People hate ChatGPT until they learn this trick
I remember watching a former colleague Sarah throw her laptop across the conference room. Ok, not literally, but close. She'd been fighting with ChatGPT for twenty minutes trying to get it to write a simple client email, and every response was worse than the last.
"This thing is useless," she muttered.
But here's what Sarah didn't know, and what I learned the hard way.
The Day Everything Changed
Picture this: Several years ago, I'm sitting in a coffee shop, deadline looming, trying to get ChatGPT to help me write a proposal summary. I type "summarize this proposal" and attach my 47-page document.
Back comes this generic, lifeless blob of text that could've been about selling staplers or launching rockets to Mars. Completely useless.
Then the guy at the next table, clearly eavesdropping, leans over. "Mind if I show you something?"
What He Showed Me
"You're treating it like Google," he said. "But it's more like hiring a really smart intern who knows nothing about your company."
He took my laptop and rewrote my prompt:
"I run a small marketing agency. We just finished a proposal for a local restaurant chain that wants to revamp their social media presence. The proposal includes brand audit, content strategy, and a 6-month implementation plan. I need a two-paragraph executive summary that highlights our unique approach and the projected ROI. Make it sound confident but not oversell-y, these are no-nonsense business owners who hate marketing fluff."
The result? Exactly what I needed. First try.
The Three Things That Actually Matter
After that coffee shop encounter, I started paying attention to what separates good prompts from garbage ones. Turns out, it comes down to three things:
Tell it why you're asking. Don't just say "write an email." Say "I'm emailing a client who's been waiting two weeks for an update on their project, and I need to deliver some bad news about delays while keeping them from firing us."
Be stupidly specific about what you want. Not "help with my presentation" but "create five bullet points explaining why our Q3 numbers dropped, formatted as talking points I can read directly from my phone during the meeting."
Set boundaries like you're talking to your most literal-minded friend. "Don't use words like 'revolutionary' or 'game-changing.' Keep it under 200 words. Sound like me, not like a press release."
Why This Actually Works
Think about the difference between these two requests:
"Write a social media post about our new product."
versus
"We're launching a productivity app aimed at freelancers who are drowning in client communications. I need a LinkedIn post that speaks to that specific pain point - make it personal, maybe start with a question, and include a soft pitch for our beta program. Skip the hashtag soup and emoji spam."
The first one could apply to literally any product. The second one gives the AI a character to play, an audience to speak to, and a clear mission.
My New Routine
Now when I need something from ChatGPT, I spend 30 seconds thinking through:
What's happening in my world that AI needs to understand?
What's the exact end result I'm after?
What would make this response completely wrong for my situation?
That's it. No complicated frameworks or prompt engineering courses. Just basic context, clear instructions, and some guardrails.
Your Turn
Next time you're about to rage-quit on AI, try this instead: pretend you're briefing a new team member who's incredibly capable but knows absolutely nothing about your specific situation. Give them the context, the exact task, and the style guidelines.
Then watch what happens.
Trust me, it's way better than throwing laptops.
