If you’ve ever felt like your AI-generated output just isn’t hitting the mark, you’re not alone. Often, the issue isn’t the model itself—it’s how you’re prompting it. Knowing how to ask makes all the difference. That’s where prompt patterns come in.
In this guide, we’ll explore 10 essential prompt patterns that can dramatically improve the quality, relevance, and usefulness of responses from AI models like ChatGPT, Claude, or Gemini. For each pattern, we’ll define it, show a simple example, and explain the ideal use case.
1. Persona Pattern
What it is: Instructs the AI to take on a specific role or persona.
Example:
"Act as a seasoned travel guide. What’s the best 5-day itinerary for Japan?"
When to use it: Ideal when you want responses framed with domain-specific knowledge or tone. Great for simulating experts, teachers, or fictional characters.
2. Step-by-Step (Chain-of-Thought) Pattern
What it is: Encourages the AI to reason through the problem in logical steps before giving an answer.
Example:
"Let's work through this step by step. What's the square root of 144, and why?"
When to use it: Use for tasks that require logical progression, such as maths problems, analysis, or troubleshooting.
3. Few-Shot Pattern
What it is: Provides a few examples before asking the AI to continue the pattern.
Example:
"Q: What’s the capital of France? A: Paris Q: What’s the capital of Germany? A: Berlin Q: What’s the capital of Italy? A:"
When to use it: Best for structured outputs or training the model on format. Especially helpful when precision and tone are important.
4. Instructional Pattern
What it is: Gives the AI a clear, directive instruction.
Example:
"Summarise this article in three bullet points."
When to use it: Useful for straightforward commands—summarising, translating, converting formats, etc.
5. Comparison Pattern
What it is: Asks the AI to compare two or more items, concepts, or approaches.
Example:
"Compare the pros and cons of electric vs hybrid vehicles."
When to use it: Use this when decision-making or critical evaluation is needed.
6. Refinement Pattern
What it is: Requests the AI to improve or edit a piece of content.
Example:
"Here’s a paragraph. Make it more concise: [text]"
When to use it: Great for editing, rewriting, enhancing clarity, or changing tone/style.
7. Contextual Pattern
What it is: Provides a background context before issuing a prompt.
Example:
"You are writing a blog for beginner gardeners. Write a post on how to plant tomatoes."
When to use it: Essential when the model needs to tailor its response for a specific audience or scenario.
8. Critique Pattern
What it is: Asks the AI to evaluate or review something.
Example:
"Critique this business pitch and suggest improvements: [text]"
When to use it: Ideal for receiving constructive feedback, especially for writing, ideas, or presentations.
9. Multiple Perspectives Pattern
What it is: Instructs the AI to consider a topic from several viewpoints.
Example:
"Explain the benefits and risks of AI in education from the perspectives of teachers, students, and policymakers."
When to use it: Excellent for analysis, debate prep, or balanced articles.
10. Template Pattern
What it is: Uses a fill-in-the-blank structure to standardise AI outputs.
Example:
"Create a job description using this template: Role: [Job Title], Responsibilities: [...], Requirements: [...]"
When to use it: Useful for repeatable formats like resumes, job ads, emails, or social media posts.
Final Thoughts
Mastering prompt patterns can drastically improve the quality and consistency of your AI interactions. Whether you’re drafting content, solving problems, or brainstorming, these patterns give you a toolkit to make your prompts clearer and more effective.
