Episode Details
Back to Episodes
Master Context Stacking to Transform Your AI Prompts From Vague to Perfectly Tailored
Published 11 hours ago
Description
# I am GPTed: The Prompt That Changed Everything
---
**[COLD OPEN]**
Hey, I'm Mal, and welcome back to "I am GPTed"—the show where we turn you into someone who actually knows what they're doing with AI instead of just asking ChatGPT to write your grocery list. Though, hey, no judgment. I've been there.
Today we're tackling the one prompting technique that'll make your AI responses go from "meh" to "wait, how did it know that?" Spoiler alert: it's not magic. It's just being specific. Revolutionary, I know.
**[SEGMENT 1: THE PROMPTING TECHNIQUE]**
Here's the thing nobody tells you: AI is like a really smart golden retriever. Throw it a vague command, you get vague results. Be crystal clear, and suddenly it's doing backflips.
The technique is called **context stacking**—and it's exactly what it sounds like. You don't just ask your question. You give the AI the who, what, when, where, and why first.
Let me show you the difference.
**Bad prompt:** "Write me a professional email."
**Good prompt:** "Write a professional email from a project manager to a client who's upset about a delayed deadline. The tone should be apologetic but confident—we have a plan. Keep it under 150 words. Use their name (Sarah) and reference the specific project (Website Redesign Phase 2)."
See the difference? The first one gets you corporate boilerplate. The second one gets you something you'd actually send.
**[SEGMENT 2: THE PRACTICAL USE CASE]**
Here's where most people sleep on AI: **brainstorming with constraints**. Not the "write me a novel" stuff. I'm talking real life.
You're planning a birthday party for someone who's impossible to shop for. Instead of spiraling, ask Claude: "I'm throwing a 40th birthday party for someone who loves hiking, hates small talk, and has a weird sense of humor. Give me five activity ideas that don't involve forced mingling." Boom. Actual useful suggestions tailored to a real human.
Or you're stuck on how to explain a complex work concept to your non-technical team. Feed your AI the concept, your audience, and one constraint—"no PowerPoint jargon"—and you've got a script in minutes.
**[SEGMENT 3: THE BEGINNER MISTAKE]**
The biggest mistake I see? And I'm admitting this because I did it for like three months: people don't iterate. They ask once, get a response, and think that's the final answer.
Wrong. Dead wrong.
AI responses are drafts. They're starting points. If something's off, you tell it what's wrong and ask again. "That's too formal" or "Make it shorter" or "I meant this kind of funny, not that kind of funny." Each time, it gets closer to what you actually want.
I used to think I was bad at prompting. Turns out I was just impatient.
**[SEGMENT 4: THE PRACTICE EXERCISE]**
Here's your homework, and it takes ten minutes:
Take something you wrote recently—an email, a message, whatever. Feed it to an AI and ask it to rewrite it in three different tones: "like you're explaining to a five-year-old," "like you're a skeptical journalist," and "like you're a motivational speaker on too much coffee."
Notice what changes? That's how you learn tone. That's how you get better at telling AI what you actually mean.
**[SEGMENT 5: EVALUATING AI CONTENT]**
Before you use anything an AI generates, ask three questions:
One: Does this sound like me, or does it sound like a robot pretending to be human?
Two: Is there anything factually wrong? Spot-check the claims, especially dates and numbers.
Three: Is this actually useful, or am I just using it because it's fast?
If you're hitting two out of three, you're winning. If it's all three, you've found your AI sweet spot.
**[OUTRO]**
That's what w
---
**[COLD OPEN]**
Hey, I'm Mal, and welcome back to "I am GPTed"—the show where we turn you into someone who actually knows what they're doing with AI instead of just asking ChatGPT to write your grocery list. Though, hey, no judgment. I've been there.
Today we're tackling the one prompting technique that'll make your AI responses go from "meh" to "wait, how did it know that?" Spoiler alert: it's not magic. It's just being specific. Revolutionary, I know.
**[SEGMENT 1: THE PROMPTING TECHNIQUE]**
Here's the thing nobody tells you: AI is like a really smart golden retriever. Throw it a vague command, you get vague results. Be crystal clear, and suddenly it's doing backflips.
The technique is called **context stacking**—and it's exactly what it sounds like. You don't just ask your question. You give the AI the who, what, when, where, and why first.
Let me show you the difference.
**Bad prompt:** "Write me a professional email."
**Good prompt:** "Write a professional email from a project manager to a client who's upset about a delayed deadline. The tone should be apologetic but confident—we have a plan. Keep it under 150 words. Use their name (Sarah) and reference the specific project (Website Redesign Phase 2)."
See the difference? The first one gets you corporate boilerplate. The second one gets you something you'd actually send.
**[SEGMENT 2: THE PRACTICAL USE CASE]**
Here's where most people sleep on AI: **brainstorming with constraints**. Not the "write me a novel" stuff. I'm talking real life.
You're planning a birthday party for someone who's impossible to shop for. Instead of spiraling, ask Claude: "I'm throwing a 40th birthday party for someone who loves hiking, hates small talk, and has a weird sense of humor. Give me five activity ideas that don't involve forced mingling." Boom. Actual useful suggestions tailored to a real human.
Or you're stuck on how to explain a complex work concept to your non-technical team. Feed your AI the concept, your audience, and one constraint—"no PowerPoint jargon"—and you've got a script in minutes.
**[SEGMENT 3: THE BEGINNER MISTAKE]**
The biggest mistake I see? And I'm admitting this because I did it for like three months: people don't iterate. They ask once, get a response, and think that's the final answer.
Wrong. Dead wrong.
AI responses are drafts. They're starting points. If something's off, you tell it what's wrong and ask again. "That's too formal" or "Make it shorter" or "I meant this kind of funny, not that kind of funny." Each time, it gets closer to what you actually want.
I used to think I was bad at prompting. Turns out I was just impatient.
**[SEGMENT 4: THE PRACTICE EXERCISE]**
Here's your homework, and it takes ten minutes:
Take something you wrote recently—an email, a message, whatever. Feed it to an AI and ask it to rewrite it in three different tones: "like you're explaining to a five-year-old," "like you're a skeptical journalist," and "like you're a motivational speaker on too much coffee."
Notice what changes? That's how you learn tone. That's how you get better at telling AI what you actually mean.
**[SEGMENT 5: EVALUATING AI CONTENT]**
Before you use anything an AI generates, ask three questions:
One: Does this sound like me, or does it sound like a robot pretending to be human?
Two: Is there anything factually wrong? Spot-check the claims, especially dates and numbers.
Three: Is this actually useful, or am I just using it because it's fast?
If you're hitting two out of three, you're winning. If it's all three, you've found your AI sweet spot.
**[OUTRO]**
That's what w