Prompt Engineering Essentials
shares
We’ve all had that moment: you ask an AI a question, and it responds with something vague, generic, or just off-base. It’s not that the model isn’t smart, it’s that we didn’t ask the right way.
That’s where prompt engineering comes in. Think of it as the art of talking to AI in a way that gets the best out of it. It’s not about being technical. It’s about being intentional.
What Is Prompt Engineering?
Prompt engineering is the practice of crafting your input (aka “prompt”) in a way that guides the AI to give you a better, more useful output. It’s less about tricking the model and more about setting it up for success.
You’re essentially telling the AI:
Who it should be (e.g., a travel agent, a senior software engineer)
What you want it to do
How you’d like the answer delivered
What the final outcome should be
The more context you give, the better it performs. Think of it like working with a talented but literal intern: if you’re vague, you’ll get a vague result.
Why Generic Prompts Fall Flat
Let’s look at a simple comparison. These are two prompts asking the model to write about climate change. One is generic, and the other gives it more structure and direction. The responses of the model have been reduced to about a third of the entire response, but you get the idea by just looking at how differently the AI approaches the task.
Prompt A:
“Write about climate change.”
Response:
Climate change refers to long-term shifts in global temperatures and weather patterns, primarily driven by human activities. The main cause is the increased concentration of greenhouse gases like […].
One of the most visible signs of climate change is the rise in global temperatures. Over the past century, the Earth’s average surface temperature has increased significantly, leading to more frequent and severe heat waves, droughts, and wildfires […].
The impacts of climate change are already being felt worldwide. Developing nations, which are often the least responsible for emissions, tend to suffer […].
Efforts to combat climate change focus on both mitigation and adaptation. Mitigation involves reducing emissions through renewable energy […].
The global community, through agreements like the Paris Accord, has committed to limiting global warming to well below 2°C above pre-industrial levels […].
Climate change is one of the defining challenges of our time. Addressing it is not only about protecting the environment but also […].
Prompt B:
“You are an environmental science teacher writing a 300-word blog post for high school students. Explain climate change clearly, using everyday examples, and end with a call to action.”
Response:
Understanding Climate Change: What It Means for You and the Planet
Climate change is one of the biggest challenges our world faces today–but what exactly is it? In simple terms, climate change means long-term […].
When we drive cars, use electricity, or produce goods in factories, we release greenhouse gases like carbon dioxide […].
You’ve probably seen some of its effects already: hotter summers, stronger storms, more wildfires, […].
But here’s the good news: we can still make a difference.
Start small–turn off lights when you’re not using them, ride a bike or take the bus […].
Climate change isn’t just something happening in faraway places–it’s affecting us here […].
What’s one change you can make this week to help fight climate change?
Which one do you think gives a more relevant, actually usable result? Exactly.
Models respond better when they’re given structure, tone, and purpose. Generic prompts don’t offer any of that. They leave the model guessing—and that usually means you get something bland, predictable, or off-target.
Different Models, Different Styles
Not all AI models respond the same way. Some excel at creative storytelling, others are tuned for technical tasks. But even within a single model, the style of your prompt matters.
For example:
Some models do better with step-by-step reasoning (“chain of thought”) when solving problems.
Others benefit from format-based prompts, where you define exactly how the output should look.
Larger models tend to pick up on subtle patterns more easily, but they still benefit from clarity.
By learning how to “speak their language,” you can dramatically improve the quality of what you get back.
The RISEN Framework
When you want a detailed, context-aware response from an AI, RISEN is one of the best ways to structure your prompt. It’s a five-part framework designed to help you guide the AI toward a useful, tailored output.
Here’s what it stands for:
R – Role
Start by assigning the AI a role. This sets expectations for tone, expertise, and perspective. Instead of just asking a question, you tell the model who it should be while answering.
“You are a senior data analyst…”
“Act as a UX writer with experience in mobile apps…”
I – Instruction
Give a clear task. What exactly do you want the model to do? This is your action request.
“…explain the results of an A/B test…”
“…write an onboarding copy for a budgeting app…”
S – Steps
Break the task into parts or give specific directions on how to do it. Think of this as scaffolding.
“Start with a short summary, then give three key takeaways, and close with a recommendation…”
“Use simple language, avoid jargon, and include a checklist at the end…”
E – End Goal
Why are you asking for this? What outcome do you want from the answer? Stating this helps the model align its choices with your purpose.
“The goal is to help a non-technical stakeholder understand the impact…”
“This copy will go on the first screen users see after downloading the app…”
N – Narrowing
Set constraints or boundaries—length, tone, audience, format, etc. This keeps the output focused.
“Use no more than 200 words, write in a friendly but professional tone…”
“Avoid technical terms and keep it suitable for high school students…”
When you put all of these together, the model gets a full picture of what you need—who it’s pretending to be, what it’s doing, how it’s doing it, what the end result should accomplish, and what rules to follow. Here’s a full example in action:
“You are a seasoned digital marketing strategist with 10+ years of experience working with small businesses.
Create a social media strategy plan for a new artisanal coffee brand launching in the U.S.
Break the plan into the following sections:
Brand positioning and audience targeting
Recommended platforms with justifications
Content themes and posting frequency
Engagement tactics and growth strategies
Metrics for tracking success
The goal is to deliver a professional strategy document the brand can use to launch and grow its online presence over the next 6 months. It should feel customized, actionable, and client-ready.
