Jens Poelman – 780820jp
Starting point
Before we started the project, I didn’t really know what to expect from a project about information strategy. I had a basic knowledge of how AI could be used, but I didn’t really know what the frameworks are that exist. When we talked about what subject we should choose to create an AI agent for, I found it difficult because there are a lot of problems and directions you can go to. So then we considered which market we could create value through an AI agent, and which problem we needed to solve. And I think that the guideline of the short project proposal helped us in a way to understand what and where we need to look for and think of.
Contribution
It was my job to address the problem and then what value our solution would add for the insurers. To define the problem and opportunities I used the “Capability -> Workflow -> Value ” model from the third lecture, which links AI capabilities to workflow changes and business value. So with Claim AI we want to automate initial damage assessment to process claims faster and reduce administrative costs. For the disruption that I needed to look into I used the information of the first lecture where I analyzed Porter’s Five Forces model. I picked the two changing forces. The first was the threat of entry, where the accessible AI models lower the technical barrier for specialist claims software providers. The second changing force is the competitive rivalry, where new AI solutions challenge the traditional way of manual claims handling on speed and cost. After looking into Porter’s Five Forces, I applied the newly vulnerable market model, that shows us why the routine vehicle claims market is interesting for our new venture. I dove into the position of Claim AI in the ecosystem, where we position ourselves between the policyholder and the insurer. We are a B2B software layer and not a two-sided marketplace. With those models I think the most important learning for me was, that it is not only about the AI model itself, but that AI creates value by changing a workflow, by making a certain process more efficient for example.
An important change
In the beginning we started with a slightly broader scope, where I thought the model would not only analyze the photos of the motor damage and the description. And if these 2 points of information match each other, will this analysis also help to prevent fraud. But I thought as well that it should look into the policyholder’s insurance documents and then combine this with the results of the analysis of the damaged vehicle and the description. But after creating the model I saw that the part of looking into the policyholder’s insurance documents using AI and from that point determine the payout was not included. This was also a bit harder to do because of concerns about privacy laws. We discussed it with the group and from that point we changed the direction a bit, towards an agent that analyzes the damage and checks for possible fraud. The code settles claims under €1500 only if all the checks pass, otherwise, an adjuster handles the claim.
What surprised me
We had a misunderstanding about what our AI model should really do and therefore I needed to change my expectations. Because the idea about our AI model was a bit different from the reality. Therefore I talked about this issue with my teammates. First to check whether I understood it correctly or not, and afterwards look how to deal with the change for my parts that I needed to do. And I think that was a good way of doing, cause if as a team we deal with a change, we need to make sure we understood the problem correctly. Then discuss with each other about the alternative to change the expectations and understand the problem correctly.
Collaboration and Feedback
Claim AI could create value for both insurers and policyholders. After the adjustment that we made during the project, I needed to change the definition of the problem a bit. The position of our new venture within the ecosystem remained the same. Next time I would be a bit more clear about the direction we want to go to. At the end of our meeting I summarized the solution I thought we wanted to create, so everyone would have the same understanding. In the end this still was not enough. Next time we need to really understand each other and if something changes that we need to communicate to the whole team. And if I need to communicate in multiple languages I will ask it in both languages so everybody will work towards the same goal.
What I learned about AI and Strategy
Before starting this project my idea about AI was that it was useful to get information or ask for help for certain problems that you face. But that AI actually can be used to create value in a lot of different ways by changing a certain workflow, I didn’t know. Claim AI could create value by reducing human labor, what will give the insurers more capacity for the more complex claims. I also realized that a prototype will show how the end product could look like, what will make it more visual for the client. The client has a better idea what he can expect and so it will be easier to understand the developer. I think that this is a big lesson that I learned, cause if I had maintained regular contact with the developer of Claim AI I would’ve known that the direction changed. When in the future I have a project like this I will define the expectations more clearly and check throughout the project if our work still aligns with the shared goals.
