Month 65

Four years in, you'd think one month would be hard to tell apart from the one before it. June wasn't. May sprawled — I said as much at the time — and June narrowed; it's the better month for it. This was when strackt started doing the boring, load-bearing things a managed platform is supposed to do without being asked. And near the end of it, for the first time, I stopped shipping features and started laying out the launch.

What Moved

None of what shipped this month is exciting, and that's the point. strackt learned to watch for known vulnerabilities in a stack, tell you what it found, and — with a nod — patch them. Certificates issue and renew themselves now, in production, with nobody touching DNS or a dashboard. And when a page in the app breaks, it surfaces instead of rotting quietly in a log. It's the chore work — 'your servers, our problem' getting literal — the maintenance I built the whole thing to take off your plate, finally taking it off mine.

The new interface kept filling in, too. Live application logs, so you can watch what your app is actually doing without SSHing in; a rebuilt backups surface; a clearer view of the services and capabilities a server is running. Last month I said the app should be the subject and the server the supporting cast — June was the month that stopped being a slogan and started being screens you can use.

And then the small thing that felt bigger than it was: on the 30th, strackt got its first piece a stranger can install without me in the room. Run one command — curl a script, or brew install it — and you've got the CLI. It's pretty small on its own; it's also the first public, installable part of strackt that exists in the world. The last week of June went to the rest of the way in: the production cutover, the deploy-and-rollback pipeline, the OS upgrade underneath.

What Didn't

The invite. Last month I wrote that I keep building the door and not opening it, and I'm not going to write that line again — it's true, and it's getting old. What's honest this month is narrower: I put a handle on it anyone can grab, and I mapped the opening down to the step, but no stranger has actually run strackt yet. Planning the launch is not launching.

And the remove-half is still undone. Last month I promised I'd start taking the old interface away — the 'remove' side of 'add then remove' — and I mostly didn't. There are still two of everything; I finished more than I subtracted. Adding is the fun part, and it shows.

Would I Pay For This Today?

Closer than last month, and for a different reason. Last month's yes was about stability and how the thing felt to use. This month it's about the maintenance gap — the actual pitch. My certificates renew themselves; my apps' errors come to me instead of sitting unseen; the vulnerability watch is wired in. That's the stuff I built strackt so I'd never have to think about, and it's running on my own servers right now. The asterisk hasn't moved, though: I'm still the only person who has ever used it. One user who wrote it isn't proof; it's dogfooding. I'd pay for what it does — I just can't yet tell you a stranger would.

Next Month

Open it. And this time there's a date on it: August 28, early access. Between now and then — get the production cutover done, get one stranger from an empty account to a running app, and finally start the remove-half I keep deferring: make the new interface the only interface. 'Mid-2026' has been the launch target for months; now it's a day I can count down to.

For four years, the honest answer to 'has anyone else run this?' has been no. Next month I want a different answer.