Try this tonight. Open whichever AI you use and ask it to review your week.

It will congratulate you.

That is not a prompt problem, and no amount of prompt engineering will fix it. An AI with nothing to read gives you a horoscope. It works from whatever you happen to mention in the moment; it cannot see last month, and nothing accumulates, so the answer you get in December is worth exactly as much as the one you got in January.

The fix turned out to be much more boring than a better prompt. It is six short lines a day.

Thirty seconds at the end of the day, in your daily note. You are not journaling here; you are leaving evidence for later, and the test of a good entry is whether a stranger could act on it six weeks from now.

Useless: Busy day.

Good: Blocked, waited 90 minutes for Sam to approve the draft, could not start the next section.

Do that for a fortnight, and the AI stops guessing, because it finally has something to read. It can tell you that you committed to nine things and finished four. That one item has rolled over five weeks and is therefore not a task at all; it is a decision you have been avoiding. That every week containing a meeting before 10 am shows an energy score of two or below, across five separate dated entries.

Nobody has ever read your last eight weeks side by side before, including you.

Copy those six lines into whatever notes app you already use and start tonight. It costs nothing, it works anywhere, and it is genuinely the part that matters.

What took me months to work out was everything that came after. Which fields a review actually needs, how to structure them so the AI can retrieve properly, and how to write prompts that refuse to flatter you.

That last part matters more than it sounds. Roughly half of every prompt in the set is spent on what the AI must not do. Do not tell me it was a productive week. Do not credit me for partial progress. Do not end on an encouraging note. Strip those constraints out, and all sixteen prompts collapse into the same agreeable summary you already get for free.

Eight linked databases, sixteen prompts written against those exact fields, a thirty-day path that switches on one capture line at a time so the habit survives week two, a recovery protocol for the weeks you miss, and eight weeks of filled-in demo data so you can see the payoff before you have earned it.

I built it twice, once for Craft and once for Notion, so no one would be stuck in the wrong app. Both versions are complete and included in whichever tier you pick.

FIRSTWEEK takes $10 off until Friday at midnight.

One honest warning before you buy anything. The whole thing rests on capturing thirty seconds a day, and if you are not going to do that, none of the rest works.