[Day 10 of 30] Think Like a Teacher When You Design Your System

Ask “What if” and use the right question format so your system teaches, not just delivers.

The situation

You are growing a personal system: a vault, rules for your AI, folder conventions, capture and review habits. It is easy to think like a PM (”what do I ship next”) or a developer (”how do I make it correct”). The system grows, but it does not teach you or the AI how to use it. When something odd happens (a note fits two projects, you skip review for a week, or two rules apply at once), things break or you patch on the fly.

Why does that keep happening? Because PM and developer minds optimize for different goals.

• The PM asks “what do I ship”;

• the developer asks “is it correct.”

• Neither asks “will the person or the AI who hits this rule in six months understand what to do and how to know it worked?”

If you design each rule and folder so that the “student” (future you or the AI) can see when it applies and when it is done, you get fewer surprises at edge cases, rules that survive context switches, and a system that compounds because both you and the AI can extend it without reinventing. That is what a teacher optimizes for: transfer of understanding, not delivery or correctness alone.

---

The 1% upgrade

Think like a teacher, not a PM or a developer.

• PM focuses on shipping and scope. Risk: the system “closes tasks” but does not teach context or boundaries. You get brittle rules and constant workarounds.

• Developer focuses on correctness and structure. Risk: the system is technically right but not explainable. You and the AI do not know “why it works that way.”

• Teacher focuses on understanding, spotting confusion, and step-by-step clarity. Goal: the “student” (you or the AI) can use and evolve the system on their own. Result: rules that work in edge cases and improve over time.

One practical way to do that is to ask questions in a clear format before you add a rule or a folder. Below: four formats and a prompt you can use today.

---

Question formats (use these when you add a rule or a folder)

All of these are about your system: vault structure, rules in CLAUDE.md, naming, capture and review habits.

• What if [scenario]? Surfaces edge cases. Example: “What if a note fits two projects?” “What if I skip daily review for a week?”

• When should [X] happen? Sets when a rule applies. Example: “When should this rule apply only to job-search chats?”

• How would [I / the AI] know that [criterion]? Checks clarity and testability. Example: “How would I know where to put this file in six months?”

• Why do we [action] here? Keeps the purpose explicit. Example: “Why do we capture this in the inbox and not in the project?”

You can stretch “What if” and related prompts to expected results (”What if I do X expecting Y?”), integration (”How can we connect a tool so that we do A and B?”), and behavior (”What should we do to make the feature work in a certain way?”). Same idea: stress-test the system before it breaks.

Anti-pattern: Questions that shut down thinking (leading questions, yes/no without context, “how do I do this faster” without “what should the user understand”). Prioritise “What if” and the formats above so you explore options instead of locking in one answer.

---

Copy-paste: five questions before you add a rule

Before you add a new rule or folder, ask yourself at least one of each type:

1. What if one edge case happens? (e.g. note in two projects, no review for a week, rule read in six months with no context.)

2. When should this rule or folder be used? (e.g. only in job-search conversations.)

3. How would I (or the AI) know the rule was applied correctly?

4. Why do we do this step here? (so the purpose stays in the rule or a one-line comment.)

Write one line into the rule or a comment so the system stays teachable for future you.

---

Copy-paste: prompt to generate three “What if” questions

Paste your rule, folder name, or a short description of a screen/workflow into the prompt below. The AI returns three “What if” questions you can use to stress-test it.

I am evolving a personal system (vault, rules for AI, folders, habits). Here is one element:

[Paste your rule, folder name, or a 2–3 sentence description of a screen/workflow]

Generate exactly three “What if” questions that help me find edge cases or confusion before they happen. Each question should start with “What if” and be about how someone would use this in real life (e.g. conflicting rules, missing data, wrong context, or future-me reading it). Output only the three questions, one per line.

Example output (for a rule like “Always put meeting notes in 00-Inbox/Meetings/”):

• What if the meeting is both a 1:1 and a project kickoff? Where does the note go?

• What if I have no time to move the note from Inbox for a week? What happens to my triage?

• What if someone else reads my folder structure in six months? Will they know why Meetings live under Inbox?

---

10-minute action

Pick one rule or one folder in your system. Paste it (or a short description) into the prompt above. Get three “What if” questions. Answer one of them in one sentence and add that sentence to the rule or as a comment next to the folder. Done.

Capturing habit. Do not leave the answer only in your head. Write it down so the system teaches future you and the AI.

Pick one rule or one folder you use often. Ask yourself one "What if" about it (e.g. edge case, expected result, or "how would I know it worked?"). Write the question and a one-line answer into the rule or as a comment next to the folder. If you like, reply here with that one "What if" so I can use it to tailor future prompts.

---

Skills that help you build skills

The same idea (stress-test with "What if," keep purpose explicit) applies when your system includes skills for Claude, Cursor, or your vault. If you write or refine those, the list below helps you do it in a structured way: process, templates, and clarity.