The Art of Prompting

Hey friends,

A few months ago, an employee at the Commonwealth Bank of Australia used ChatGPT to look up a client’s contact details.

It confidently gave them an answer — except it was the wrong business.

Same name. Different company.

Classic AI hallucination.

Here’s what happens when AI hallucinates: it tries to fill in the blanks. When it doesn’t have all the information, it guesses — confidently. It pieces together patterns, fills gaps with “best guesses,” and presents them as facts.

And if you don’t stop to fact-check, you can easily end up with a polished, professional-looking answer that’s completely wrong.

So yes, in that moment, AI failed. But so did we.

Because AI doesn’t know what we know. It doesn’t have context, or judgment, or memory. It only has the data and direction we feed it.

And that’s why so many businesses panic and ban AI altogether. They see the hallucinations, the misinformation, the mistakes — and they think the safest thing is to shut it down.

But that’s not leadership. That’s fear.

If you ban AI because it fails, you also ban the chance to learn how to use it well.

So instead of avoiding it, we need to learn how to talk to it differently. How to prompt it better.

🧩 The Real Dumb Monkey Move: Banning AI

When I talk to business leaders about AI adoption, someone always says,

I get it — there’s fear around privacy, misinformation, and risk.

But here’s what usually happens next.

In one company, after hearing that, we ran an anonymous survey. Turns out 40% of employees were already using ChatGPT anyway.

That’s the real problem.

When you don’t give people a safe, guided way to use AI, they’ll still use it — just without any guardrails.

You can’t ban curiosity. You can only train it.

My son is at that age where he wants to touch everything — switches, remotes, the stove, you name it. If I tell him “Don’t touch,” it works for about three seconds. But if I show him why it’s hot, or how to use it safely, he learns.

People are the same with AI. The more you try to ban it, the more curious they become.

So instead of banning AI, we need to teach people to use it safely.

Because if you don’t provide them with an alternative, they’ll still use AI — just dangerously.

💬 Prompt Engineering: But It Isn’t Engineering

So if banning isn’t the answer, what is?
It starts with how we talk to AI.

That’s where “prompt engineering” comes in — at least, that’s what most people call it. But the truth is, that term makes it sound far more technical than it really is.

Prompting isn’t engineering. It’s communication.

The word “engineering” makes it sound like something only experts or data scientists can do. But prompting is more like typing or sending an email — a foundational skill that everyone can learn.

In the next few years, being able to communicate clearly with AI will be as basic as knowing how to use a computer. It’s not a specialist skill. It’s a human one.

Think of prompting like delegating to a new intern.

You wouldn’t just say, “Make a report.” You’d tell them what it’s for, who it’s for, and what success looks like.

That’s why I use a simple framework called SCORE:

S – Situation: What’s happening right now?

C – Context: Who’s involved? What’s the background?

O – Objective: What do you want to achieve?

R – Role: Who should the AI act as?

E – Expectation: What kind of output do you need?

Here’s an example in action 👇

That’s not a prompt — that’s a brief.
And when you start briefing AI instead of commanding it, the results get dramatically better.

If you’d like more frameworks like this, I’ve shared all of them — SCORE, CORE, and LEARN — in my book The CEO Who Mocked AI (Until It Made Him Millions).
Or better yet, join the Dumb Monkey AI Community to learn and practice these with others building smarter workflows. (Links at the end 👇)

🔁 The Power of Asking Questions

Here’s a small tweak that changes everything.

Ask the AI to ask you questions.

Before you give it a task, try saying:

That one line turns your AI from a task-doer into a thought partner. It starts probing like a consultant — asking smart questions, identifying what’s missing, and giving you far more relevant results.

When I do this, the responses I get are smarter, sharper, and usually more creative.

By asking ChatGPT to ask questions, you’re encouraging it to think deeper about your problem — and that’s where the best insights come from.

🧠 Teaching AI Your World

Most businesses don’t realise their biggest asset isn’t their technology — it’s their people’s knowledge.

We call it tacit knowledge — the know-how that lives in people’s heads but rarely gets documented.

Here’s how we capture it at Enterprise Monkey:

1️⃣ Record internal meetings in Teams or Zoom (with consent).

2️⃣ Export the transcript.

3️⃣ Use ChatGPT or Copilot to turn those transcripts into detailed SOPs (Standard Operating Procedures).

This is how you start building AI that actually understands how your business runs. But understanding doesn’t stop at knowledge — it’s also about context.

Knowledge tells AI what we do. Context tells it how and why we do it.

That’s why every team member here builds three context files:

1️⃣ Company Context – our history, values, services, and tone.

2️⃣ Department Context – how each team works and measures success.

3️⃣ Role Context – the individual’s goals and responsibilities.

When you combine these, AI stops being generic. It starts thinking like your team — and speaking like your brand.

And that’s where tone and personality come in.

AI doesn’t sound human because it doesn’t know you yet. I fix that by training it on my voice — uploading my best-performing blogs and LinkedIn posts, setting up custom instructions, and turning on memory in ChatGPT.

I tell it:

• Who I am

• How my company communicates

• What tone to use

• To ask questions when uncertain

• And to always include a “confidence level” with every answer

Prompting AI is like talking to a person with short-term memory loss — you have to keep reminding it who you are until it sticks. But once it does, it becomes an incredible collaborator that thinks, writes, and speaks with your team’s shared brain.

🧭 Final Thought

AI will sometimes get it wrong — confidently wrong.

But that doesn’t make it the enemy. It just means it needs better direction.

And none of this matters if your team is too scared to use it — or worse, using it unsafely.

The fix is simple:

✅ Use enterprise-grade tools like Copilot or ChatGPT Pro (with privacy settings on).

🚫 Never paste confidential info into free models.

💡 Most importantly, teach your people why this matters.

AI isn’t dangerous. Ignoring how it works is.

When we learn to guide it with context, curiosity, and intention, it stops feeling risky and starts becoming useful — a partner that reflects the clarity of our thinking.

And if you’re ever unsure where to start, try this:

Because sometimes, the smartest move isn’t knowing all the answers — it’s knowing how to ask better questions.

📲 Resources & Links

🎧 Listen to the Podcast Episode 5 on: Spotify | Apple Podcasts | YouTube | Other Platforms

📘 Book: The CEO Who Mocked AI (Until It Made Him Millions) by Aamir Qutub

🚀 Dumb Monkey AI Academy

📱 Dumb Monkey AI Academy App: Apple | Android