Stirling-pdf. You can self-host it.

using this myself. its pretty good.

Only last week there was a HN thread about how the author said they just used chatgpt to make the entire thing and as a result the code is beyond bad. I don't think I'd trust it.

They didn't say that. They said they wrote the first version in a few days, using ChatGPT. Then worked on it almost another year since then. Something of that nature. Pretty big difference.

>I don't think I'd trust it.

You can audit the code yourself then. What's stopping you?

Am not a Coder :-P

That's a fair point lol, my bad.

Nothing is stopping me using something else that isn’t ChatGPT hope and pray code.

The files are deleted immediately after processing I'm considering implementing WebAssembly (Wasm) to do most of the work on the client's device and enable offline use

You say that here, but its not in the privacy policy as far as I can see.

When you get to the stage of monetizing the site, I expect the most obvious starting point is monetizing the information inside the pdfs.

Then there's the obscurity of how much (if any) is passed on to other services (like google). You may have one policy about the PDFs, they may gave another.

So yeah, I'm in the "not for general use" camp myself. (Although there are edge cases where it may be useful.)

Don't get me wrong, I can see the upsides, and your web site looks professional, but alas the downsides are too significant to overcome the inconvenience of searching out something local.

Pandoc, ocrmypdf, libreoffice, pdftk, pypdf2

For PDF to markdown, I recently released V2 of my tool marker - https://github.com/vikparuchuri/marker

This is very effective - it consistently yields great results.

There is a site listed elsewhere in these comments that does the work entirely in-browser. That's what I would use.

I use Preview on mac or even Spotlight for a good portion of these functions