AI Guides › Playbooks
By Nigel Guy · 6 min read
The usual way to "reinvent yourself" is a big planning session: a new identity, a ten-year vision, a colour-coded plan. It feels productive because the document looks impressive. By Monday it has changed nothing, because nothing in it was ever a thing you did.
A day is not enough to rebuild a life. It is enough to decide what has stopped working and to make one real move that someone or something outside your head can see.
The rule: in 24 hours you may make one cut, one move and one visible trace. Anything beyond that is planning theatre.
The Card is a single page with five fields, filled in order. Claude is useful for two of them and should be kept away from the rest. You can use a paper notebook if you prefer; nothing here needs software.
| Field | What goes in it | Who does it |
|---|---|---|
| 1. The Stop | One thing that is no longer working, stated as a plain fact | You |
| 2. The Evidence | Two or three concrete observations behind it | You, then Claude challenges |
| 3. The Move | One action, doable in under 90 minutes, that changes the situation | You, Claude sharpens |
| 4. The Trace | Something visible that proves it happened | You |
| 5. The Review | A date and time, within seven days, to look at what happened | You |
Finish this sentence without a "because" clause: "What is no longer working is ___." Examples of the right size: "My current job's weekly review meetings," "Saying yes to freelance work below my rate," "Evenings spent on a side project I no longer want."
Too big: "My career." Too vague: "I feel stuck." If you cannot get it down to one thing, pick the one that irritates you most this week. You are choosing a starting point, not ruling on your life.
Write two or three observations that are facts, not feelings: dates, hours, amounts, things people said to you that you can recall precisely. "I have dreaded the Thursday meeting for most of this quarter" is a feeling with a date on it; "I have spent about four hours a week on it and cannot name a decision it produced" is closer. If you are not sure of a number, write "not sure", do not round it into certainty.
This is where Claude earns its place. It is good at asking the awkward question you skipped and at turning a vague intention into a small action. It does not know your life, and it will happily build a tidy case for whatever you tell it. So you instruct it to push back.
You are a direct, practical thinking partner. I have one hour to decide
one change and one action. Do not give me a life plan.
What is no longer working: [THE_STOP]
Evidence I have (facts only): [EVIDENCE_LIST]
What I cannot change right now (money, contract, caring duties, etc.): [FIXED_CONSTRAINTS]
Time and energy I realistically have in the next 24 hours: [HOURS_AND_ENERGY]
Do this in order:
1. List the two strongest reasons my conclusion could be wrong or
incomplete. Quote my evidence when you do.
2. Tell me which piece of my evidence is weakest and what I could check.
3. Propose three possible Moves. Each must be doable in under 90 minutes,
within my constraints, and must change something outside my head
(a message sent, a thing cancelled, a booking made, a draft shared).
4. Recommend one, and say what a good outcome and a bad outcome would
look like within a week.
Rules: if I have left something out that you need, ask me before you
answer. Do not invent facts about my situation. If a Move would affect my
health, legal position, finances or someone else's wellbeing, say so and
tell me who to speak to first. Before replying, check that every Move is
small, specific and has a clear first action.
Fill in the four bracketed items. Read the answer sceptically: if the "reasons I could be wrong" are all generic, tell Claude to be specific to your evidence and run it again.
Do the Move before you do anything else tidy. Then leave a Trace: a sent email, a cancelled subscription confirmation, a calendar block, a photo of the cleared shelf, a one-line message to a friend saying what you did. The Trace is the difference between a decision and a mood. It also makes it harder to quietly undo.
If you want Claude to help, use it to draft the message or checklist, then send it yourself.
Put a 20-minute slot in your diary within seven days. At the review, answer three questions on paper: what did the Move change, what did it cost, and is The Stop still the right Stop? Then choose one of three outcomes: repeat, adjust, or pick a new Stop. A second Card is a second 24 hours, not a continuation of the first.
Using Claude's Projects (at time of writing, available on the free plan too, with a limit of five projects) lets you keep your Cards in one place. In Claude, go to Projects, choose "+ New Project", then use "Set project instructions" to paste a short standing brief, for example: "Act as a challenger, not a cheerleader. Ask for missing facts before advising." Anthropic's help page notes that context is not shared across chats within a project unless it is added to the project knowledge, so upload or paste your previous Card there if you want Claude to see it.
Memory is on by default for Free, Pro and Max plans at time of writing. If you would rather a sensitive review did not feed it, start an incognito chat (the ghost icon, outside projects), or pause memory under Settings > Memory. Paid plans start at about $20 a month (about £15 at time of writing; check the £ price at checkout), but nothing in this guide needs one.
It will not tell you whether to leave a job, a relationship or a city. It is built for a reversible decision you can test in a week. If The Stop turns out to be a large one, the Move is to take the first small step toward finding out: a conversation, a quote, a trial.