We started the project with a brainstorming phase to come up with realistic use cases, which can be done in the given time and with regard of our skills. There were multiple different ideas like a chat assistant and including the OpenAI speech recognition Whisper. Finally we decided on 3 use cases to explore: curated shopping list, product comparison and a product summary.
The first week was focused on onboarding, reading OpenAI documentation and prompt engineering to test feasibility of our selected use cases. To be able to interact with the Idealo site without risking to break anything or cause security problems, we decided to build a Chrome Extension. After creating the first prototype of the product summary and the product comparison it was clear that these use cases are realistic, but need refinement. From there our tech stack grew to include NPM, React and Webpack to get a more dynamic and interactive experience. With this a good foundation was being built to get additional functionality on the Idealo page, that we used to refine the look and feel of the extension.
Our team worked in an agile team structure and our uses cases were split into smaller tasks which were worked on individually. The project was implemented with close collaboration with our supervisor Jakob Panten, who helped us with technical questions and gave us feedback on our progress and direction.