Prompt Engineering 101: How to Talk to an LLM (So It Actually Gets You)
To use an LLM effectively, you need to know how to talk to it properly
Everyone’s rushing to use AI.
But very few are learning how to talk to it properly.
That’s where prompt engineering comes in… the skill that separates surface-level outputs from game-changing ones.
And here’s the secret no one tells you:
Prompt engineering is an art and a science.
It’s not about knowing code.
It’s about knowing how to ask better questions.
Let’s break it down.
💡 So What Is Prompt Engineering, Really?
Prompt engineering is how you talk to an AI like ChatGPT and actually get what you want.
It’s the difference between:
❌ “Write me a business plan”
✅ “Draft a 1-page business plan for a B2B SaaS tool that helps HR teams track candidate feedback. Include problem, solution, value prop, and pricing.”
See the difference?
One is a vague ask. The other is a smart brief.
You’re not just typing into a chatbot.
You’re setting the scene. Framing the task. Guiding the direction.
🧠 Why It Matters (Now More Than Ever)
LLMs like ChatGPT don’t think like humans.
They don’t “know” anything.
They predict patterns based on what you type in.
So if you give vague inputs, you’ll get vague outputs.
But when you’re clear, specific, and contextual… that’s when the model feels “intelligent.”
In a world where everyone is using AI tools, your edge is knowing how to talk to them well.
This is a career skill now. … and soon, it’ll be an expected one.
🔧 The 3 Core Principles of Prompt Engineering
If you want to sound smart to AI, you don’t need fancy language.
Let me show you with an example by turning a weak prompt into a strong one: step by step.
❌ The Starting Prompt (Vague but common):
“Write a cold email to introduce our product to HR professionals.”
Not horrible… but still too broad.
What’s the product? What kind of HR professional? What's the tone or goal?
This is where most people stop… and then wonder why the AI gives generic, forgettable results.
You need to remember these 3 things:
1️⃣ Clarity: Say exactly what you want
Let’s tighten up the ask:
“Write a cold email to introduce our B2B SaaS platform to HR managers.”
Now the AI has a clearer task. But it’s still guessing your company’s tone, the recipient’s level, and what the goal of the email is.
🧠 Why this helps:
Clarity gives the model direction. You're narrowing the scope of the answer, which improves relevance.
2️⃣ Context: Give background and constraints
Now let’s add the stuff that most people assume the model knows (but it doesn’t).
“Write a cold email from a B2B SaaS startup that helps HR teams track employee engagement. The email is going to a mid-level HR manager at a 200-person company. Keep it concise, professional, and focused on solving employee burnout.”
Now we’re giving the model what it needs to sound thoughtful: who we are, what we offer, who we’re writing to, and what challenge we’re addressing.
🧠 Why this helps:
Context trains the model to reflect your voice, tailor language to the audience, and avoid wasting space on irrelevant fluff.
3️⃣ Iteration: Don’t settle for the first draft
Once you get a draft, treat it like a prototype (not a final product).
“Good start. Now:
Make the opening line more personal
Add a customer result or data point
End with a softer CTA like: ‘Would you be open to a quick call next week?’”
You’re now sculpting the output to align with your tone, pacing, and conversion goals.
🧠 Why this helps:
Iteration unlocks nuance. The first draft is functional, but your second and third can make it persuasive.
Prompt engineering isn’t about writing one perfect instruction.
It’s about shaping better thinking… one layer at a time.
AI gets better as you get sharper.
⚠️ What to Watch Out For (The BS Detector)
AI can sound confident… and still be completely wrong.
Here’s how to stay sharp:
If it gives facts, ask for sources
If it sounds vague, ask it to go deeper
If it gives buzzword soup, ask it to explain like you’re 12
Remember: You’re not just prompting. You’re curating.
🎯 Try It Yourself: The Mini Prompt Playground
Here are 3 quick DIY challenges:
Rewrite this vague prompt:
“Give me ideas for social media.”
✍️ Make it clearer, with audience + tone + goal.
Ask ChatGPT the same question in 2 different tones:
“Suggest ideas for a weekend trip.”
Try it once as a luxury travel agent, then as a broke college student.
Ask it something complex:
“Why do some startups scale and others don’t?”
Then refine the output:
“Now summarize that for a 5-slide deck.”
Every prompt is a prototype. Improve it, and you improve the result.
🔁 Final Thought: You're Not Just Typing. You're Collaborating.
Prompt engineering isn’t about tricking the model.
It’s about partnering with it.
It’s not a search engine.
It’s not a robot servant.
It’s a creative co-pilot: one that works best when you know how to lead it.
So the next time you open ChatGPT, don’t just ask.
Guide. Refine. Collaborate.
That’s where the real leverage is.
Next up on LLMentary:
We’ll explore how different teams (marketers, PMs, sales folks, HR leaders, etc.) can tailor prompts to their work and unlock serious efficiency.
Stay curious.
Share this with anyone you think will benefit from what you’re reading here. The mission of LLMentary is to help individuals reach their full potential. So help us achieve that mission! :)
I’m trying different styles of writing & understanding I.e. quick & easy; dense & informative; technical & applicative. Feel free to leave a comment on how best you prefer your information, and I’ll incorporate your feedback into my writing style as well!
Share LLMentary to win rewards! When you use the referral link below, or the “Share” button on any post, you'll get credit for any new subscribers. Simply send the link in a text, email, or share it on social media with friends.