I accidentally turned GPT into my “annoying but genius coworker” and it fixed my workflow

So I’ve been using GPT for months like most people do.

Me: “Write me a landing page”
GPT: “Sure! Here is the most generic landing page ever written by a polite toaster.”

I kept blaming the model, the updates, the internet, mercury retrograde, everything.

Then last week I was tired and typed something like:
“Before you answer, ask me what you need to know. And if my idea is dumb, say it.”

GPT instantly replied with questions I should have asked myself 2 days earlier.

I tried it again with a client task (Upwork). Instead of “write proposal”, I forced a structure:

  • ask clarifying questions first
  • challenge assumptions
  • give 3 versions (safe, bold, and “if I had confidence”)
  • summarize then expand into steps

Result: I sent a proposal that actually sounded like a real person with a brain.
Client replied in 20 minutes. I stared at the screen like it was witchcraft.

The funny part is, once you start using these prompt “frameworks”, GPT stops feeling like a chatbot and starts feeling like that coworker who is slightly annoying but always right.

Anyway, if you want the collection of prompt frameworks I’m using lately, I put them here (more examples there too):
https://allneedshere.blog/prompt-pack.html

If you have a task you’re stuck on, drop it below and I’ll share which prompt structure I’d use.

Leave a Reply