Today, almost everyone is using AI in some form. Some use it to draft emails, others to write code, and many just explore it for fun or to clarify questions. Some even chat with AI about life goals. But here’s the catch: while millions are using it, very few truly understand how it works—or why it sometimes gives answers that sound smart but are completely wrong.

Imagine asking a stranger for directions. If you aren’t clear about where you want to go, they might confidently point you the wrong way. You could end up completely lost, even though the stranger sounded sure of themselves. AI works the same way. It isn’t magic—it’s a tool. And like any tool, its usefulness depends on how you use it. If you don’t know the right way to “talk” to it, you risk wasting time or getting misleading results.

That’s why learning the art of prompt engineering isn’t just a nice-to-have skill. It’s the key to unlocking AI’s true potential. Done right, prompt engineering transforms AI from a simple assistant into a powerful partner that can help you generate ideas, solve problems faster, and even surprise you. It has become a career path in its own right, and understanding it can give you a real edge—whether you’re exploring new opportunities or looking to level up your technical skills.

Have you ever received an AI answer that sounded confident but made no sense? Imagine how much more you could achieve if you knew exactly how to guide it.

What Is Prompt Engineering?

At its core, a prompt is simply the conversation you have with AI. Prompt engineering is the skill of designing these instructions so that AI gives you the output you actually want.

Think of it this way: if your prompt is vague, the output will likely be vague too. But if your instructions are clear, detailed, and well-structured, AI can generate responses that are useful, relevant, and reliable.

Prompt engineering isn’t about programming AI or changing how it works internally. It’s about giving clear instructions so the AI can help you solve problems faster, brainstorm creatively, or automate repetitive tasks without frustration.

Once you master prompt engineering, you’ll see AI differently. It’s no longer just a tool; it becomes a collaborator, one that can amplify your ideas and help you achieve more in less time.

Mental Model for Talking to AI

To get the best results from AI, assume it's your fellow teammate. Don’t rely completely on AI and trust that it gives you the best result on your first prompt. Instead, collaborate and give specific instructions until you are confident in the answer, just like you would with an assistant.

You may use this mental model to understand:

• Goals: What exactly do you want? Define your desired outcome clearly.

  • E.g.: “I want a 3-day travel itinerary for Bangkok that’s budget-friendly and family-safe.”

• Constraints: What boundaries should the AI respect? This could be word count, tone, style, or format.

  • Example: “Keep it under 300 words, in bullet points, and easy to read.”

• Examples: Give reference points or sample outputs if you want the AI to match a style or structure.

  • Example: “Write it like this sample itinerary I provide.”

• Format: Decide how you want the information structured: table, list, code snippet, paragraph, or JSON.

  • Example: “Provide a table comparing three hotels with price, distance from attractions, and guest rating.”

• Audience: Who will use or read this output? Tailoring your prompt for the audience ensures clarity and relevance.

  • Example: “Explain blockchain concepts to a high school student, not a tech gig.”

Vague Prompt: E.g., “Tell me about web development.”

Well-Structured Prompt (Using the Mental Model): You are a friendly coding instructor. Explain the basics of web development to a junior developer. Cover HTML, CSS, and JavaScript in 3 short sections, each with a brief example. Use simple language but include technical terms where necessary, and format each section with a heading and bullet points.

The Five-Principle Framework

The best way to prompt your LLM model is to use a five-principle framework. As a quick reminder when crafting your prompts:

• Task: Specify exactly what you want the AI to do.
• Context: Provide all necessary background and relevant details.
• References: Give examples, data, or style guides to guide the output.
• Evaluate: Check the AI’s output critically for accuracy, relevance, and tone.
• Iterate: Refine your prompt based on evaluation to get better results.

1. Specify the Task

Defining a clear task is the first step to writing a great prompt so the LLM can give you a clear response. A precise task helps the AI understand your goal. Include:

• Action: What do you want it to do? (Write, summarize, explain, generate)
• Audience or Persona: Who should it “act as”?
• Output Format: Paragraph, list, table, JSON, code block.

Example:

You are an experienced recruiter helping a developer land a new job. Your task is to write a compelling cover letter for a Backend Software Engineer role at a mid-sized retail software company. The letter must be approximately 300 words.

To make it more clear you can write it as follows:

• Persona: Act as a professional recruiter with deep knowledge of the software and retail industries.

• Task: Write a cover letter that highlights the developer's experience in the retail and e-commerce sectors. The letter should specifically emphasize their skills in backend development related to the retail industry.

• Structure: Format the letter into three distinct sections:

  • Introduction: Greet the hiring manager and state the purpose of the letter.
  • Body: Detail the developer's relevant experience and skills. Specifically mention their work with e-commerce platforms and other retail-related software.
  • Closing: Express enthusiasm for the role and a desire for an interview.

• Tone: The tone should be professional yet confident and persuasive.

2. Include Necessary Context

The more relevant details you include, the better your output will be. Providing background information—like your goals, the reason for the task, or what you’ve tried before—rounds out the model's understanding of what you're looking for. These details are called context, and they’re a vital part of creating great prompts.

