Our initial idea was to look into the construction industry because we assumed that it was probably lagging behind in technological development. Although none of us had real experience in this field, we found it interesting to explore. Based on our research, we identified multiple issues that construction companies face: coordination among suppliers and contractors, contract and project risk evaluation, and competition for contracts. Our next step was to determine where an AI or IT system could improve these processes. This brought us to the final idea, which was kind of a mixture of these issues. Improving the IT project system with an AI agent for procurement evaluation would help estimate the costs and risks of an ongoing project, which would give contractors a competitive edge. All of this would be part of a system we would manage, where we could connect contractors and suppliers.
After we all agreed on this solution and project, we had to decide which path to choose. At first, Track A seemed a bit more appealing because we would improve the already existing system, but we ditched this idea for multiple reasons: it would have been impossible to gain insight into a real contractor’s system and collect sufficient data; our team also had a bit of an entrepreneurial approach to this assignment. We decided that it would be better to pick the venture path because it offered more creative freedom and because, as we progressed through the lectures, applying the course concepts required a slightly more innovative approach. This would be simpler to do with a newly built venture. Up to this point, we had all put our heads together, brainstormed, and tailored the idea until we had a solid, detailed proposal.
The next steps for this assignment required us to divide the tasks. I stepped up to divide the tasks among the group because I already had an interesting proposal for the prototype. Luckily, all of us had preferences regarding the tasks, so dividing them was pretty natural. The hardest task of the assignment was coordinating among the members because each task could be done individually, but the overall project needed to be coherent and aligned. This brought up multiple issues during the individual tasks. We had to revisit our solution multiple times because, week after week and lecture after lecture, we understood more of the concepts, and it became harder and harder to “fit” our idea into the course concepts. Communication, patience, and trust helped the team to build this project. After our team got picked as an example of a strong idea in the lecture, all of us were more motivated to put the most effort into this project. This public feedback about our main assumption helped me refine my initial idea of the prototype. It shifted my focus from the features and interface of the agentic AI to testing the assumptions that could undermine our entire competitive advantage: whether our feedback loop would work and whether the dataset was useful at all.
My contribution to the group project was building the prototype interface and the initial logic behind it to test our assumption. Furthermore, I built the final slide deck. The reason why I took over these tasks was that I had the most experience with building something using AI. For the slide deck, we decided to have one person put it together to keep the storytelling coherent and easy to follow, and to maintain the same writing and design style. Because I was already deeply involved in the prototype work (and because I truly enjoyed the process), I took over the task of putting the slides together and creating a story for our project journey.
Although my task didn’t really involve much application of the course concepts because it had to be quite practical, I have to emphasize the influence of what I learned during the course. In designing the prototype, the lessons really shaped my way of thinking, teaching me to always think about AI as a tool rather than the “secret formula” for all the issues constructors face. For example, to build something that can actually bring a sustainable competitive advantage, I had to think beyond the technology and start thinking more strategically. This shaped the idea of creating a feedback loop for our solution that would make our venture hard to copy.
The hardest challenge for me was to come up with a prototype that would be useful for demonstrating our strategic advantage (and testing our riskiest assumption). The process involved a lot of brainstorming with Claude and testing multiple approaches. The first approach was the one the IS team suggested: to pick one case and run it through our system, changing one variable at a time. Unfortunately, I didn’t manage to design a system that would effectively test our assumption, mainly because the test would have been circular: I would have been building the prototype in a way that proved our assumption by design. After a couple of failed attempts, I managed to come up with a solution that could both test our assumption and showcase, in a simple way, how our agent would actually work in real life. Once I had the logic clear, designing the interface was the easiest part of the process. Unfortunately, another problem arose: How could I explain the thought process and methodology behind building the agent in a simple, easy way? I always struggled to put my thoughts into clear words, which is why I was happy to take on the challenge of putting together the slides and explaining our journey in an understandable way.
Another challenge that I really enjoyed overcoming was the platform test for our venture. Platforms were introduced in Session 5, and we realized that our idea would be stronger if we were a platform. However, the progress we had made by that point made this really hard, so we had to come up with something. During the team brainstorming meeting, we managed to solve the issue fairly simply (what I wouldn’t be able to do alone) by creating a future scenario in which we could potentially become a platform. Essentially, this meant that we could use the course concepts and also retain the progress we had already made. This type of problem-solving made me realize that working in a team really helps us because of the variety of perspectives and the complementary qualities we share.
In conclusion, I deeply enjoyed every part of the process of creating this assignment. Collaborating with people who had completely different personalities, giving and receiving feedback, and using the actual concepts and theories we discussed in the lecture were the highlights of my experience during this assignment.
