​
1. Persona: Defining Who You Are (or Who You Want the AI to Be)

Sometimes, to get the best help, it’s useful to tell someone what kind of role they should take on. The same applies to AI! Defining a persona in your prompt means assigning a specific role to the AI. This helps the AI understand the tone, style, and expertise you’re looking for.

Imagine you need help planning a fun day trip. You could just ask, “Suggest some things to do near me.” But if you want more specific and exciting ideas, you might say, “Act as a friendly and enthusiastic local travel guide. I’m interested in outdoor activities and good food. Suggest three places to visit near me.”.

By assigning the role of a “friendly and enthusiastic local travel guide,” you give the AI a blueprint for how to respond. It will likely use more descriptive language, focus on local gems, and adopt an upbeat tone. You can ask the AI to be a book editor, a kindergarten teacher, a motivational speaker, or any other persona that fits your needs. Defining a persona helps the AI craft its responses to align with that specific role, leading to more relevant and informative output.

​
2. Task: Specifying What You Want

This might seem obvious, but clearly stating what you want the AI to do is crucial. Think of it as the main instruction you’re giving. Are you asking a question? Do you want the AI to write something? To summarize information? To brainstorm ideas?

A vague prompt like “Tell me about cats” might give you a very general overview. But if you have a specific task in mind, be direct. For example, instead of the vague prompt, you could say:

  • “What are the main characteristics of Persian cats?” (Asking a specific question)
  • “Write a short poem about a playful kitten.” (Requesting creative writing)
  • “Summarize the key points of the provided article about cat care.” (Asking for summarization)
  • “Brainstorm three potential names for a new cat breed that is known for its large size and gentle temperament.” (Requesting ideas)

Clearly phrasing your objective helps the AI understand what kind of response you’re expecting. Be as specific as possible about the action you want the AI to take.

​
3. Context: Providing Relevant Background

Imagine asking a friend for advice without giving them any background information. It would be hard for them to help, right? Providing context in your prompt is like giving the AI the necessary background information to understand your request fully. The more relevant context you provide, the better the AI can tailor its response to your specific situation.

Let’s go back to our day trip example. Simply asking a travel guide for suggestions is okay, but providing context makes it much better: “Act as a friendly and enthusiastic local travel guide. I’m visiting for the weekend, staying downtown, and I’m interested in outdoor activities and good food. I have two free days and prefer not to drive more than an hour outside the city. Suggest three places to visit.”

In this revised prompt, the context about the duration of the visit, accommodation, travel preferences, and time constraints helps the AI narrow down its suggestions and provide more useful recommendations. Context can include:

  • Specific details about your situation or needs.
  • Background information related to your task.
  • Any constraints or limitations (like time, budget, or location).
  • Existing information or resources the AI should consider. You can even provide text or images as context.

Providing adequate context gives the AI an idea of where to start and what kind of output would be most helpful to you.

​
4. Format: Structuring the Output

Think about asking someone for a list. Would you prefer it as a jumbled paragraph or neatly organized with bullet points? Specifying the format of the response you want from the AI can make the information much easier to use.

You can instruct the AI to provide its output in various formats, such as:

  • Bullet points or numbered lists: Useful for clear, concise information or steps.
  • Tables: Great for organizing data with rows and columns.
  • Short paragraphs: Suitable for explanations or descriptions.
  • Code blocks: Necessary for receiving computer code.
  • Specific structures: You can even define a specific structure, like asking for output in JSON format with particular fields.

For example, instead of just asking “List the ingredients for chocolate cake,” you could prompt: “List the ingredients for a classic chocolate cake in a bulleted list.”. This ensures the AI provides the information in an easy-to-read format. Being clear about the desired length and format of the response helps the AI deliver output that is structured and useful for your needs.

​
5. The Role of Examples in Prompts

Sometimes, the best way to explain what you want is to show it. Providing examples in your prompt can be incredibly powerful, especially when you have a specific style or structure in mind. This technique is known as few-shot prompting because you’re giving the AI a few examples of the desired input-output pairs.

Let’s say you want the AI to write product descriptions with a humorous and enthusiastic tone. You could provide a few examples:

Prompt:

“Write product descriptions with a humorous and enthusiastic tone, like the examples below:

Example 1: Product: Self-Stirring Mug Description: Tired of your coffee just…sitting there? Our revolutionary Self-Stirring Mug brings the party to your morning brew! Just press a button and watch your beverage whirl into a vortex of deliciousness. Perfect for the lazy genius or anyone who appreciates a good caffeine cyclone!

Example 2: Product: Glow-in-the-Dark Socks Description: Say goodbye to boring feet! Our Glow-in-the-Dark Socks are the ultimate way to light up your life (and your ankles). Whether you’re navigating a dark room or hitting the dance floor, these socks will make sure you’re seen. Warning: May cause spontaneous disco parties.

Now write a description for: Product: A foldable travel toothbrush”

By providing these examples, you show the AI the style, tone, and even the length of the descriptions you’re looking for. The AI can then learn from these examples and apply a similar approach to the new product. Examples are a great way to clarify your expectations and guide the AI towards a specific kind of output when words alone might not be enough.

By understanding and effectively using these five core components – Persona, Task, Context, Format, and Examples – you’ll be well on your way to crafting prompts that unlock the true potential of AI and get you the results you need. Remember that prompt engineering is often an iterative process; you might need to experiment and refine your prompts to achieve the best outcomes. In the following chapters, we will explore various techniques and best practices to further enhance your prompting skills.