About the Project: AskTube

Inspiration

We wanted a faster way to learn from YouTube videos without spending hours watching them. The need to save time while still getting valuable insights led us to create AskTube.

What it does

AskTube quickly summarizes YouTube videos and allows users to ask questions about the content. It includes direct links to video timestamps for easy navigation and deeper learning.

How I built it

We used OpenAI GPT4 and LLAMA3 for generating summaries. For the chat feature, we implemented a Retrieval-Augmented Generation (RAG) model to provide relevant context to the language model. The extension was designed with a focus on speed and simplicity.

Challenges I ran into

Balancing accuracy and speed in summarizing long videos was a major challenge. Ensuring the Q&A feature provided reliable answers and worked smoothly with YouTube's interface was also complex.

Accomplishments that I'm proud of

I'm proud that AskTube is a tool my friends and I use daily. It has proven to be helpful and time-saving for everyone who tries it.

What I learned

I learned a lot about crafting effective prompts for summaries and various summary techniques using LLMs and RAG. I also gained experience in developing Chrome Extensions and working with Next.js applications.

What's next for AskTube

We plan to add more features, including generating different types of content, supporting additional languages, and integrating with other video platforms.

Built With

• gpt4
• groq
• llama3
• nextjs
• prisma
• react
• tailwindcss