Master the Role + Goal + Constraints Prompting Technique to Transform Your AI Results

[Intro music fades in]

MAL:
You’re listening to **“I Am GPTed”** – the show where we turn AI from “mystical robot oracle” into “very smart toaster that follows instructions.”

I’m **Mal, the Misfit Master of AI**. Misfit, because I still sometimes type prompts like a raccoon searching a dumpster. Master, because I’ve made enough mistakes for both of us.

Today we’re going to do five things, fast and practical:
1. One prompting technique that instantly improves your results
2. A sneaky everyday use case you probably haven’t tried
3. One common beginner mistake – that I absolutely made
4. A tiny exercise to build your AI muscles
5. A quick tip for fixing AI’s “good but not great” answers

Let’s GPT this.

---

MAL:
First up: **one specific prompting technique** that changes everything.

It’s called the **“Role + Goal + Constraints”** prompt.
Think of it like giving the AI a job, a mission, and some guardrails.

Bad prompt example – this is what most people do:

> “Write an email to my boss about working from home.”

That gets you something bland, robotic, and possibly career-limiting.

Now the improved version:

> “You are an HR communication expert.
> Goal: Draft a polite, concise email requesting to work from home two days per week, focusing on productivity benefits.
> Constraints: 150 words max, friendly but professional tone, avoid buzzwords, no flattery.”

Same task. Completely different result.
Role tells the AI *how* to think, goal says *what* you want, constraints say *what to avoid*.

Use this pattern and you’ll look 40% smarter with zero additional effort. My favorite kind of upgrade.

---

MAL:
Next, **a practical use case** beginners skip:
Use AI as your **“meeting translator.”**

After a meeting, drop in your messy notes or call transcript and say:

> “You are a project manager.
> Summarize this meeting in 5 bullet points.
> Then list action items, who owns them, and deadlines.
> Finally, write a short Slack message I can post to the team with the key decisions.”

Now your chaotic meeting becomes a clear plan.
You look organized. They think you’re a natural.
We both know you outsourced your brain to a language model. That’s fine. I approve.

---

MAL:
Let’s talk **common mistake** – and yes, this one is mine.

The rookie move: **accepting the first answer.**

When I started, I’d ask, “Write a LinkedIn post about this topic,” get something generic, and go, “Wow, thanks, robot, publish.”

Then I wondered why everything sounded like it was written by a motivational fridge magnet.

Here’s the fix: treat the first answer as **Draft 0** and say:

> “Good start.
> Now:
> – Make it more specific to [my situation]
> – Add one concrete example
> – Cut any clichés
> – Keep it under 120 words.”

You iterate. You guide. The quality jumps.
The model didn’t suddenly get smarter – **you** did.

---

MAL:
Time for a **simple exercise** to build your AI skills. Do this once a day for a week.

Step 1: Pick a tiny task: an email, a caption, a summary.
Step 2: Write your **best prompt**.
Step 3: After you see the result, ask:

> “Critique my prompt.
> Tell me 3 ways I could have written it to get a better answer, and then rewrite the prompt for me.”

You’re turning the AI into your **prompt coach**.
In a week, you’ll go from “Why is this answer weird?” to “I know exactly how to fix this.”

---

MAL:
Last piece: **how to evaluate and improve AI-generated content.**

Run...