Turbocharge your writing and content marketing efforts with Unsummary, an AI-powered tool that generates quick summaries of 40m+ books, 630k+ movies, 230k+ TV shows, 4m+ podcasts, and 1.2m+ people.
If you've ever had to trawl through a million books, movies, podcasts, or people to write a snappy summaries for an article or a report, you know it's a total time-eater. Plus, it can expensive if you're running a company that pays people to write these summaries.
That's where Unsummary comes in.
Think of it like an AI-powered sidekick that does the heavy-lifting for you. It dives into over 40 million books, 630k+ movies and TV shows, and millions of podcasts, then pops out with a short-and-sweet summary. All in a blink!
Why?
I wanted to build Unsummary to give people more time for the fun stuff - getting creative with their work - and less time on the grunt work of providing the intro summaries of books, movies, podcasts etc.
A handful of my own freelance clients have been beta-testing Unsummary and they're using it to help speed up production of their blog posts, book reviews, podcast recommendation articles, newsletters etc.
I hope Unsummary can be a time-saving, money-saving, headache-saving solution for more busy writers.
Thanks, Pete.
Hi @graeme_fulton
Thanks very much for your kind comments.
Tbh, I'd never even considered an API. But I will certainly have a good think about how that might work.
Thanks again, Pete.
@peterbrady Yeah I feel like a lot of products want to summarize large content, but don't want to build this part. Also, did you see any limitation with the AI, for example is it allowed to summarize news and political articles?
And can it handle different file types such as PDFs? (that one is because my friend just had to summarize a bunch of PDFs and we didn't find a good tool for it)
Hi @graeme_fulton
Re. news and political articles:
I did play with adding a new summarisation type: web page. That would probably take care of online articles, blog posts etc. The only problem is the token limit of Open AI. I never know how much text is going to be on the actual webpage. So a summary could be generated for only half an article.
There are ways of dealing with this. Chunking the document and then having the AI API summarise the summaries. Langchain and Llama Index can deal with this and so it's is certainly something I will be working on.
P.S I never hit any limitation issues with political content.
Re. PDF's
Same problem, really. I never know how big the PDF will be so there's a speed and cost consideration.
Btw, did you friend try https://pdf.ai/? I never tried it myself but it does summaries of PDF's, I think.
Just as a point of interest, I played with summarising YouTube video transcripts using the chunking method but for a long video with a lot of talking, the AI summary got quite slow and quite expensive. Again, it's certainly something I'll be looking more into. Just need to figure out the costing model.
I wanted to get this first version of Unsummary out with the basics to see if anybody actually cares about a tool like this. Now that I have some great feedback - thanks :) - I can start figuring out next steps.
Thanks again, Graeme. I really appreciate it.
Pete.
Just on that, @graeme_fulton
You could copy the text of an online article or a political article and paste it into the text summariser at https://unsummary.com/text to see how it behaves.
Thanks, Pete.
Hi Chris @shortcipher
Thanks for this.
I do store the summaries with the view of surfacing them on the user's dashboard for later. My thinking there is that once somebody has created a summary and used a credit, they should be able to retrieve that summary. That's next on my list.
Re. revising and improving summaries: hmm... maybe an additional prompt beneath the summary? So users can say "make this less formal", "make it shorter", maybe even fine-tuning a summary for a particular audience?
Is that where you were coming from?
Thanks, Pete.
Great service. But it's a pity that youtube videos are missing. youtube videos come with subtitle data, so I guess you can just summarize them by referring to them, right? (Those without subtitle data are not covered and return an error.
Unsummary
Prototypr
Unsummary
Prototypr
Unsummary
Unsummary
Unsummary
Unsummary