Keep the tone professional but approachable. Limit the strategy to platforms most relevant for consumer brands (e.g. Instagram, TikTok, maybe Pinterest). The entire response should be under 300 words.”
Response:
Artisanal Coffee Brand – 6-Month Social Media Strategy
1. Brand Positioning & Audience Targeting
Position the brand as a modern, ethically sourced artisanal coffee rooted in craftsmanship and lifestyle. Emphasize transparency, sustainability, and experience. Target health-conscious, design-savvy consumers aged 25–40, primarily urban dwellers with interests in specialty food, wellness, and slow living.
2. Recommended Platforms
Instagram: Visual storytelling, reels, and community building make it ideal for lifestyle branding.
TikTok: High discovery potential among younger audiences; use for behind-the-scenes, trending formats, and product education.
Pinterest (optional): Strong for driving website traffic and inspiring at-home coffee rituals.
3. Content Themes & Posting Frequency
Brand Story: Founder vision, sourcing journey – 1x/week
Product Features: Beans, brewing guides, bundles – 2x/week
Lifestyle & Aesthetic: Coffee moments, minimal design, café vibes – 2x/week
UGC/Community: Reshares, reviews, barista tips – 1x/week
Trendy/Reels/TikToks: Coffee hacks, POVs, seasonal drinks – 2–3x/week
Frequency: ~5 posts/week per platform, with 3–5 stories on Instagram daily.
4. Engagement Tactics & Growth Strategies
Launch giveaway with influencers and micro-creators
Comment daily on related niche accounts to boost visibility
Use trending audio and hashtags on TikTok
Host “Ask Me Anything” or “Coffee Talk” Lives monthly
Encourage tagged photos for UGC and reshares
5. Metrics for Tracking Success
Engagement Rate (likes, comments, shares)
Follower Growth Rate
Video Views & Watch Time (TikTok/Instagram Reels)
Website Clicks & Conversions (tracked via UTM links)
User-Generated Content Volume
Generic prompts get generic results. RISEN prompts get results that sound like they were written for you, not just by AI.
(Thanks to Kyle Balmer for creating and sharing this framework. Definitely give him a follow.)
Other Prompting Approaches That Work
While my personal favourite, RISEN isn’t the only way to guide an AI effectively. Depending on what you’re trying to achieve, there are a few other techniques worth knowing. These aren’t rigid templates—they’re more like habits you can build into your prompts when you want more useful, coherent responses.
RTF: Role, Task, Format
Think of RTF as the lightweight sibling of RISEN. It’s simple, fast, and especially useful when you want to get a well-structured answer without writing a long prompt.
Role: Who should the AI be?
Task: What should it do?
Format: How should the response be delivered?
Here’s a quick example:
“You are a hiring manager. Summarize this CV in bullet points and highlight red flags.”
RTF works because it answers the three things every AI model needs to know to stay on track: what perspective to take, what to do, and what shape the answer should have. When you don’t need all the details of RISEN, this is a great shortcut.
Chain of Thought
This one’s less about how you phrase your prompt and more about how the AI thinks through the problem.
Chain of Thought (CoT) is a technique that encourages the AI to reason step-by-step instead of jumping to a conclusion. You can trigger this by either:
Asking the AI to “explain your reasoning” or “think step-by-step”
Writing your prompt with examples that model step-by-step thinking
This works incredibly well for complex questions, logic puzzles, math problems, and even decision-making scenarios. It helps the model avoid hasty, shallow answers and instead walk through the problem in a structured way.
Without CoT:
“What’s the best time to post on Instagram?”
→ “Around 11am on weekdays.”
With CoT:
“What’s the best time to post on Instagram? Think step-by-step about user behavior, engagement trends, and time zones.”
→ “First, let’s consider your audience. If they’re mostly in the U.S., you’ll want to post during peak hours in those time zones. Next, think about engagement patterns: weekdays at mid-morning tend to perform well because people check their phones during breaks…”
The second answer isn’t just a guess—it’s a rationale.
Chain of Density
Chain of Density is about making a piece of writing richer with information, layer by layer.
It’s often used to revise or rewrite a draft by packing in more details, insights, or facts, without making the text longer. You can think of it as “upgrading” the density of meaning in a paragraph.
Say you start with:
“The company had a good quarter.”
Then you ask:
“Revise this to include more specific numbers, context about the industry, and mention any notable achievements.”
Now it might become:
“The company reported a 17% increase in revenue this quarter, outpacing competitors in the renewable energy sector and securing three major international contracts.”
Same length. Way more value.
This technique works best when you’re editing something you’ve already written, or when you want to take an initial AI draft and make it sharper, more informative, and more impactful.
Final Thoughts
If there’s one takeaway here, it’s this: AI is powerful, but it’s not a mind reader.
Prompt engineering isn’t about coding or complexity. It’s about clarity, context, and communication skills. When you give the model structure—like with RISEN or RTF—or encourage it to reason or revise—like with Chain of Thought or Chain of Density—you’re giving it the tools it needs to meet you halfway.
The good news? It’s a skill anyone can build.
Start small. Try giving the model a role. Add a bit of structure. Experiment with your phrasing. You’ll start to notice the difference right away.