TL;DR

A prompt is your question or instruction to an AI. Good prompts are clear, specific, and give context. Small tweaks to how you ask can dramatically improve the quality of answers.

Why it matters

AI tools respond to what you ask—but they can't read your mind. Learning a few simple prompting patterns helps you get useful answers faster, with less frustration.

What makes a good prompt?

A good prompt has three ingredients:

  1. Clarity: Say exactly what you want
  2. Context: Provide relevant background
  3. Constraints: Specify format, length, or style

Example:

  • Weak: "Tell me about dogs."
  • Better: "Write a 3-paragraph summary of golden retriever care for first-time dog owners, covering diet, exercise, and grooming."

Pattern 1: Be specific

Vague prompts get vague answers. Narrow down:

  • What you want (a summary, a list, code, an explanation)
  • Who it's for (beginners, experts, kids)
  • How long or detailed
  • What style (formal, casual, technical)

Examples:

  • Instead of: "Explain AI"

  • Try: "Explain AI in 2–3 sentences for someone with no tech background."

  • Instead of: "Write code to sort a list"

  • Try: "Write Python code to sort a list of dictionaries by a 'price' key, in ascending order."

Pattern 2: Give context

The AI doesn't know your situation. Fill in the blanks:

  • Why you're asking
  • What you've tried already
  • What you know so far
  • What you need help with

Example:

"I'm writing a blog post about coffee for people new to brewing. I want to explain the difference between pour-over and French press in a friendly, non-technical way. Can you draft 2 paragraphs?"

Pattern 3: Use examples

Show the AI what you want by giving an example:

"Rewrite this sentence to be more concise: [your sentence]. For example, 'The reason why we are doing this is because...' becomes 'We're doing this because...'"

Or:

"Generate 3 headlines for an article about productivity tips. Style: punchy, actionable. Example: 'Stop Multitasking: The One Focus Trick That Actually Works.'"

Pattern 4: Ask for formats

Tell the AI how to structure the answer:

  • Bullet points
  • Numbered steps
  • Table
  • Code block
  • Markdown
  • JSON

Example:

"List 5 benefits of regular exercise. Format: bullet points, one sentence each."

Pattern 5: Iterate and refine

First answer not quite right? Refine:

  • "Make it shorter."
  • "Explain it like I'm 12."
  • "Add more detail about X."
  • "Rewrite in a formal tone."

Think of it as a conversation, not a one-shot query.

Pattern 6: Role-play

Ask the AI to take on a role or persona:

  • "You are a patient tutor explaining calculus to a high schooler."
  • "You are a senior software engineer reviewing this code."
  • "You are a friendly librarian recommending books."

This sets the tone and style.

Pattern 7: Break it down

Big tasks are hard for AI (and humans). Break them into steps:

  • Instead of: "Write a business plan."
  • Try:
    1. "What are the key sections of a business plan?"
    2. "Draft an executive summary for a coffee shop."
    3. "Now draft the market analysis section."

Pattern 8: Ask for evidence

If you need facts, ask the AI to explain its reasoning or cite sources (though it may not always have real citations):

  • "Explain why this is true."
  • "What evidence supports this claim?"
  • "Can you provide an example?"

This helps catch hallucinations.

Common mistakes to avoid

  • Too vague: "Tell me about X" → Specify what aspect of X
  • Assuming knowledge: The AI doesn't know your project or context unless you tell it
  • One-and-done: Treating it like a search engine instead of a conversation
  • Ignoring errors: If the answer is wrong, refine instead of giving up
  • Over-trusting: AI can sound confident even when wrong—verify important info

Prompt templates (copy and adapt)

For explanations:

"Explain [topic] in simple terms for someone who [context]. Focus on [specific aspect]."

For writing:

"Write a [length] [type of content] about [topic] for [audience]. Tone: [formal/casual/etc]. Include [specific elements]."

For code:

"Write [language] code to [task]. Requirements: [list constraints]. Include comments."

For summaries:

"Summarize this [article/document] in [N] bullet points. Focus on [key themes]."

For brainstorming:

"Generate [N] ideas for [topic/problem]. Make them [creative/practical/etc]."

Advanced tip: Chain of thought

For complex reasoning, ask the AI to "think step by step":

"Solve this problem step by step: [your problem]."

This encourages the AI to show its reasoning, which often leads to better answers.

Use responsibly

  • Don't share sensitive info in prompts (passwords, private data, trade secrets)
  • Verify facts before relying on them
  • Be ethical: Don't use AI to generate spam, misinformation, or harmful content

What's next?

  • Evaluating AI Answers: How to spot hallucinations and check accuracy
  • Prompting 201 (coming soon): Structured prompts, JSON output, and advanced techniques
  • How Chatbots Work: Understand what's happening under the hood