Prompt Engineering Fundamentals: Write Prompts That Actually Work

Why Prompt Engineering Matters

The difference between a mediocre AI response and a brilliant one is almost always the prompt. Prompt engineering isn’t just about asking nicely — it’s a systematic approach to communicating with AI models.

The CRAFT Framework

Use this framework for every important prompt:

• Context — Set the scene. Who are you? What’s the situation?
• Role — Give the AI a persona. “You are a senior DevOps engineer…” (this ties directly into system prompts, which we cover separately)
• Action — Be specific about what you want done
• Format — Specify the output format (table, list, code, JSON)
• Tone — Set the communication style

Example Using CRAFT

Context: I'm launching a SaaS product for small business accounting.
Role: You are a growth marketing strategist with 10 years of experience.
Action: Create a 90-day go-to-market launch plan.
Format: Present as a week-by-week timeline with milestones and KPIs.
Tone: Strategic and actionable, no fluff.

Essential Patterns

Few-Shot Learning

Show the AI examples of what you want:

Convert these product descriptions to compelling one-liners:

Input: "A 500ml stainless steel water bottle with double-wall insulation"
Output: "Keep your drinks ice-cold for 24 hours with our bulletproof steel bottle."

Input: "A wireless bluetooth speaker with 20-hour battery life"
Output: "Your soundtrack, untethered. 20 hours of wireless audio bliss."

Input: "A mechanical keyboard with cherry MX switches and RGB lighting"
Output:

Chain of Thought

Force step-by-step reasoning:

Solve this problem step by step. Show your reasoning at each stage before giving the final answer.

A company's revenue grew 15% in Q1, declined 8% in Q2, grew 22% in Q3, and grew 5% in Q4. If they started the year at $1M monthly revenue, what was their revenue at the end of Q4?

Constraint Setting

Set clear boundaries:

Write a technical blog post about Kubernetes autoscaling.

Constraints:
- Target audience: mid-level developers
- Length: 800-1000 words
- Include at least 2 code examples
- Avoid jargon without explanation
- End with 3 actionable takeaways

Common Mistakes

1. Being too vague — “Write something about AI” gives you generic content (we break down more of these in You’re Using AI Wrong)
2. Not specifying format — You’ll get walls of text instead of structured output
3. Ignoring iteration — The first response is a draft. Always refine.
4. Overcomplicating — Sometimes a simple, clear prompt beats an elaborate one

Practice Exercise

Try this prompt on your favorite AI model and compare the results:

I'm a freelance web developer building a portfolio site. Help me write the "About Me" section.

About me:
- 5 years experience
- Specialize in React and Node.js
- Based in Melbourne, Australia
- Passionate about clean code and accessibility

Requirements:
- 150 words max
- Professional but approachable tone
- Include a subtle call-to-action
- First person perspective

Prompt engineering is a skill that compounds. Each conversation teaches you what works and what doesn’t. Start applying these patterns today.

Related Articles

• Mastering System Prompts — The most powerful application of prompt engineering
• Getting Started with Claude — Practice these techniques with Claude
• You’re Using AI Wrong: 5 Mistakes That Kill Your Productivity — The anti-patterns to avoid