The AI Strategy Lab was a new experience for me. Coming from another faculty, I had not previously encountered a course project that developed gradually each week. This made it a valuable opportunity to experience a way of working that felt closer to a professional project than a traditional university assignment.
The first challenge I encountered was identifying a real-life workflow problem that emerging technologies could help solve. Although I have a strong background in economics, I had little experience with developing business models or translating an idea into a business plan. I therefore began by thinking about problems I had personally experienced when interacting with businesses and service providers. I remembered the times I had visited a physiotherapist and been sent home with exercises that I then performed incorrectly. As a result, I had to return to the physiotherapist to practise them with guidance. This experience formed the basis of the information asymmetry and lack of guidance during home rehabilitation that our group eventually aimed to address. Starting from a personal experience helped me understand the problem and explain why it mattered to both patients and physiotherapists.
The lecture on agentic AI helped me think more clearly about how AI could contribute to our proposed solution. We began holding online meetings to develop the project together. These meetings were very useful for discussing how the system should work and which functions the application should provide. Explaining our ideas to one another helped us organise our thoughts and identify where our ideas differed. This made sure that we were working towards the same goal, which was important because we completed some parts of the project individually between meetings.
My main contribution was developing the first mock-up. This was challenging because I had a clear picture of the application in my mind, but translating that picture into a detailed design required a lot of precision. There were many interconnected functions and small details that could easily be overlooked. For me, the hardest part of the project was describing the application clearly enough to ensure that the design accurately showed what we had discussed as a group. The first mock-up generated from my prompt was too simplistic and did not include all the functionalities we had thought of. For example, it did not show where the patient could start the camera to start the AI-supported movement assessment, even though this was a central part of the application. I realised that I had not described the requirements clearly enough. Something that seemed obvious to me was not necessarily obvious from the instructions I had given the AI. I revised the prompts to describe the functions and the connections between screens more explicitly. This involved repeatedly reviewing the output, identifying what was missing or incorrect, and adjusting my instructions. With each adjustment, the mock-up came closer to what I had in mind. This experience taught me that having a clear idea is not the same as being able to communicate it effectively. I had to write down all my assumptions instead of expecting the AI to guess them correctly.
Feedback from my teammates also shaped the design. During our meetings, we continually discussed how the application should work and what patients and physiotherapists would need from it. These discussions helped me refine the requirements and translate our shared ideas into more precise instructions. The mock-up therefore developed through both my own revisions and the group’s feedback. The resulting mock-up included both a patient interface and a physiotherapist interface, with a clear layout showing the application’s intended functions. A teammate then used it as the basis for developing the working prototype. We worked closely together: I helped specify the required functionality, while she focused on building the prototype. I appreciated how this collaboration allowed us to use our different strengths. Visualising a solution and providing clear instructions and plans suited me, while my teammate was stronger at implementing the prototype. Together, we translated the ideas from our meetings into something tangible. This showed me that an effective contribution does not necessarily mean doing every part of a project yourself. Providing a clear design and communicating requirements can be just as important as building the final product.
Another part of my contribution was thinking about trust, privacy, and governance. I expected these topics to be challenging, but they required more detailed consideration than I thought at first. Because our proposed application processes sensitive patient information, we had to consider ethical concerns, data protection, and the reliability of AI-generated information. I realised that the application’s usefulness would depend on whether patients and physiotherapists trusted it. Building that trust required a lot of discussion about how the system should handle personal information, communicate its limitations, and involve the physiotherapist. Human oversight became particularly important in our design because AI-generated information could support professional judgement, but treatment decisions would still require the physiotherapist’s involvement. This changed my view of governance: it needed to shape the application and business plan from the beginning.
The project also changed my understanding of the relationship between AI and business strategy. I learned to start with a problem and the opportunity arising from it, rather than with an interesting technology that I wanted to implement. The starting point should be what needs to improve in a workflow. Only then should the team consider which technology best supports the solution. Our physiotherapy idea made this lesson concrete because the value of AI depended on whether it could address the lack of information and guidance between appointments.
Overall, I am pleased with what we developed and how we collaborated. In a future project, I would organise more frequent meetings. Our busy schedules sometimes limited how often we could meet, although we used the meetings we did have to discuss the project thoroughly. I would also like to be more involved in the economic aspects of the business plan. Although I contributed to these discussions, I did not work directly on that part of the project. Given my background in economics, I could have contributed more in this area. The project helped me recognise the strengths I used effectively, the expertise I could apply more actively, and the importance of connecting a clear business problem with an appropriate technology and responsible implementation.
