My week

The method behind it is GTD, and its centerpiece isn't the task list: it's the weekly review. It's what keeps the list trustworthy — and it's the first thing everyone abandons, because in the traditional version it takes hours.

This screen exists to fit in the gap between two meetings. It doesn't ask you to organize anything: it gathers the data, you look, decide and write down three commitments. If it regularly takes more than ten minutes, the problem is almost never you — it's too much accumulation between reviews, and the number in the footer is there precisely to show that.

Choose a fixed time (Friday afternoon or Monday morning) and protect it. A ritual that depends on remembering isn't a ritual.

A ten-minute ritual in five steps: empty the inbox, look at what's due, chase whoever owes you, find what's stalled and choose the three things for the week. The system gathers the data; the decision is yours.

It's the percentage of the last eight weeks in which you completed the review. A week you never opened doesn't count against you. The ritual's effect comes from REPETITION: eight ten-minute reviews are worth more than one two-hour one.

1 · Inbox — just the count of what's open in the inbox; emptying it happens there.
2 · What's due — your open requests due within seven days, overdue ones included, most urgent first. Shows up to 12.
3 · Who owes me — what you noted as "waiting for", oldest to newest. No cap: this list is meant to bother you.
4 · What's stuck — projects where you're the manager and that have no open request. It's the classic sign of a project with no next action: nobody abandoned it, it's just that nobody decided what the next step is. Shows up to 12.
5 · The week's three — three free fields, yours.

Two cuts to keep in mind: steps 2 and 4 cut off at 12 items (the review is for choosing, not taking inventory), and step 4 doesn't see projects you take part in without being the manager.

The duration of each review is recorded. If the ritual regularly goes past ten minutes, what failed was the screen's design — the number exists to hold the product accountable, not you.

You can't finish the review with the three fields empty: a review that doesn't end in a commitment is just reading a list. One is enough, but write at least one.

The format that works is what + by when. "Take care of the proposal" isn't a commitment; "send customer X's proposal by Wednesday" is. Three is the ceiling on purpose — whoever picks six hasn't picked anything.

Commitments are free text and don't automatically become requests. If one of them needs a deadline, an owner or follow-up, it deserves to become a real request — the review is the place to notice that, not to replace the board.

You can move between steps in any order and leave halfway: each step you check is saved immediately, and "save and continue later" returns to where you left off.

When you note something you're waiting for, it starts coming back on its own. The ruler is the interval you set (default every 3 days, from 1 to 60): counted from the last reminder, or from the request if there hasn't been one yet.

The reminder arrives through the platform's bell. When the debtor is someone in-house and you point them out in the dedicated field, they get the notice too. Customers and suppliers never receive automatic chasers — chasing a third party by automation is a relationship decision, not a software one; you're the one who gets the nudge, so you can chase in the way the relationship calls for.

The routine runs once a day, handling one batch at a time and starting with whoever has gone longest without a reminder — in a large base, an item may be a day late, never left behind forever. Running twice on the same day doesn't generate a repeated notice.

Each row's age gets a label: recent (up to 2 days), "worth a nudge" (3 to 7), "past due" (8 to 21) and "probably forgotten" (over 21). The bell on each row turns that item's reminder on and off.

Nobody reads what you write here — not your manager, not HR, not the platform administrator. It's a design decision: a review that someone else can read stops being honest in the first bad week, and a dishonest review isn't useful even to the person who writes it.

Time is added up in segments: at each action of yours (completing a step, saving, requesting the draft, touching "waiting for", completing the review) the interval since the previous action is added. An interval longer than 3 minutes adds nothing — during that stretch you were somewhere else.

In other words: opening the review, going to a meeting and coming back three hours later records the time you were actually on the screen, not the three hours. That's why the footer talks about "minutes on screen".

The number doesn't evaluate you: it audits the screen's design. If reviews regularly go past ten minutes, either the screen failed or the interval between reviews is too long — and for that reading to stay useful it has to measure work, not calendar time.

There's also no way to reopen a review that's already completed. Whatever was left to do goes into next week, which is where it should be anyway.

It receives exactly what's on the screen — the inbox count, the requests coming due, what you're waiting for with each item's age, the stalled projects — and returns an honest sentence about the week, up to three candidate commitments (each saying which concrete item it came from) and what can be let go.

It's instructed not to invent an item that isn't on your list and to say when information is missing. And it doesn't fill in the three fields: it stays alongside, and you copy whatever is useful.

If AI isn't active, the draft doesn't appear and nothing else changes — the review works perfectly without it. It just saves you the first draft.

The review is strictly personal: there is no manager, HR or platform administrator view. Not even the superuser reads what you write here.

It's a design decision, not an oversight: a review that someone else can read stops being honest in the first bad week — and a dishonest review is useless, even to the person who writes it.

It's the percentage of the last eight weeks in which you completed the review. A week you never opened doesn't count against you — only weeks in which there was some review are included.

It's the screen's most honest indicator: the ritual's effect comes from repetition, not from the depth of a single review. Eight ten-minute reviews are worth much more than one two-hour review.