Artificial Intelligence (AI) is a powerful tool, but to get the best results, you need to ask the right questions in the right way. OpenAI's president, Greg Brockman, recently shared a game-changing framework for crafting the perfect AI prompt. Originally developed by Ben Hylak, this four-step method ensures clear, precise, and optimised responses from AI models.

Whether you’re using AI for research, content creation, or business insights, this structured approach eliminates vague responses and maximises efficiency. Let’s break it down.

The Four Pillars of a Perfect AI Prompt

1. Defining the Goal: Clarity is Key

The first and most crucial step in writing an AI prompt is to be specific about your objective. AI thrives on clarity—vague requests lead to vague answers.

Example Request (Too Vague):

❌ "Tell me about coffee."

This could lead to an overwhelming amount of generic information. Instead, make your goal clear:

✅ "I want a comparison of Arabica and Robusta coffee beans, highlighting their taste, caffeine content, and best brewing methods."

A well-defined goal ensures that the AI filters out unnecessary details and provides a focused, relevant response.

2. Specifying the Return Format: Get Responses the Way You Want

AI can present information in various ways—paragraphs, bullet points, tables, or even code snippets. If you don’t specify, you might receive an answer that doesn’t match your needs.

Example Request Without Format:

❌ "Compare Arabica and Robusta coffee beans."

This might result in a long, unstructured response. Instead, define the format:

✅ "Provide the comparison in a table format with three columns: Coffee Bean Type, Key Characteristics, and Recommended Brewing Method."

📌 Expected Output:

Coffee Bean TypeKey CharacteristicsRecommended Brewing MethodArabicaSmooth, mild, lower caffeinePour-over, French pressRobustaStrong, bitter, higher caffeineEspresso, instant coffee

By structuring your request, you save time and receive well-organised information.

3. Adding Warnings: Improve Response Accuracy

AI can sometimes generate outdated, misleading, or incomplete responses. Adding warnings helps refine accuracy and credibility.

Example Request Without Warnings:

❌ "List the top coffee-growing countries."

This may provide a general list without current data. Instead, add a warning:

✅ "List the top five coffee-growing countries based on 2024 production data. Exclude any outdated information."

Including a verification step ensures that AI double-checks details, reducing errors.

4. Context Dump: Personalise the Response

Providing context helps AI generate a response that aligns with your specific needs and preferences.

Example Request Without Context:

❌ "What’s the best coffee for beginners?"

This is too broad—"best" depends on personal taste, budget, and brewing method. Instead, refine it with context:

✅ "I’m new to coffee and prefer a mild, low-acidity option. Recommend three coffee types that are beginner-friendly and easy to brew at home."

With this added information, AI can suggest tailored recommendations rather than generic ones.

Putting It All Together: A Perfect AI Prompt Example

Let's combine all four pillars into a fully structured prompt:

👉 Goal: I want a guide on beginner-friendly coffee types with a focus on taste and easy brewing methods.

👉 Return Format:

• Coffee Name
• Flavor Profile
• Best Brewing Method
• Ideal for (e.g., mild taste, low acidity, rich flavor)

👉 Warnings: Ensure the coffee recommendations are widely available in 2025 and exclude rare or expensive varieties.

👉 Context Dump: I’m new to coffee and prefer a mild, smooth taste with minimal bitterness. I have a basic coffee maker and a French press at home.

With this structured prompt, AI delivers a relevant, personalised, and well-organised response.

Why This Framework is a Game-Changer

Using AI is not just about asking questions—it’s about asking them the right way. Whether you’re:

A marketer needing AI to generate SEO-friendly content

A student researching complex topics

A business owner using AI for data analysis

A coffee enthusiast looking for personalised recommendations

Applying the Goal, Return Format, Warnings, and Context Dump method will help you unlock better results, save time, and improve accuracy.

Final Thoughts: AI Works Best with Clear Inputs

AI is only as good as the prompts you give it. A poorly structured question leads to vague and unhelpful responses. But if you take a moment to refine your request using this four-step approach, you’ll receive clear, structured, and highly relevant answers.

So, next time you use AI, try this method—and watch the quality of responses skyrocket! 🚀