Prompting AI Like a Pro
Mark CodewellJuly 28, 20253 min readProgrammingAI Generated
🧠 Prompting AI Like a Pro
📍 Part 1: Understand the Input-Output Contract
- Be explicit, not implicit. Say “Write a summary in three bullet points” instead of “Tell me about it.”
 - Include constraints. Word count, format, audience, tone—these help the AI shape its response.
 - Context is code. Referencing previous messages or providing background yields more tailored answers.
 
🛠️ Pro Tip: Use line breaks, lists, or labeled sections to structure complex prompts—AI handles semi-structured input like a champ.
  🧪 Part 2: Prompt Patterns That Work
| Type | Example | Best For | 
|---|---|---|
| Instructional | “Explain quantum computing like I'm 12.” | Teaching or simplifying topics | 
| Comparative | “Compare Python and JavaScript for web dev.” | Analysis and decision-making | 
| Creative Seed | “Give me three sci-fi world-building ideas.” | Ideation and inspiration | 
| Roleplay | “Act as a sarcastic history professor.” | Tone control and engagement | 
| Debugging | “Why does this code fail? # code” | 
        Problem-solving and clarification | 
🛠️ Pro Tip: Prefixing with “Act as…” often improves tone alignment and response depth.
  📦 Part 3: Packaging Prompts Like a Pro
- 🧹 Start simple. Use numbered steps or labeled goals.
 - 🧭 Give direction. Ask for opinions, strategies, or implementation details.
 - 🪚 Trim ambiguity. Replace vague verbs with specifics like “generate,” “summarize,” or “compare.”
 
🎯 Example Upgrade:
❌ “What’s the best framework?”
     ✅ “Compare Next.js and Nuxt.js for server-side rendering with developer experience in mind.”
⚙️ Part 4: Debug Your Prompts
- 🤖 Too verbose? Break into parts: one prompt for explanation, another for examples.
 - 🎭 Wrong tone? Add role cues like “You’re a startup founder pitching to VCs.”
 - 🔁 Messy output? Ask for “step-by-step” or “structured response with headings.”
 
🧠 Part 5: Lin’s Favorite Prompt Starters
💡 Act as a [role] and help me [task]
🔍 I want to understand [concept] through [example]
🧰 Improve this code’s readability without changing behavior:
📊 Summarize this data and suggest optimizations:
🎙️ Write a blog post titled “[Topic]” in a cheeky tone for engineers