Context can include:

• Purpose of the task
• Constraints or limitations
• Previous attempts or ideas
• Level of audience understanding

For example, you could rewrite the context for the above example as:

“The developer has 5 years of backend experience, primarily in e-commerce and retail platforms, and wants a cover letter that highlights their technical skills, project experience, and fit for a mid-sized retail software company.” Also highlight his database and cloud infrastructure experience.

Note: Context has the potential to be the longest piece of a prompt. One of the most powerful and reliable ways to provide an AI tool with context is to give the model specific reference materials to use.

3. Provide References

References are examples or any additional resources that guide an AI tool towards the output you want.

Depending on the AI tool you’re using, you can include text, images, and even audio references to sharpen your input. But simply pasting in a reference isn't enough. The key is to clearly label and structure your references so the tool knows exactly what they are and how to use them.

They can be:

• Text snippets (sample articles, product descriptions)
• Data tables or JSON
• Images or audio clips, etc.

Best Practices for Structuring References:

• Label references clearly using headings, XML-style tags, or markdown.
• Include only 2–5 relevant examples to avoid overwhelming the AI.

E.g.:

2 (Structure Example – Three Sections):

• Introduction: Greet and state purpose
• Body: Highlight experience and achievements
• Closing: Express enthusiasm and invite an interview

“Write a backend developer cover letter for a mid-sized retail software company. Use the structure and tone shown in the provided examples. Highlight experience in backend development, e-commerce platforms, and relevant technical skills.”

4. Evaluate Your Output

Different AI models are trained on unique data and rely on different programming techniques. Some models may be better suited to specific uses like writing code or brainstorming ideas, while others might have limited outputs because of their training sets. No matter the model, running the same prompt multiple times will likely render different results because of how AI tools process data.

That’s why it’s so important to evaluate your output. Before you use any AI-generated information, text, or materials, critically evaluate that the output is accurate, unbiased, relevant, and consistent before incorporating it into your workflows. If the output isn’t what you’re looking for, you should iterate on your prompt.

Evaluation ensures quality by checking the following:

• Is it accurate? Does the information make sense and is it factually correct?
• Is it relevant? Did the AI stay on topic and give you what you actually needed?
• Does it sound right? Does the tone and style match what you asked for?

After evaluating, you know exactly what to fix or improve.

5. Take an Iterative Approach

There will be times when your prompt simply isn’t leading to the output you need. That’s where our ABI advice comes in: Always Be Iterating. If you find an output lacking, continue clarifying what you need until it’s just right.

Iteration is where magic happens. Refine your prompts using feedback from the evaluation stage (stage 4) above.

First Prompt (Initial Attempt):

“Write a cover letter for a backend developer applying to a retail software company.”

Result:

• Output is generic, misses specifics about e-commerce experience.
• Tone is too casual.
• Achievements are vague.

Evaluation Findings:

• Missing measurable impact or examples.
• Not tailored to the retail domain.
• Tone not professional enough.

Refined Prompt (After Iteration):

“You are a professional recruiter. Write a 300-word cover letter for a backend developer applying to a mid-sized retail software company. Highlight 5 years of experience in backend development, specifically with Node.js and Python. Include achievements in e-commerce platforms, such as optimizing database performance and integrating payment gateways. Use a professional yet confident tone. Format the letter into three sections: Introduction, Body (experience and skills), and Closing (express enthusiasm and request an interview).”

Result:

• Output is structured in three clear sections.
• Highlights retail-specific experience and measurable achievements.
• Tone is professional and persuasive, suitable for a hiring manager.

Key Takeaways for Iteration:

• Refine the task: Make it more detailed or specific.
• Add context: Include relevant experience, tools, or skills.
• Use references/examples: Show the style or structure you want.
• Specify tone & format: Ensure it matches your audience and purpose.
• Repeat if necessary: Don’t settle for the first output; iterate until it’s polished.

Iteration transforms a vague or generic AI response into a precise, professional, and highly tailored output.

Prompting for Images

Tools like Midjourney and DALL-E have made AI image generation incredibly popular. Many people's first experience with generative AI is creating an image, so it's essential to cover.

Key Elements of an Effective Image Prompt:

• Subject/Content: What should appear in the image?

  • Example: “A Monkey eating a banana”

• Style: What artistic style or mood do you want?

  • Example: “A city with a park and people in a good mood and enjoying”

• Perspective & Composition: How should the scene be framed?

  • Example: “Bird’s-eye view showing skyscrapers and flying cars.”

• Details & Attributes: Add specifics to avoid vague results.

  • Example: “Include neon signs, reflective glass surfaces, and bustling streets.”

• Format/Resolution: Specify the output type if needed (square, portrait, landscape).

Mini Example:

Vague Prompt:

“Draw a cat.”

Result: Could be anything from a realistic cat to a cartoon cat, without control over style or mood.

Well-Structured Prompt:

“Create a digital illustration of a playful dog wearing a tiny hat, sitting on a stack of food, in a cozy, sunlit room. Use a warm, painterly style reminiscent of classic storybook illustrations.”

