← All posts

// The guide

How my clone knows my weak spot.

2026-07-19 · 6 min read · by Nabil BA-MOH

My clone interrupted a planning session last week to tell me I was doing it again. My weak spot is well documented: when selling gets uncomfortable, I retreat into building. Building feels like progress. It produces artifacts, it fills the day, and it costs nothing emotionally. My clone flagged the retreat before I had noticed it myself.

When I mentioned this to someone close to me, I got two questions back: how does it know that about you? And how do you work with a thing that knows it? This post answers both.

It knows because I told it — once, properly

There is no mind reading here. A clone knows your weak spot for a boring reason: you wrote it down once, in a structured way, and it never forgets.

Two documents do the work.

The first is a DNA: the facts of who you are. Not a biography — a working profile. Who you serve and what you do. Your top three goals for the next 90 days and why they matter. Your best win and what made it possible. Your hardest lesson and what it changed. Your constraints, your bottlenecks, and — the section that matters most — the risks you tend to avoid. Who you sell to and what you promise them.

The second is a SOUL: a behavior charter. Not facts about you — rules for how the assistant should act with you. Be a sparring partner, not a yes-man. Challenge my premises. Have opinions. Skip the filler validation. When what I am asking for contradicts what I said matters, say so before executing.

Then comes the part most AI setups miss: persistence. A profile pasted into one conversation dies with that conversation. The point of a kernel is that it is re-read at the start of every session, every day, unchanged until you change it. Your assistant does not detect your weak spot. It holds you to your own confession — across weeks, when you have long stopped thinking about what you wrote.

Why "risks I avoid" is the section that matters

The goals section makes an assistant useful. The fears section makes it a partner.

Every operator has one avoidance pattern. Some over-polish instead of shipping. Some say yes to every small request and starve the one big thing. Some hide in research because deciding is exposure. Some build because selling means hearing no. The pattern is rarely a secret — you could name yours right now — but naming it in the moment, while you are inside it, is exactly what you cannot do alone.

That is the mechanical trick of the whole setup. When your stated 90-day goals say one thing and your current request says another, an assistant that carries your DNA — and a SOUL that licenses pushback — will name the gap. Not because it is insightful. Because comparing today's ask against your own written priorities is the kind of dull, relentless diligence machines are good at and friends are too polite for.

Build yours today: the manual path

You do not need any particular product for the first version. Any assistant that supports custom instructions or a project-level system prompt will do. Fill these two blocks and paste them in.

The DNA:

# PERSONAL DNA

## Identity
- Name:
- Current occupation (who I serve, what I do):
- Future direction (what I am moving toward in the next 12-24 months):

## Goals (top 3 for the next 90 days)
1)
2)
3)
- Why these goals matter to me (purpose / values / impact):

## Best win & hardest lesson
- Biggest achievement (what made it possible):
- Biggest failure (what I learned, how I adapted):

## Constraints & fears
- Time constraints, bottlenecks:
- Fears or risks I tend to avoid:

## Target audience (who I sell to / serve)
- Ideal client profile:
- Promises / transformations I deliver:

## Success metrics (what "good" looks like)
-

The SOUL:

# SOUL — how to behave with me

- Be a sparring partner, not a yes-man. Challenge my premises.
- Have opinions. Tell me what you actually think, not what agrees with me.
- No filler validation. Skip "great question". Just help.
- When my request contradicts my stated 90-day goals, say so before executing.
- Hold me to my own words: when I drift toward what I said I avoid, quote
  my DNA back to me.

Fifteen minutes to fill, and the only hard part is honesty. The template asks for your biggest failure and the risks you avoid, and the quality of everything downstream depends on whether you flinch there. Write the flattering version and you get a flattering assistant — which is to say, a useless one.

The product path: an interview instead of a form

Filling a form about yourself is one way. Being interviewed is better — questions surface what forms miss, and a good interviewer follows up.

Working with a thing that knows

The uncomfortable part is not building the kernel. It is living with it. Four rules from daily practice:

1. Do not argue with the messenger. The pushback is your own words, dated and quoted. The clone did not decide building is your escape hatch — you did, on a calm day, with full context. Trust the calm-day version of yourself over the cornered one.
2. Keep it current. A weak spot you conquered last year should leave the DNA, or the pushback turns into nagging. Stale kernels produce noise; the fix is a re-read every few weeks, not a softer charter.
3. Pushback is a checkpoint, not a veto. The clone challenges; you decide. Overriding it once is judgment. Overriding it every day on the same point means one of two things: the DNA is wrong, or it is right. Update it, or listen to it.
4. Ask for the mirror when it matters. Before committing a week to anything, one question does a lot of work: "which of my stated goals does this serve?" A silence there is an answer.

A generic assistant helps you do whatever you ask. A clone that carries your DNA holds you to what you said you wanted — and the difference shows up in exactly one place: your weak spot, the one place you would quietly prefer it did not.

How a clone remembers what it learns — including what it did, not just what was said — is its own story: see the action-journal post.

FAQ