different marine products, all requiring large volumes of data to process a claim, which makes it difficult to automate an already complex line of business. Any automation system needs to identify the different product, fill the relevant fields and triage to the relevant specialist handler. Having evaluated three vendors, it became clear that developing an FNOL automation product was beyond the capability of two of them.
MS Amlin’s UK hub, which is more focused on digital products, was already working with omni:us on a non-marine product. When I spoke to omni:us it was clear that they understood the complexity of the project and had the technological capability to offer a complete solution, so we went with them.
Daniel: omni:us had previously only worked in retail insurance, which is more streamlined, digital and standardised. From the beginning, we made it clear that although we had the capability, we did not have the intimate knowledge of processes and requirements. The success of this project has been made possible through excellent collaboration between us and the MS Amlin claims team.
What was the development process like?
Daniel: The whole process took around four and a half months. The first step was to gather enough relevant and diverse claims data to train the algorithm. To do this omni:us began extracting and exporting data from MS Amlin Marine NV’s claims system. This was the quickest step, taking only a few weeks.
The next step was to learn MS Amlin Marine NV’s decision-making process, which took a month. The extracted data and final claims outcomes need to be analysed to create a detailed outline of each individual decision that was made to reach an outcome. These individual decisions need to be assessed to find what can and cannot be automated and which decisions needed to be flagged to a specialist or require greater flexibility.
Once we identified decision paths and what could and could not be automated, we needed to train the algorithm. This took two months and required a lot of collaboration with the team as we produced various decision iterations and had to check with MS Amlin Marine NV to see if these versions were delivering the right outcomes.
Daniella: It is important to note that we were implementing omni:us’ solution during a remote working period. This meant we had to collaborate in a very agile way and remain responsive to their requests. Working in small teams helped as it meant high levels of empowerment to make decisions and keep up the momentum.
How did MS Amlin Marine NV integrate omni:us into its system?
Daniel: In MS Amlin Marine NV’s case it took three months to complete the integration and the decision-making to go live. The integration period will be a function of each customer’s internal IT environment and will vary in every case.
Daniella: For most insurers, there are a lot of compliance hurdles that will need to be overcome before you can unlock APIs to link a solution like omni:us. Fortunately, MS Amlin Marine NV had developed its claims management system in-house. The value of building the solution in-house meant the IT team was more comfortable in creating APIs to link the products than if the claims system had been sourced from an external provider.
What has been automated so far?
Daniella: So far, we have automated the initial triaging process once the FNOL is received. Currently, we receive claims via a phone call or email with a small number of claims coming through a separate FNOL system. omni:us’ AI extracts the relevant information from all these sources. From this, it can identify who sent the claim, whether the claim is entirely new or an amendment to an existing claim, what the relevant policy is and whether this policy period has expired. Using this information, omni:us can recommend whether a claims handler is needed and what type of specialist they appoint to handle the claim. The algorithm has also been trained to indicate how much should be paid for less complex claims worth €5,000 or less.
How is this automation speeding up the claims process?
Daniella: Automating the FNOL decision process has made the entire journey a lot simpler. The data extracted and input by omni:us is of better quality than if it was when entered by a human. As a result, claims assistants only need to key FNOL data for cases that the algorithm deems exceptional. This creates a feedback loop for the algorithm as claims handlers can further train it on how to extract data from irregular claims to improve the claims process in the future.
Was there cultural resistance to the changes or were they embraced?
Daniella: You definitely need to be considerate of those working in FNOL when implementing these changes. These people will be using the system daily, so you have to ensure that it actually improves their lives and the quality of work. I am lucky that my team is open to change and trying new things, but you need to ensure that people do not feel threatened and explain the benefits of no longer having to perform administrative tasks and having more time to spend on more interesting work.
Daniel: In my previous experiences I have seen that employees can find this type of automation intrusive. It is important to dispel this notion early on as if key employees don’t buy into the project it can go from taking a few months to up to a year as people make it difficult by withholding the data, tools and knowledge needed to train the algorithm and automate claims effectively.
What is the greatest benefit for claims handlers?
Daniella: Overall, claims handlers have greater job satisfaction. Being free of such administrative tasks means that we can upskill assistant claims handlers to manage more complex claims and further interrogate the existing claims process to make it more efficient. This makes people feel that they are adding value to the business as they are supporting the development of our overall claims strategy.
How far can you take this partnership in marine claims?
Daniella: Going back to the 80:20 rule, I have an ambitious target of automating around 85% of the FNOL process for hull and cargo insurance products. We are in the early stages of preparing for the next step in our automation journey to meet this target. That step will be using omni:us to facilitate the entire straight-through processing of simple cargo claims, only passing cases which require more attention to a claims handler. We are even brainstorming the idea of automating a suggestion as to the next step after a simple claim is passed to a handler. For example, omni:us would suggest that the claims handler should write a letter to request additional information. In the longer term, I am keen to work with omni:us to handle more complex claims, where third parties might be involved such as liability and property and indemnity cases.
Can you implement a similar solution in other specialty classes?
Ironically, by starting with our marine business, we have worked with omni:us to implement their solution for one of the most complex P&C classes at MS Amlin. I’m therefore confident that the same approach can be implemented by other teams in other P&C classes.
We intend to foster a long-term partnership between omni:us and MS Amlin. As mentioned, omni:us has an existing integration with our platform. This will make future integrations a lot easier as the system has existing APIs for relevant data points. Given the success of automation in marine insurance, there are many more opportunities for us to explore together in 2024 and 2025.