I built SumThatUp (https://sumthatup.com/) to generate summaries, key takeaways, and mind maps from YouTube videos. No sign-up required.
Why I Made This
I often get YouTube links from friends, but watching long videos isn’t always practical. I wanted a way to quickly grasp the main points without sitting through the entire thing.
How It Works
SumThatUp pulls the video transcript and generates a summary from the text. This has two advantages:
• Efficiency – Processing text requires fewer tokens than analyzing video directly.
• Speed – Summarizing transcripts is faster than processing raw video.
Limitations
• It only works with videos that have transcripts.
• It can’t capture purely visual information.
• Extracting transcripts from very long videos can be a challenge.
I also tested Gemini’s YouTube summarizer for videos without transcripts, but it ran into token limits—a 15-minute video needed over 1 million tokens.
Open Questions
• What are other good approaches for summarizing videos?
• How to handle videos without transcripts?
• How could the mind map visualization be improved?
Next Steps
I plan to add PDF and text exports soon. Feedback is welcome!
Comments URL: https://news.ycombinator.com/item?id=43360320
Points: 1
# Comments: 0
Leave a Reply
You must be logged in to post a comment.