RecapAI

Landing Page
Loaders
Summarize and Quiz example

Inspiration

The idea of creating a Generative AI tool goes back a few months ago where we had all gathered for GenAI Buildathon conducted in Bangalore, India. This got us engaged into the field of GenAI. We as students have had difficulties in following and keeping up with video lectures, thus we thought of coming up with a solution to this.

What it does

RecapAI currently takes in English YT videos having captions enabled and generates a transcript along with timestamp based key points.
It also enables the user to summarize the entire video, quiz them about the video and ask questions about the video.

How we built it

We made use of OpenAI's GPT-3 Turbo API Key for all Generative routes.
We used the youtube-transcript-api pip package to transcribe the YouTube video.
It took a lot of team effort to take this project to a basic production level.

Challenges we ran into

All of us were unaware of Docker's functionalities and usages. We took some time out to explore the tool.
The CI/CD development stages for the entire project resulted in a lot of anxious hours.
We did run into initial merge conflicts on GitHub but eventually resolved those by easing out the process by creating forks and multiple branches.

Accomplishments that we're proud of

Completing a base version of the project just the way we thought of.
Successfully implementing Docker technologies for hosting it out.
Achieving the end goal for this hackathon together as a team!

What we learned

Docker images vs containers.
Pushing images onto DockerHub and using its link for production deployment.

What's next for RecapAI

Next up, we look forward to allowing users to upload any file they want, not restricted to YouTube.
We would also work on enhancing the design to smoothen the UI/UX experience.
Adding on OpenAI Whisper for better transcriptions and timestamps.
Open to public suggestions and contributions!