Looking for pdf to text api. Gonna use 3-8 pages per api call on avg. Quality and speed of setup more important than price. Any good or bad experiences to share, or recommendations? #webdev #api #developer #buildinpublic - ThreadSky

michielberk.com • 85 days ago

Looking for pdf to text api.

Gonna use 3-8 pages per api call on avg. Quality and speed of setup more important than price.

Any good or bad experiences to share, or recommendations?

#webdev #api #developer #buildinpublic

Comments

biptopposts.bsky.social•84 days ago

Hey @berksky.bsky.social👋,

This post was a runner-up for a top #buildinpublic posts for the past 24hrs. You almost got most replied (4)!

See who had the top posts on our feed @biptopposts.bsky.social. Keep up the great work!🎉

dontcookidontclean.bsky.social•84 days ago

https://gotenberg.dev/

michielberk.com•84 days ago

hey thanks!

krasny.co•85 days ago

Maybe take a look at Eden AI. At least for quick comparisons of outputs.

tony-m-meijer.bsky.social•85 days ago

Use python, it has many good libs for this. I can show you tomorrow. ;)

sabihsarowar.bsky.social•83 days ago

Gpt4o worked great for me

rafal.fyi•85 days ago

I was planning to add one to Supadata based on what I’ve made for Chatwith…

just an exploratory question, but how would you expect to push the PDF to the API? upload it via multipart, url, something obscured via an SDK?

michielberk.com•85 days ago

Word of caution though: I am exploring options and it looks like this is a REALLY crowded space. And I suspect half of them are just layers on top of Google Cloud Vision api which is dirt cheap. I think its only worth it if you can upsell it to existing customers.

rafal.fyi•85 days ago

thanks for the info - mind sharing some links to competitors? :)

michielberk.com•85 days ago

Currently looking into;
-pdf.co
-Cloud Vision api
-pdfcrowd
-Convertapi

rafal.fyi•85 days ago

And what are these solutions lacking?

michielberk.com•85 days ago

Cloud Vision: needs to use google cloud for storage, cant send url
Convertapi: fixed pricing on top of pay per use which I think is dumb. Havent checked it out yet (last on list).
https://pdf.co did not get it right, but did extract text. With an additional open ai prompt I can make it accurate.

rafal.fyi•85 days ago

and yeah, my product is ofc a convenience wrapper for people who don’t have the time/skills to figure out how to make it all work

michielberk.com•85 days ago

Btw, I am processing image based pdf's. A lot of services that had a free no registration demo could not handle these either.

michielberk.com•85 days ago

Right now I am extracting text on my own server but am struggling with img based pdfs (the ones where you cant select text). Currently I have the source docs so sending url is probably easiest/fastest.

rafal.fyi•85 days ago

to explain the URL, say your customer uploaded to your S3, you could just share a signed URL to that asset

rafal.fyi•85 days ago

feel free to DM me if that easier :)

Comments

Posting Rules

Reply