What do you think of its potential for use in regions with slow and/or expensive Internet (South America seems to be the worst off here), where I could come to some arrangement with a local NGO/GO to fund server costs and development?

You absolutely could, Opera browser was doing something (I think they proxied images and similar files) before it gets to someone's screen. If you can compress large images (and you'd have to know they're images cause you gotta render them right?) in such a way that they're not visually distorted it would be worth paying for.

The harder thing is trust, trusting you to not change any images. On the other hand that'd make for a fun VPN service for general browsing (but again then it comes back to trust) but if you target it as a browser plugin bundled VPN service that saves on bandwidth you could hit more of an audience and use that income to fund Browsh. I figure you dont want to get too far outside of that scope though (development wise) since it's a lot more work.

I also like the other ideas of using some of the Browsh capabilities to scrape the web, maybe as a paid API, sure you can write a script to do that, but if you can curl / postman an API that does it for you and returns what you need, then why wouldn't you? Trick is can an API be done just perfectly enough? How do you tell it to give you whatever part of the screen with whatever requirements? It could sound simple but get complicated quickly enough. If it's too complex maybe a compression VPN is a better target.

Yeah the Opera-style compressing VPN idea is definitely high up on my list. That's why I released Browsh along side text.brow.sh and html.brow.sh At the moment they only return static content just as CDN would, so there's no privacy concerns, but to make those services truly VPN-like then they'll need to allow logins, sessions, etc, which then very much raises privacy concerns. And there's just nothing Browsh can do to assuage that, other than good branding, because Browsh by definition has to read every character on page in plain text.

I've got a good feeling about text.brow.sh, it's so simple. As you say I just need to work on the consistency of what it returns.

Thanks for your thoughts!

Maybe this would be useful to commercial shipping companies who run thier net over satellite and similar.

That would work nicely as a hosted service with an app. Like a more aggressive version of opera's bandwidth reduction.

The military might have similar problems.

It's not really ready fir consumers yet though.

Great ideas, I'd not thought of them. So might be worth researching some contacts in those fields.