vidsnap.ai

Inspiration

I had to catch up on Math lectures, and I tried watching hour-long youtube videos that, in most cases, didn't answer all the questions I had. So I thought of a platform that would allow me to ask any questions I have from a video. And here we are, with the help of vidsnap.ai, I can not only have all my questions answered, but also test my overall understanding of the concepts in the video using the Quiz Me feature of the platform.

What it does

vidsnap.ai takes a YouTube video link and retrieves the transcripts of the video, which it uses to prompt a ChatGPT API to answer any questions that come up based on the transcript of the video. So the user types in the link of the YouTube video in interest, and hits submit. And the transcript of that video will be retrieved by using an API, and be given to the Chat GPT as complementary information. Then, can Ask a question from the video by writing in the question on the input field on the chat box on the right hand of the window. The user can ask to be quizzed by clicking the "Quiz me!" button. There is also another option the "Quiz Me! (MC)" which quizzes the user in a multiple-choice format. The interaction between the user and the bot is shown inside the chat box, and the user can always go back and see what the previous questions and responses were.

How we built it

vidsnap.ai was built using Next.js, Tailwind CSS, and ChatGPT API. We also used an API that retrieves the transcript of a YouTube video. So by taking that transcript, the ChatGPT API was used to answer questions and quiz the user based on the contents of the video. We used Next.js and TailwindCSS for the overall functioning and UI of the platform.

Challenges we ran into

Some of the challenges we ran into include trying to get the transcript of any YouTube video. So that included any YouTube video that users put in the YouTube Link submission. We were able to do that after we found an API that does that. But we didn't quite make it work for longer YouTube videos of length 20 minutes or higher. We also ran into some issues with pulling and merging requests. We were able to finish and deploy the project on GitHub. We also initially planned on having a database. But we were not able to fully implement it.

Accomplishments that we're proud of

We're really proud of making an accessible platform for students to ask and have their answers quickly and a more efficient way of studying. We're also proud of how the UI turned out to be: very clean and straightforward. The app works for any kind of video with a length of around 20 minutes or less, so that's something we're proud of too.

What we learned

We got to work on something that involved git hub. We learned a lot about git repositories, pull requests, merging, and more from GitHub. Getting to work on an Open Source was really helpful. We also learned more about UI design.

What's next for vidsnap.ai

We hope to make a database that would store users' videos, notes, questions, and summaries. We also plan to make the app work for longer videos of length 20 minutes or higher. We also hope to deploy it on the web, like Vercel or GitHub.