This guide accompanies the Prompt Engineering video course. All frameworks, templates, and examples from the course are documented here.
LESSON 1: WHY PROMPT ENGINEERING MATTERS
Most people get mediocre AI results because they write mediocre prompts. Prompt engineering is the skill of communicating precisely with AI models to get professional-grade output.
The quality of your output is directly determined by the quality of your input.
Bad prompt: "Write me something about marketing." Good prompt: "You are a conversion copywriter with 10 years of experience. Write a 200-word product description for [PRODUCT] targeting [AUDIENCE]. Focus on the emotional benefit, not the features. End with a specific CTA. Avoid these overused words: 'innovative', 'game-changing', 'seamless'."
LESSON 2: THE ANATOMY OF A GREAT PROMPT
Component 1 — ROLE "You are a [SPECIFIC ROLE] with [SPECIFIC EXPERIENCE]." Examples:
- "You are a senior data analyst specializing in e-commerce metrics."
- "You are a personal trainer who works with busy professionals."
- "You are a legal assistant specializing in contract review."
Component 2 — CONTEXT "Here is the relevant background: [BACKGROUND]. The situation is: [SITUATION]. My goal is: [GOAL]."
Component 3 — TASK "Your task is to: [SPECIFIC, CLEAR TASK]. The output should: [WHAT IT NEEDS TO ACCOMPLISH]."
Component 4 — FORMAT "Format your response as: [NUMBERED LIST / TABLE / PARAGRAPH / STRUCTURED DOCUMENT]. Include these sections: [SECTIONS]."
Component 5 — CONSTRAINTS "Important: [LENGTH LIMIT]. [TONE]. [WHAT TO AVOID]. [WHAT TO INCLUDE]."
LESSON 3: THE 7 PROMPTING FRAMEWORKS
FRAMEWORK 1: ROLE-TASK-FORMAT (RTF) The simplest framework. "Act as a [ROLE]. [TASK]. Format: [FORMAT]."
Example: "Act as a marketing strategist. Create a 90-day marketing plan for a new product launch. Format: week-by-week action list."
FRAMEWORK 2: CHAIN OF THOUGHT (COT) Force step-by-step reasoning. "Think through [PROBLEM] step by step. Show your reasoning before giving the final answer."
Best for: Math, logic, complex decisions, multi-step analysis
FRAMEWORK 3: FEW-SHOT PROMPTING Give examples to show the desired output style. "Here are 3 examples of the output I want: Example 1: [EXAMPLE] Example 2: [EXAMPLE] Example 3: [EXAMPLE] Now create [NEW ITEM] in the same style."
Best for: Consistent tone, specific formats, creative styles
FRAMEWORK 4: TREE OF THOUGHTS (TOT) Explore multiple solution paths. "Generate 3 different approaches to [PROBLEM]. For each: describe the approach, pros/cons, and when it's best. Then recommend which approach best fits [MY SITUATION]."
Best for: Strategic decisions, complex problems, creative challenges
FRAMEWORK 5: THE SOCRATIC METHOD Use questions to generate better output. "I'm trying to [GOAL]. Before giving me a solution, ask me the 5 most important questions you need answered to give me the best possible advice."
Best for: Personalized advice, strategy, complex situations
FRAMEWORK 6: CONSTRAINTS FRAMEWORK Force creativity through constraints. "[TASK]. Constraints: must be under [LENGTH], must include [ELEMENT], must avoid [ELEMENT], must [REQUIREMENT], must target [AUDIENCE]."
Best for: Writing, content creation, focused problem solving
FRAMEWORK 7: PERSPECTIVE SHIFTING Get multiple viewpoints. "Analyze [TOPIC/DECISION] from 4 perspectives: (1) a [STAKEHOLDER 1], (2) a [STAKEHOLDER 2], (3) a [STAKEHOLDER 3], and (4) a skeptic. Then give your overall recommendation."
Best for: Business decisions, writing, balanced analysis
LESSON 4: ADVANCED TECHNIQUES
TECHNIQUE 1: PRIMING Set up the AI's mindset before the real request. "Before we start, I want you to know: [CONTEXT ABOUT YOUR SITUATION]. Your responses should always consider [KEY FACTOR]. Now, [YOUR REQUEST]."
TECHNIQUE 2: NEGATIVE PROMPTING Tell it what NOT to do. "Write [CONTENT]. Do NOT: use buzzwords, start sentences with 'I', use passive voice, include generic statements, or use any of these phrases: [LIST]."
TECHNIQUE 3: PERSONA CREATION Build a consistent character. "From now on, you are [PERSONA NAME]. Background: [BACKGROUND]. Personality: [PERSONALITY]. You always [BEHAVIORS]. You never [AVOID BEHAVIORS]. Respond to everything from this persona's perspective."
TECHNIQUE 4: META-PROMPTING Ask AI to help you write better prompts. "I want to [GOAL]. Help me write the best possible prompt to achieve this. Ask me any clarifying questions first."
TECHNIQUE 5: ITERATIVE REFINEMENT LOOP Build a feedback loop into your prompt. "[INITIAL REQUEST]. After you give me the output, ask me: 'What would you like to improve or change?' Continue refining until I say 'Final version.'"
LESSON 5: COMMON MISTAKES
Mistake 1: Being too vague Fix: Add role, context, specific task, format requirements
Mistake 2: Too much in one prompt Fix: Break complex tasks into a conversation
Mistake 3: Not iterating Fix: Every first draft is a starting point, not a final product
Mistake 4: Ignoring the format Fix: Always specify how you want the output formatted
Mistake 5: Not setting constraints Fix: Length, tone, what to avoid — constraints dramatically improve output
Mistake 6: Generic roles Fix: "Expert copywriter" < "Senior copywriter with 10 years of B2B SaaS experience specializing in email sequences"
LESSON 6: MODEL-SPECIFIC TIPS
ChatGPT (GPT-4):
- Excels at: Following complex instructions, versatility, plugins
- Use for: Most general tasks, content, coding, analysis
- Tip: Use System messages (Custom Instructions) to set persistent context
Claude:
- Excels at: Long documents, nuanced writing, following constraints
- Use for: Editing, long-form writing, analysis of large texts
- Tip: Give it the full context — it handles long prompts well
Gemini:
- Excels at: Google Workspace integration, real-time search
- Use for: Research, documents in Google Workspace, data
- Tip: Use with Google Docs/Sheets for seamless integration
QUICK REFERENCE: PROMPT TEMPLATES
Analysis: "Analyze [X]. Give me: key insights, main patterns, risks I might be missing, and a clear recommendation." Writing: "Write a [FORMAT] about [TOPIC] for [AUDIENCE]. Voice: [VOICE]. Constraints: [CONSTRAINTS]." Problem solving: "I'm facing [PROBLEM]. Context: [CONTEXT]. Think step by step. Give me 3 possible solutions and recommend the best one." Research: "Explain [TOPIC] at a [LEVEL] level. Include: key concepts, real examples, common misconceptions, and 3 practical applications." Editing: "Edit this for [CLARITY/CONCISENESS/IMPACT/TONE]. Show changes and explain why: [PASTE TEXT]"
Downloaded from DavideGPT.ai — Follow @DavideGPT for daily AI tips