5 Power Techniques

These five techniques are the highest-leverage additions to any prompt. Learn these and you'll outperform 90% of AI users.

1. Chain-of-Thought: "Think Step by Step"

Adding "think step by step" or "reason through this before answering" dramatically improves AI performance on complex tasks — analysis, math, planning, debugging.

Without it: "Should we expand to the German market?" → Quick, surface-level answer

With it: "Should we expand to the German market? Think step by step: consider market size, competition, our current resources, regulatory requirements, and risks before giving a recommendation." → Structured, thorough analysis

Use it for: decisions, analysis, debugging, anything where reasoning matters.

2. Few-Shot: Give an Example

Show the AI one example of what you want before asking it to produce the real thing. This calibrates tone, format, and style better than any description.

Example:

"Here's an example of the style I want: 'Notion: your second brain, finally organized.' Now write 5 taglines for our project management tool in the same style. Punchy, benefit-focused, under 8 words each."

Use it for: writing style, formatting, tone, any output where you have a clear model in mind.

3. Role Assignment

Assigning a role tells the AI which "mode" to operate in. The more specific the role, the better.

• ❌ "Review my code"
• ✅ "You are a senior backend engineer with 10 years of experience. Review this Python function for security vulnerabilities, performance issues, and readability. Be direct and specific."

Use it for: technical reviews, specialized advice, when you need domain expertise.

4. Constraints

Constraints force the AI to make better editorial decisions. Without them, it defaults to safe, generic middle-ground output.

Useful constraints:

• "No jargon — if you must use a technical term, define it immediately"
• "No bullet points — write in flowing prose"
• "Maximum 100 words"
• "Do not use the words 'leverage', 'synergy', or 'innovative'"
• "Assume the reader has never heard of our company"

Use it for: whenever the default output feels too generic or too long.

5. Output Format Specification

Tell the AI exactly how to structure the response. Don't leave it to guess.

Examples:

• "Format as: Problem | Root Cause | Fix | Prevention"
• "Give me a table with columns: Feature | Benefit | Competitor Comparison"
• "Structure as: 1 headline, 2-sentence summary, 3 bullet points, 1 call to action"
• "Write this as a series of tweets, max 280 characters each, numbered 1/5, 2/5 etc."

Use it for: reports, comparisons, content with consistent structure.

Combining Techniques

The real power is combining these. A single prompt can use all five:

"You are a UX researcher [Role]. Review this onboarding flow and identify the top 3 drop-off risks. Think step by step [Chain-of-thought], working through each screen before drawing conclusions. Here's an example of the format I want: [Few-shot example]. Avoid UX jargon [Constraint]. Format your response as: Screen | Risk | Recommended Fix [Output format]."

Try It

Pick any prompt you've written and add at least two of these techniques. Run the comparison.