Each morning, Lisa stopped by the same neighborhood coffee shop. On her first visit, she ordered simply, “Just a latte, please.” It arrived warm and pleasant, but not quite what she had hoped for. The flavor was too light.
The next time, she asked for a stronger one. It came back bitter and overpowering.
It wasn’t until two weeks later that she figured out how to ask properly: “Double-shot oat milk latte, no sugar, medium foam, to go.” This time, it was exactly right.
The beans hadn’t changed. Neither had the barista. What changed was how she made her order clear.
That moment mirrors what many people are discovering in the world of AI. Large language models are powerful, but their results depend entirely on how we speak to them. This quiet craft of asking with clarity and intent is known as prompt engineering.
In the sections ahead, we’ll explore how this discipline is shaping the future of human-computer interaction. Through four essential lenses: goal, context, source, and expectation. Mastering prompts is not just about better answers, but better thinking.
The first stitch in any prompt begins with the goal. Not just what you want the AI to say, but what you’re hoping to accomplish. Are you trying to explore an idea? Summarize a report? Create something persuasive, informative, or emotionally resonant? Without clarity, even the most advanced AI will fumble.
One of the most common pitfalls I see is the assumption that the AI will “figure out” what they meant. But AI doesn’t infer emotion, nuance, or subtext the way humans do. It takes you quite literally. That means if you say, “Write something about our business,” you might get a generic paragraph that sounds like it was lifted off a template. It’s not wrong, just not useful.
Let’s take an example. Imagine you’re launching a new line of sustainable skincare. You want marketing content to describe your product line.
You type: “Describe our skincare products.”
The result? Likely generic and uninspiring.
Then, you try this: “Act as a content creator in cosmetic and beauty industry. Write a product description for our sustainable skincare line, focusing on ingredients that are native to Southeast Asia, cruelty-free certification, and the brand story around eco-conscious self-care. The tone should be warm and appealing to urban millennial women.”
The difference isn’t just in the words. It’s in the goal.
The second version isn’t just asking the AI to describe. It’s asking it to sell, connect, and carry a tone. You’ve gone from a mechanical command to a strategic design brief.
A good prompt goal answers these unspoken questions:
Once your goal is clear, the rest of the prompt begins to take shape more naturally. It becomes easier to decide what context to provide, what data to feed it, and what expectations to set.
The next layer of thought goes into context. This is where you give your prompt meaning beyond the words. Who is the output for? In what situation will it be used? What kind of world does it need to fit into?
If the goal is what, the context is why and for whom.
Here’s where a lot of prompts fall short. People often ask AI to “write a proposal,” “create a plan,” or “summarize a report” without specifying the audience or the setting. The result might be technically accurate but strategically off-mark.
Imagine telling a friend, “Can you help me explain this to someone?” and your friend replies, “Sure. Who are we talking to?” That’s what AI is silently asking too.
Let’s take a typical scenario. You’re preparing a business pitch. The bad prompt might look like this: “Create a pitch deck for our new fintech app.”
It doesn’t say who the audience is. Is it for venture capitalists? Bank regulators? University students?
Now compare that with:
“Create a 5-slide pitch deck for potential angel investors in Malaysia. Focus on how our fintech app solves the problem of digital payment fragmentation for small businesses. Include market size, product demo points, business model, and traction. The tone should be confident but not boastful.”
This prompt gives context: the audience, the problem space, the local relevance, and even the emotional tone. You’ve told the AI not just what to do, but why it matters.
Context also ensures the output speaks the language of your audience. A sales deck for enterprise software buyers must feel different from a TikTok ad script. A policy brief for a government committee should sound different from a community blog post.
Great prompt engineers learn to walk in the shoes of the reader before they even write the prompt. They ask: Who needs to understand this? What do they care about? What tone will earn their trust?
Even the best-structured prompt will produce poor results if it doesn’t have the right information to work from. AI doesn’t “know” your brand, your product, your dataset, or your unique circumstances unless you tell it.
This is where sourcing becomes critical.
Think of the AI like a brilliant intern who’s willing to help with anything but doesn’t have access to your internal files, notes, or backstory unless you hand it over. If you say, “Write our company’s vision,” and give it no material, it will guess. It might do an okay job, but it will fill in gaps with assumptions and sometimes, pure fiction.
Let’s consider someone who wants a company profile. They ask: “Can you write a profile for my company?”
That’s like telling a stranger to write your life story without any photos, journals, or conversations.
Now compare it to: “Using the content below from our About Us page and our CEO’s recent statement, write a company profile suitable for a sustainability-focused investor report. Keep it under 300 words and focus on our impact, growth, and mission.”
Here, the AI has something real to work with. You’ve pointed it to a source, clarified the purpose, and defined the scope. This is how you reduce hallucinations because AI is prone filling up the gap when it lacks detail.
In tasks involving data (especially summarization or analysis), pasting the source or uploading documents (when tools support it) is essential. Even for generative tasks like design inspiration or naming exercises, a source can be mood boards, taglines, past campaigns, or customer personas.
Good sourcing isn’t always about more. It’s about its relevancy.
You don’t need to give the AI your whole brand bible, just the right excerpt that helps it speak your truth, not invent one.
The last but equally crucial element of prompt engineering is setting expectations. This is where you define the form of the response. Even if the goal, context, and source are perfect, unclear expectations will dilute the output.
Think about asking someone for a recommendation. Do you want a single suggestion? A ranked list? A one-liner? A paragraph? A pros-and-cons comparison? Without telling them, they’ll default to whatever feels natural to them not necessarily to you.
AI works the same way. Left to its own devices, it may output in ways that surprise you. That’s ambiguity. Unless you are doing some intentional exploration, otherwise it’s always good to specify clearly what you want from AI.
Instead of just saying, “List our top 5 products,” say, “List our top 5 products in a table with columns for name, customer segment, average rating, and unique selling point. Keep the tone casual but informative.”
Or if you’re asking for analysis, instead of “Summarize this,” say, “Summarize this research article into three paragraphs: one on findings, one on implications, and one on next steps. Make it readable for a non-expert audience.”
Setting expectations means defining:
You don’t need to micromanage every detail but think like a director giving stage notes to a talented actor. The more clearly you describe the performance you want, the better the show will be.
Prompt engineering isn’t technical wizardry. It’s a soft but powerful skill that turns intent into impact. As AI becomes embedded in how we learn, communicate, write, plan, and even feel, it becomes more urgent for more women, especially, to take the lead in shaping how these tools are used.
Because prompting isn’t about coding. It’s about vision. It’s about communicating with clarity, empathy, and precision. These are skills women have long honed in the margins, often without recognition. Now, those very strengths are central to how we teach AI to speak, act, and think.
To prompt well is to lead well.
It’s to know what you want, understand who it’s for, give it what it needs to succeed, and set the stage for results that matter.
Here’s a quick guide on how to construct a meaningful prompt:
I want to generate a list of strategic growth opportunities (the goal). This is for a pitch to the board of a Malaysian logistics company that is exploring expansion into last-mile delivery for e-commerce clients (the context). Use the industry trend data pasted below (the source) and provide three opportunities, each explained in one paragraph. Use formal business language and include one market insight per point (the expectation).
If prompting is about asking better questions, maybe the real question is this: Are we training AI or is it training us to be more intentional, more empathetic, and more exacting in our thinking?
What began as a shortcut to content is becoming a mirror to our inner logic. When you prompt an AI well, you’re not just learning how to get answers. You’re learning how to ask better of yourself.
If the way we prompt reflects how we think, what does your prompting style say about you?