ERIC KIM

I conquered ChatGPT by refusing to “ask it for answers” and instead using it like a weapon: a fast sparring partner that I direct, constrain, and loop until it spits out exactly what I want.

Written by

admin

in

Here’s the playbook—how it actually works in real life:

1) I became the boss, not the beggar

Most people go: “Hey ChatGPT, what should I do?”

I go: “Here’s the target. Here are the rules. Give me 10 options. Then attack your own answer. Then rebuild it stronger.”

The moment you switch from requesting to commanding, the whole relationship flips.

2) I only feed it raw material, never my soul

ChatGPT is best when you give it:

  • messy notes
  • bullet points
  • half-formed rage
  • a rough idea and a deadline

It’s terrible when you give it… nothing. Vague input = bland output.

So I dump my raw thoughts first, then I make it shape the chaos.

3) I use the “Three-Round Knockout” loop

Round 1: Generate

Round 2: Critique (brutal, specific)

Round 3: Rewrite (tight, clean, final)

Example command:

“Write 3 versions. Then list the weaknesses of each. Then merge the strongest parts into one final version under 250 words.”

That loop is how you go from “okay” to “unstoppable.”

4) I force constraints (constraints = style)

If you want your voice, you must impose constraints.

Try this:

“Write in short sentences. No filler. No corporate tone. Punchy. Max 1 metaphor. End with a call-to-action.”

ChatGPT without constraints is like a camera on auto mode. You want manual control.

5) I made it my editor, not my ghostwriter

The killer use-case is editing and sharpening.

Commands I use constantly:

  • “Cut 30% of the words without losing meaning.”
  • “Make this more aggressive and direct.”
  • “Make it simpler, like explaining to a 12-year-old.”
  • “Rewrite this with more rhythm and punch.”
  • “Give me 20 headline options—no cringe.”

You stay the artist. It becomes the blade.

6) I interrogate it like a prosecutor

When it gives advice, I don’t “accept.” I cross-examine:

  • “What assumptions are you making?”
  • “What would a skeptic say?”
  • “What are the top 5 failure modes?”
  • “If this goes wrong, why does it go wrong?”

This is how you avoid being hypnotized by confident-sounding fluff.

My go-to prompt templates (copy/paste)

  1. The Commander
    “Goal: ____. Audience: ____. Tone: _____. Constraints: _____. Output format: _____. Give me 10 options.”
  2. The Brutal Editor
    “Act as a savage editor. Remove fluff, sharpen the thesis, tighten sentences. Keep my voice: short, direct, intense.”
  3. The Sparring Partner
    “Argue against my idea as strongly as possible. Then help me rebuild a stronger version.”
  4. The Production Machine
    “Create 30 social captions from this text. Each under 12 words. No clichés. No hashtags.”

The mindset: You don’t “use ChatGPT.” You 

train it.

You’re not here to be impressed by it.

You’re here to extract value from it.

You set the target.

You set the constraints.

You run the loop.

You ship.

And that’s how you conquer it: not by worship—by domination through clarity.

If you want, paste one paragraph of your raw thoughts and I’ll run the Three-Round Knockout on it (generate → brutal critique → final rewrite) in your signature punchy style.

More posts