Thirty days, one small thing a day

What a day looks like

One exercise, named and numbered, with a real target — never “just practise”. Then a few seconds of video to show you did it. That is the whole loop, and it is short on purpose: the thing that ends most attempts at learning something is not difficulty, it is not knowing what today’s version of the work is.

Nobody has to be on camera on day one. The exercises start hands-only, move to your voice, and only later ask for your face — and every day is checked against the minutes you said you had before you ever see it.

One take. No do-overs.

There is no delete and no second attempt. That is a product rule rather than a setting, and the app is built so it cannot be got around — the route that would replace a clip does not exist. A record of the work as it actually went is worth more than a highlight reel of the takes that went well.

Who decides what we launch

An agent does. It researches who might want a thirty-day challenge and why now, proposes an audience with the evidence for it, writes the exercises, and reads the results afterwards to decide whether to scale that challenge, pause it, or stop. Every proposal it makes is cited, every decision it takes is recorded — including the ones it was refused.

It is refused quite often, and deliberately. It can never put a challenge in front of more than a small share of people until the evidence says so. It cannot touch health, financial or legal subjects on its own, whatever it thinks. And it can propose all it likes without anything appearing here.

No exercise reaches anybody until a person has read it and approved it. That is not a policy we intend to follow — it is a gate in the software, and publishing is refused while a single exercise in a challenge is still unread. Everything on this site has been through it.