I didn't know Alphabetum's PandocRuby but it seems pretty good and it shouldn't be hard to integrate, or provide a separate PDF generation function (analogous to gitprint) to work in tandem with any Wiki that can generate output in one of those document formats. That probably means any modern Wiki.
Were you thinking of any document formats in particular, Steve?

There's also wkhtmltopdf which uses the webkit engine to convert a webpage to PDF. I used it for a project recently, it's really convenient as you can stick it in front of any web app regardless of stack, and you can customised the output using a print media stylesheet.

Well, for myself, I was thinking markdown to PDF. HTML to PDF is a low priority, though I'm sure some folks would like to have it. Gitprint seems cool, but I can't find any documentation about how it works.

I only brought up gitprint as one of a number of possible tools (Mayel suggested another one). gitprint doesn't do much else but rendering markdown to PDF. Anyway, Alphabetum's PandocEuby, gitprint, or wkhtmltopdf, it's certain that we're capable of exporting to PDF taking input from practically any modern wiki that is able to produce a known format such as markdown or html. Guaranteed.
Here's the gitprint source code in case it's of interest
https://github.com/adamburmister/gitprint.com

OK, that makes me feel better.

To swing this back around to the original question, it seems to me that the natural first step is to install one of the so-called "meta-platforms," which I think is a misnomer. Nextcloud seems like the natural choice to me. (Nothing against Sandstorm, but Nextcloud looks more polished and cohesive.) It comes with a lot of what people would expect from an online office suite: word processing, a calendar, an email interface, and quite a bit more. Plenty of other things on the list look useful/cool, but this would take us the longest distance in a single step.

Another thought that's been bouncing around my brain today... If even a significant portion of these ideas were implemented, we could host a whole variety of new platform co-ops. The podcasting capacity would work for Taru's audio publishing idea. Riot with Jitsi would work for my remote interpreting idea. There are probably dozens more we can't even see yet. That calls to mind a phrase the Indianos use to describe themselves: "a community WITH businesses, not a community OF businesses."

And that might be where @ntnsndr 's idea of an international, federated umbrella co-op could come in. If we could get the incorporation step out of the way for start-up platform co-ops, then the only hurdles to clear are the ones we set up: the principles of cooperation and free software.

That might be the ultimate service.

As you know, I fully support this and I am eager to contribute (financially, at least) towards this becoming a reality. Also, I agree with @dougbelshaw when he says we should begin by assessing member needs before doing anything... Once we nail that, we can try our hand at expanding towards new areas of interest.
Count on me for this! :)

As for email though, while there is plenty of email server & webmail software, there is no easy solution if we want security (especially end-to-end encryption) built-in.

Protonmail has built a pretty cool solution, but it is not full open-sourced (and read their transparency page to see how much of a headache running an email service can be).

Even the security of services used by activists like Riseup, despite throwing in every trick in the book, ultimately relies on people's trust in the admins, and offloads any end-to-end encryption onto users (having to use browser plugins and PGP keys).

The ideal scenario might be to somehow negotiate a deal with Protonmail to run another instance of their server stack (or simply have accounts on their infrastructure, linked to our domain(s)), with public key federation (so that emails between social.coop and protonmail.ch would automatically be end-to-end encrypted).

I agree 100% with Mayel. Email is one of the trickiest, most demanding, and most unmerciful services to run, on top of that it's also the least profitable as corporations have created an expectation of gratis email (which they subsidise through their surveillance business model), and a kind of service that offers zero opportunity for differentiation (it either works and delivers emails properly, or it doesn't).

From every point of view except for the benefit of branding, email is a problem that is best outsourced. There's a cost to operating email services for hundreds of users, but I'm sure it's possible to find an ethical ISP keen on providing that service at a discounted rate.

Another example to illustrate the issue with email: scramble was a cool federated FLOSS project a while back, which shut down this year with the following message:

I apologize to the (apparently very few) people who may be inconvenienced by this :)

Scramble started back in late 2013 after the Snowden stories. We were shocked to learn how intrusive surveillance had gotten, and wanted to do something about it.

I'm happy to report that 3½ years later, end-to-end encryption is mainstream. The cool people at Signal and WhatsApp delivered strong crypto to over a billion people. For email, ProtonMail started with a very similar mission to ours and delivered a solid service. Check them out!