Our teamwork began with a kick-off meeting, where we brainstormed project ideas. While there were several promising ideas, we particularly liked one problem which focused on the difficulty migrants and other Non-Dutch speakers have in navigating the Dutch healthcare system, particularly the post-consultation steps they need to take regarding physician instructions such as their medication intake. After dividing tasks to work on the project, I decided to work on the platform and ecosystem analysis, the digital markets and the agentic future of our idea. I decided to focus on these areas as I had previously done an internship where I had to conduct various market analyses and thus wanted to apply my knowledge to this new case.
Furthermore, during the later stages of the group project I took on the informal role of project coordinator, as our team was struggling to coherently synthesise our progress and knowledge we had developed throughout the project for the final slide deck. For this, I created the outline of the report by synthesising all relevant course materials with our project. During that work I learned that the development of such novel ideas encompasses not only the technology and the difficulties it brings itself, but also the coordination of various ideas from multiple people to achieve the best outcome.
The most important decision was taken at one of the final meetings, which focused on the software’s functionality. It was not easy to agree on the exact scope of the app’s functionality, as it became apparent throughout the project that each team member had slightly different ideas about data input mechanisms, workflows and the general user experience. For example, whilst the final report was being drafted, a meeting was held to discuss the possible integration of health insurers as an additional party. There was a discussion as to whether their integration should be presented as a feature that had already been “implemented” as part of the project, or as an integration planned for the future product roadmap. With the feedback from the TA Wei Wang, it was decided that such an expansion would not be beneficial as it would go beyond the scope of this project and could thus lead to a confusing product architecture. Thus, it was decided to include the integration with health insurers as a future step needed for the development and integration of CareClear. This iteration taught me how to prioritise the most important features within extremely limited time and resource constraints.
I was particularly surprised by the level of detail required to actually develop an AI business idea. Due to my background in business administration, it was relatively straightforward to formulate a general concept for the app. However, translating this concept into a technical framework and understanding how exactly AI is increasing value in this case proved to be surprisingly complex. This became clear, for example, when a decision had to be made on how patient data should be integrated into the system, as the chosen method would have completely altered the underlying business model. There were two options, with either patients actively recording their doctor’s appointments in the app, or the tool operating passively in the background and retrieving existing electronic health records and patient data. While first option required more active work from the patient and thus a higher likelihood of errors occurring, the integration with electronic health records would necessitate complex collaboration on the server side with existing providers of IT solutions for the healthcare sector. While we decided to go with the integration of electronic health records, this example illustrates the difficulty I had in thinking through such technical issues. As I mostly had to develop business strategies in previous projects, the new technical and IT considerations were rather new to me. However, this would also be my greatest source of learning as I have found thinking through such integration to be extremely interesting, and the application in this project has allowed me to deeply develop my skills in this area.
During one meeting, discussing an enormous number of edge cases led to a bloated array of potential features. By this point, the team had already moved on from brainstorming basic ideas to discussing extremely specific details, such as the exact design of the user interface and the definition of specific roles for various external medical institutions.
Subsequently, feedback from our TA fundamentally altered the group’s approach. When asked about the complex scheme governing how these institutions interacted with one another and to what scope to include them, we were told to rather reduce the scope and complexity of the final solution. This realistic assessment forced a change of course. We then refocused on the fundamentals of the value proposition, by asking ourselves what the main problem was that we were solving. Through that, we were able to eliminate unnecessary complications, with the key functions then being successfully discussed and developed, and these clear and optimised workflows were ultimately implemented in a functional and fully developed HTML file as our MVP.
On a personal level, I learned how important active listening is in such group projects. During meetings where we were developing the core workflow, I found it to be extremely enriching to step back and listened attentively. Through that, we were able to develop many small and innovative details while building on top of each other’s ideas. Such details often determined the final approach to the app’s development. For example, we were able to add a separate field for explanations of difficult medical terms with which even native Dutch speakers could have trouble, thus expanding the potential user base to include not only Non-Dutch speaking patients but rather potentially even all patients.
This project has fundamentally changed my understanding of the relationship between AI, data management and business strategy. While I used to think that the integration of AI into a business was rather straight-forward, I have found out that when putting an AI application at the core of your business idea it is extremely difficult to integrate it with all of the required stakeholders. Specifically issues such as data privacy in the healthcare sector have made the conceptualisation of this idea more difficult than a standard business plan. As can be seen from the text above, one of the main concerns throughout the project was the issue of technical feasibility. From a purely business strategy perspective, it became clear just how detailed and meticulous most technical workflows need to be to ensure that all stakeholders would be on the same page.
Ultimately, I learned that while having a good business idea is crucial, integrating AI at its core creates complex interplay among various parties. Issues such as data privacy and ownership of decision-making are increasingly complex and are likely to become even more difficult with the emergence of agentic AI. I am very thankful that this project has allowed me to develop the skills required to navigate this volatile and complex environment.
