Most of my conversations revolve around the same two topics: AI and sustainability.
Which probably makes me unbearably boring to some people 😅
but it also keeps me focused on what matters to me.

Last week, I was in a discussion with friends and we reached that critical moment—the moment where someone says:

“Just Google it.”

But now it’s more like:

“Ask ChatGPT.”

So I was the one who had to prove the point.

I started typing my prompt. And to everyone else it looked painfully boring because I first asked a very general question…
and only then narrowed it down to the specific issue we wanted to solve.

(You’ll understand exactly why when you read Hack #4.)

But suddenly the discussion wasn’t about the topic anymore.
It was about how I prompted, why I started broad, why I layered my questions, and how the answer kept getting better.

At first, I thought they were joking.
Then I realised they were genuinely interested.

I always assumed everyone prompts this way.
Turns out—I was the only one in the room doing it.

And that’s when it hit me:
I actually have a few small “hacks” when it comes to asking GPT.

If my friends found them surprising and useful, I’m pretty sure you will too.

So in today’s article, I’m sharing the 4 hacks I use every single day:

Simple changes. Big difference.

Most people don’t get bad answers from AI because the model is weak.
They get bad answers because the prompt had no strategy behind it.

We expect AI to magically understand the context, the industry, the nuance, the question behind the question.

But AI is still just a reflection of what we ask.

And if we ask casually, we get casual answers.
If we ask with clarity, we get clarity back.

This has nothing to do with writing 20-line prompts or being a “prompt engineer.”
It’s simply approaching AI like a professional tool—not a guessing machine.

That was the moment I realised I had a few habits that consistently improved my output… even though I never named them.

And that’s where these 4 hacks come from.

So instead of accepting the response at face value, you force the model to validate its own answer and reveal uncertainty.

It’s the same principle we apply in sustainability reporting:
don’t trust data without assurance.

In sustainability and ESG the risk of “confident mistakes” is high:

This simple habit lets you prevent hallucinations and unreliable answers before they happen.

You get:
✔️ transparency
✔️ better accuracy
✔️ assumptions exposed

Here’s the exact prompt I use:

It takes a few seconds to add but drastically reduces misleading answers.

If you haven’t subscribed to my YouTube channel yet, you’re seriously missing out on practical, no-fluff content at the intersection of sustainability, career growth, and AI.

We’re just 334 subscribers away from hitting 500 — a milestone I’ve set for myself by the end of this year.

If you’ve found this article helpful, I’d love for you to join me on that journey. Let’s build a smarter, more skilled sustainability community together.

Do you think we can make it?

Most people ask AI for answers.
But for complex sustainability topics, it’s often better to ask AI to design the perfect prompt first.

You’re basically turning the model into your prompt engineer before you begin the real task.

This transforms the output completely.

You are already assigning the role before you even prompt

A vague prompt leads to vague results.

But when AI helps you create a stronger prompt framework,
you start with clarity and depth instead of trial and error.

Your output becomes:
✔️ structured
✔️ comprehensive
✔️ relevant to the scope

This is the exact one I use:

And if I want an even better outcome:

Always check the prompt and follow up with clarifications to get the best outcome

It doesn’t take more time—just more intention.

Not all AI tools are designed for the same tasks.
So before you start typing, describe your goal, the outcome you want, and the format—then ask which tool is best.

This helps you stop guessing and start choosing intentionally.

In sustainability and ESG work, we deal with different types of tasks:

One tool is rarely the best for everything.

Sometimes GPT is best.
Sometimes Perplexity is better.
Sometimes NotebookLM is better for document reasoning

Using the wrong model can ruin the output before you even start.

Using the right one:
✔️ improves accuracy
✔️ saves hours
✔️ improves clarity

Here is the one I use:

Optional add-on:

It’s such a small step—but it changes everything.

Instead of jumping straight into a detailed or complex question, start with a simple prompt that sets the context and role.

Think of it like “warming up” the model by anchoring it to the right topic and industry before diving deeper.

AI doesn’t always know what your real question is.
If you start too specific, you risk:

But when you set the scope first, the accuracy increases dramatically.

For example:
Start with a broad question like:

Then go deeper:

You’re guiding the model step-by-step instead of hoping it gets the full picture instantly.

This simple flow gives you:
✔️ less noise
✔️ better clarity
✔️ industry-specific answers

Here is the one I use:

Optional add-on:

It sounds simple, but it’s one of the fastest ways to improve AI output without writing long prompts.

If you haven’t seen my most popular AI use case for sustainability professionals—especially those working on audits—go check it out next.
And if this article helped, give it a like, leave a comment, and subscribe so I can keep bringing more practical use cases.