Prompting Techniques
Improve AI Output with Five Techniques
Reach for these techniques when your prompt isn’t getting you what you need.
-
Role Prompting
When a task calls for specialized vocabulary, a professional’s focus, or a specific perspective, assign the AI a role.
Examples
- “You are a leadership coach who specializes in talent development.”
- “You are a customer success manager writing to a client.”
- “You are a skeptical investor reviewing my pitch.”
-
Few-Shot Prompting
When you need to follow an exact template or the format you want is hard to explain, give the AI examples of what you want.
Example
- “Format each name as Last, First:
Maria Chen → Chen, Maria
Jamal Brooks → Brooks, Jamal
Priya Patel → [AI fills in]”
- “Format each name as Last, First:
-
Chain-of-Thought
When the task has math, several steps, or a decision to check, add a phrase like “think step by step” and the factors to weigh.
Examples
- “Map the timeline step by step. Include the days between each deadline.”
- “Show your reasoning before the recommendation so I can check it.”
- “Check the quantities, rates, and totals in this invoice step by step.”
-
Output Formatting
When you want the output in a specific layout, tell the AI what format to use and how long to make it.
Examples
- “Format it as a numbered list.”
- “Put this in a table with columns for Price and Support.”
- “Give me the five key takeaways as bullet points.”
-
Meta-Prompting
When you’re not sure how to start or your first prompt came back generic, ask the AI to interview you or draft the prompt for you.
Examples
- “Before you start, ask me questions to fill in any gaps.”
- “Help me write a prompt to create a 30-day onboarding plan.”
- “Rewrite my prompt so the summary gives specific numbers, examples, and next steps.”