From an Idea to a Prototype: my Role in Creating a Tangible Product from Multiple Ideas (Jonah van Kempen – Team 21)

Our Project
Over the past few weeks, my team and I developed an AI extension to the Adidas platform, which allows users to get a personal size recommendation on items based on their personal characteristics, as well as a database from Adidas including information about previous returns, reviews and ordering history. Together, this information allows the tool to give a calculated size recommendation, aiming to decrease the amount of clothing returns, effectively decreasing the associated costs for the retailer and the hassle of the return procedure for the client. Within our team, I was responsible for formulating the problem and opportunity, creating the prototype, and finally putting together the slides to convey the message of our team.


Problem and Opportunity
After collectively agreeing on the scope of our project, it was my task to put our ideas into one clear statement, which showed the existing problem and how we aimed to solve it. Although it seemed easy to write down our idea, research the figures behind it, and summarise it, the truth was far from that. As we all worked on the project, ideas kept developing, and opportunities to solve the problem were constantly transforming, making it challenging to accurately define our problem within the concise space we had. I therefore decided that it was best to write down the general direction of our project to keep the team aligned, but only make it specific at the very last moment, when all other parts were finished. This way, I was sure the problem and opportunity statement actually matched the rest of our presentation, rather than giving a general description of the problem. Feedback on the introduction kept evolving the opportunity statement, increasing its quality with every revision. In hindsight, I believe this was the best way to tackle this part of the project, as the introduction depends on everything that follows it.


Creation of Prototype
Soon after coming up with the preferred solution to our problem, I realised I was most keen on working out the prototype, as I have experience in coding and like to see our solution come together. However, this process took a whole different turn, as I was very surprised by the quality of Claude Code, which changed the entire way of creating the prototype. Instead of coding it myself, I realised Claude Code is much quicker, better, and more adaptable than my own coding skills, but more attention was needed on checking the output, testing the prototype and the creative design choices. Thus, I decided to vibe-code our project using Claude Code, putting much more emphasis on the design and functionality of the prototype rather than the code behind it. However, this process took multiple attempts, as the first few prototypes failed and were of insufficient quality to put into our presentation. Rather than teaching me to code, this process taught me a lot about the structure of such an application, as I had to critically think about where to place functionalities, how to design them, and what to do if the system fails. This changed my understanding of how to use AI in business, as I realised it is much more important to think about the reasoning and requirements behind an online tool than to put time into actually coding it, as this is easily done with the support of AI.


Finalising the slides
After my team and I finished the document in which we all wrote down our findings and suggestions for the application, it was time to put it all together into a presentation, which had to explain our findings in detail, within limited space. After using Claude Code to help create the application, I was curious how far we were able to push it, and I tried using Claude Design to help create the PowerPoint, with great success. It was able to help me structure the slides in a way I could not think of myself, allowing for much more information to be presented on a slide, without overstimulating the reader with too much text. However, it also created challenges, as the AI was often thinking differently from my own thoughts, resulting in the need for very accurate prompts about the structure, design and content. It was quite challenging to arrive at the final presentation that we submitted, as all five team members had their own expectations of the final results, while it was my individual task to make all these different expectations come true.


Final Thoughts
This project has been very challenging but interesting at the same time. As my main contributions to the project included both the initial and final parts of it, I was very involved in trying to tie everyone’s input together to end with the best possible final product. Throughout meetings, feedback sessions and individual talks, I came to get a better understanding of how to bundle all inputs into one application. Throughout much trial and error, many alterations to the prototype, and many failed attempts to make the PowerPoint, I learned to effectively use AI as a support for creating the deliverables, as well as to keep the project’s scope in one direction that we all agreed on.