Result: Detailed, visually rich, and matches the intended style and context.

Tips for Image Prompting:

• Use adjectives and descriptive phrases to convey mood and style.
• Include context: location, action, or scenario.
• Reference famous styles or artists if you want a particular look.
• Combine with iteration: evaluate the image and refine your prompt for better results.

Prompting for images is a creative extension of text prompting—it’s about guiding AI to visualize your ideas with precision, style, and clarity.

Advanced Prompting Strategies

1. Zero-Shot Prompting

You ask the AI to perform a task without giving any examples—just a clear instruction.

E.g.: “Write a 300-word cover letter for a backend software engineer applying to a mid-sized retail software company. Highlight experience in backend development and e-commerce platforms.”

2. One-Shot Prompting

Provide one example of the type of output you want.

Example:

“Dear Hiring Manager, I am excited to apply for the Marketing Specialist role at XYZ Company. I have 3 years of experience managing digital campaigns and optimizing conversion rates…”

Instruction: “Now write a 300-word cover letter for a backend software engineer applying to a mid-sized retail software company. Highlight experience in backend development and e-commerce platforms.”

Result: AI mimics the style and tone of the example while tailoring it to the developer role.

3. Few-Shot Prompting

Provide a few examples to show structure, tone, and style.

Example 1:

“Dear Hiring Manager, I am excited to apply for the Marketing Specialist role at XYZ Company…”

Example 2:

“Dear Recruitment Team, I would like to express my interest in the Data Analyst position at ABC Corp…”

Instruction:

“Now write a 300-word cover letter for a backend software engineer applying to a mid-sized retail software company. Highlight experience in backend development and e-commerce platforms.”

Result: AI follows the pattern from multiple examples—maintaining consistent style, tone, and professional formatting.

4. Chain-of-Thought (CoT) Prompting

Ask AI to reason step by step before producing the final output.

Example:

“Write a backend developer cover letter for a mid-sized retail software company. First, outline the introduction, body, and closing. Then, list the key skills and experiences to highlight. Finally, generate the full 300-word cover letter using this structure.”

Result: AI explains its reasoning and structure first, then produces a well-organized and tailored cover letter.

5. Multimodal Prompting

Multimodal prompting is all about combining different types of inputs like text, images, and sometimes video, to get richer, more accurate AI outputs. Instead of relying solely on words, you can provide multiple sources of information that the AI can process together.

Human in the Loop: Ethical Considerations

As we've seen, prompt engineering is a powerful skill. But with great power comes great responsibility. The models we use are trained on huge amounts of data from the internet, and that data can contain biases, misinformation, and even harmful content. This means the AI can sometimes produce outputs that are biased, inaccurate, or just plain wrong.

This is where the final, crucial human element comes in. We call it “the human in the loop.”

Our job as a prompt engineer is not just to write a good prompt; it's to critically evaluate the output before using it. Always:

• Check for Bias: Does the AI's response reflect stereotypes or unfair assumptions?
• Verify Facts: Is the information accurate and verifiable from other sources?
• Ensure Fairness: Is the output appropriate and does it consider different perspectives?
• Protect Privacy: Never include sensitive personal information in prompts.

Remember, the AI is a tool, not an authority. Ultimately, we are responsible for the content generated and used. This critical oversight is the most important skill of all.

Common Pitfalls and Troubleshooting in Prompt Engineering

Common Pitfalls

• Hallucinations: AI sometimes makes up facts or adds details that aren’t true.

  • E.g.: Asking for a backend developer cover letter might lead the AI to invent a company name or project that doesn’t exist.

• Bias: AI reflects patterns from its training data, which can introduce unintended stereotypes or partiality.

  • Tip: Always review outputs for fairness and inclusivity.

• Vague or Ambiguous Prompts: Broad instructions like “Write something about software” often produce weak or generic results.

  • Tip: Fix by being specific—state the task, audience, and desired tone.

• Overly Complex Prompts: Packing too many instructions into one prompt can confuse the AI.

  • Tip: Break tasks into smaller, manageable steps when needed.

Troubleshooting Strategies

• Refine Your Prompt: Simplify or restructure your instructions. Add context, examples, or format requirements to guide the AI.
• Handle Hallucinations: Ask the AI to cite sources or limit responses to verified information. Cross-check outputs before using them in real applications.
• Address Bias: Include inclusive language guidelines in your prompt. Review outputs critically and revise if biased content appears.
• Iterate Step by Step: Generate multiple versions and compare outputs. Adjust wording, examples, or tone until the output matches your goals.

💡 Takeaway: Understanding these pitfalls and knowing how to troubleshoot turns prompt engineering from a trial-and-error process into a precise, reliable skill. Even a few minutes of iteration can drastically improve results, saving time and avoiding errors.

Conclusion: Practical Tips and Best Practices

1. Be Specific and Clear
2. Include Context
3. Use References and Examples
4. Evaluate and Iterate
5. Experiment with Prompting Techniques
6. Keep It Conversational (When Needed)
7. Don’t Overload the Prompt