The Work That Doesn't Show
Stripe is wired up, a test payment went through on production, self-hosted analytics is running on my own server. A week of work on the hub — and from the outside the site looks exactly the same as before.
By Georg Malahov
I keep grinding away at my malahov.io — getting it ready for launch. A surprising amount has been done this past week, by July 12; it's just not visible from the outside.
I finally wired up Stripe — the payment service that handles cards and subscriptions. Test-mode Stripe has already worked on production: sounds odd, but that's how it is — a live site, with payments still going to the sandbox, a special test mode with no real money. The paywalls — the screens that ask you to pay — look untidy and unoptimized, but the payment mechanics themselves are alive. I set up the user invitation mechanism. I stood up Umami — an open-source analytics system — on my own server: it works without cookies or sneaky pixels, and all the data stays on my infrastructure in Europe. Left to test: scheduled email sending and the subscription; and for the legal texts — Datenschutz, the data-protection texts German rules require, and the rest — I've kicked off a separate task to get a plan of what exactly to write.
The thought all of this exists for: I want to set up a shared base once — technical, legal, infrastructural. Texts, blog, changelog, sign-up, payment, emails, agreements with providers — all of that machinery built end to end, ideally properly. Then every next product will only need a landing page, not months of plumbing. So much labour to turn scattered projects into a single system.
And this is work that visually leads to nothing. I do it and do it and do it — and the site looks exactly the way it did a week ago. Not one new button, not one new screen. Worse than that: after the latest releases the master-password login that keeps the site closed until launch broke — so a week of work actually made the site look worse, until I fixed it. All that's left is to believe this leads somewhere, preferably somewhere good.
What should come out the other end. The hub will hold three products: Subtitle Editor — a browser-based subtitle editor; dictation — an app where you talk and get clean text back; and an external link to Call Copilot — an extension that transcribes calls, which for now lives outside on its old infrastructure. I want to move it inside eventually too, to cut the spending on third-party services, but for now it stays as it is. Plus my articles — one way or another, some will get published. The result is a big site with several projects, different landing pages, and a blog.
Launch is the stretch goal for next week: open everything at once. Checking myself will be easy: either the site is open without a password a week from now, or I've buried myself in details again.