I watched someone spend almost two hours writing a proposal with ChatGPT.
Write.
Edit.
Rewrite.
Start over.
Different prompt.
Different answer.
Repeat.
They weren’t using AI.
They were arguing with it.
The problem wasn’t their prompts.
It was their workflow.
Most people treat AI like one assistant that can do everything.
I don’t.
I treat it like a team where every person has one job.
This approach is called prompt chaining.
When a company launches a product, one person doesn’t do everything.
A researcher gathers information.
A strategist decides the direction.
A writer drafts the message.
An editor tightens it.
A critic tries to break it.
Each person hands their work to the next.
That’s exactly how I use AI.
Not one conversation.
A sequence of specialists.
Here’s an example.
Let’s say I want to write a LinkedIn post.
I never start with:
Write me a LinkedIn post about customer retention.
Instead, I build it one step at a time.
Employee #1: The Researcher
You’re a market researcher.
Employee #2: The Strategist
Look at these five insights.
Now I have an angle worth writing about.
Employee #3: The Storyteller
Turn this insight into a story.
I’m still not writing the post.
I’m building the foundation.
Employee #4: The Writer
Turn this into a LinkedIn post.
Now I have a first draft.
Most people stop here.
I don’t.
Employee #5: The Skeptic
This prompt has improved my work more than almost any other.
Pretend you disagree with everything I’ve written.
It hurts.
Almost every time.
It also makes the final version much stronger.
Employee #6: The Editor
Remove repetition.
Now it’s ready.
Same AI.
Six different jobs.
This workflow taught me something else.
Don’t ask AI to create.
Ask it to transform.
Creation starts with a blank page.
Transformation starts with an idea.
AI is much better at improving something than inventing it from nothing.
Don’t say:
Write a newsletter.
Say:
Here’s an observation.
Every prompt builds on the previous one.
Instead of asking AI to invent, you’re asking it to improve.
The difference in quality is huge.
One last tip as promised
Here is the key workflow and prompt
Never let one AI grade its own homework.
When you’ve finished a piece, open a fresh chat and say:
You’re reading this for the first time.
You have no knowledge of the previous conversation.
Tear this apart.
What’s confusing?
What’s weak?
What would make someone stop reading?
What questions remain unanswered?
A fresh chat doesn’t try to defend the decisions made earlier.
It evaluates the work with a clean slate.
I’ve found that this catches weak arguments, missing context, and lazy writing that the original conversation often misses.
This week, don’t open one chat and hope for magic.
Create six employees.
Give each one a single responsibility.
Pass the work from one to the next.
The quality doesn’t come from a better prompt.
It comes from a better process.
In the next newsletter, I’ll show you how I customize AI to think more like me.
We’ll cover ChatGPT Projects, Custom Instructions, and Claude Skills, and how I use them to give AI more context before I write the first prompt.
See you next week.
Pelumi