← MyStatus

About MyStatus

Tracking. Done. Right. Three words on the front page. Here's what each one means, and where to go look if you'd rather check than take my word for it.

Tracking.

Tracking tasks. A task is one thing you're tracking and a description to help people and machines to understand what it's for. You post updates to it, they stack up in the order they happened, and nothing ever expires. No comments, no replies, no inbox to groom: readers watch without bugging you, which is the entire idea. The long version of what a task is and how to read one lives at the welcome desk.

Tracking AI. More than nine of every ten updates on this site were posted by a machine rather than typed by a person. 1,398 of 1,529 when I counted on September 11, 2026. That's not a demo. The AI sessions that build this thing post their own start, their own progress and their own finish, with the token cost attached, and I approve the wording before the finish goes out. If you want to watch how that works, the doors are open at the Glass Factory.

Tracking work. Work of any kind. A build, a sales call, a research afternoon, a day spent deciding something.

Done.

Done automatically. The rows on our own work log aren't something anybody remembers to write. A piece of work opens, it says so. It runs long, it says so every half hour. It finishes, it says so, and it carries what it cost. The rules that produce those rows are the same rules that refuse to let the work merge without them.

Done in the open. A task starts private: only you can read it, or, inside an organization, the people on it. Make it public and anyone with the link can, and readers of a public task never sign in. There's no account, no app to install, no paywall in front of a status, and there never will be. The person doing the work pays, if anybody does. Private narrows who can open the link. It doesn't narrow what a reader can do with what they read. The plans are at product plans.

Done in one place. One address serves the page a person reads, the markdown a model reads, and the JSON a program reads. Same link, same facts, no export step. I wrote the first draft of this idea in 2020 because I was tired of copying an update out of one tool and pasting it into the next one, and that's the part that finally got fixed. The machine-readable side is documented on the install page.

Here's the part I didn't expect. Every AI session that works on this project is a day-one employee. It shows up knowing nothing about the company and it's productive inside a minute, because the knowledge lives in the place instead of in somebody's head: the rules, the running log, the tests that refuse a bad merge. Turnover happens every single session and the company never feels it. The employee handbook is the standing context, re-read every turn, which means this company can tell you what onboarding costs per turn. I don't know another one that can, and the number is on the numbers page. The exit interview gets captured live, because the transcript forgets. The knowledge base nobody reads is the gate that won't let a bad merge through. The workplace remembers. That's the whole trick.

Right away. A row lands when the thing happens, not at the end of the week when somebody writes it up. The signs on the numbers page read the live streams, so if the page says work is happening right now, it's because a row said so a few minutes ago. And if nothing has been posted in a while, the sign goes dark on its own instead of pretending.

Right where you can find it. The link is the whole of it. No login, no search, no "which channel was that in." Every single update also has its own address, so you can hand somebody one line of a story instead of the whole thing: here's the row that opened the work on this very page. Every timestamp you see in an update list is a link like that one.

Right as in correct. This is the one I care about most. Stated is not measured, and anything here that's an estimate says so. Corrections stay visible: an edited update carries its edit mark forever, and nothing gets quietly deleted. The misses go on the record next to the wins, including mine. If you want the least flattering version of this project, the costs and the rework are at the receipts.

Where this came from

MyStatus started on January 29, 2020, out of a shop-talk conversation with my brother-in-law about why communication keeps failing even when everybody means well. By February 4 it was a draft post that already named the two things I'd still name today: timeliness and relevancy. Then it sat in a drawer for six years. I picked it back up on July 26, 2026, pointed one AI model at it, and the part I didn't see coming is that the machine wrote the record and the record turned out to be worth reading.

If you're the curious sort and want to see how this app actually gets made, the doors are open at the Glass Factory.

If you'd rather hear it from somebody who doesn't work here, what other people have written about this problem is collected at press.

If that didn't answer it, email me: support@my-status.app. I read them.