QuizAidBot

Inspiration

We were inspired by traditional education apps such as Quizlet. While flashcards do help students retain short blocks of information, typically before an exam, there may still be gaps in students' understanding of a particular concept within a subject. As a result, we aimed to build a chatbot that would generate quiz questions relating to the inputted academic subject from the user, and then provide feedback to the response.

How we built it

We used a Vercel Template that aided with the front end for our chatbot. Then, we updated all references with OpenAI to implement a Watson Assistant from IBM Watson. Since the Vercel AI SDK does not integrate IBM Watson, we attempted to do so by updating our routes.

Challenges we ran into

Using the Watson Assistant on the Lite version was difficult for us, as most of the Task customization appears to be a paid service. As a result, a lot of our intended solutions were not viable under the Lite version. Furthermore, there was a learning curve for our team, as we have never used WatsonAI, nor TypeScript+NodeJS as a tech stack.

What we learned

We learned about new technologies in the AI space and attempted to integrate them within our project. We learned a bit more about OAuth through our updates of routes to and from Github. We also explored the use of Axios and Hoppscotch to send requests and test our API.

What's next for QuizAidBot

We think that transitioning to OpenAI would be wiser (please give us free credits), since that would allow for easier customization of our chatbot. In doing so, we would have also liked to explore Modal Labs' capabilities with generative AI. In the future, we would aim to delve more into the ibm-watson package from npm, as we did not have enough time to fully explore and understand the codebase. As a result, there may be some inaccuracies and inefficiencies within our codebase. We also have an issue with a declined pre-receive hook for a recent push