Prompt Patterns
Prompt Patterns
Reusable prompt templates that solve whole classes of problems — quickly and reliably.
What Are Prompt Patterns?
Prompt patterns are reusable prompt structures — much like design patterns in software engineering. Instead of reinventing a prompt every time, you apply a proven template that reliably solves a common type of task.
Once you learn a handful of patterns, you stop writing prompts from scratch and start composing them. This makes you faster, more consistent, and able to tackle complex workflows by combining patterns together.
A Simple Analogy
Think of a rubber stamp. You design it once, then press it onto page after page to get the same clean result every time. Prompt patterns are your reusable stamps for recurring AI tasks.
Core Patterns to Know
These six patterns cover the vast majority of everyday prompting needs.
Persona Pattern
Set expertise and tone.
Assign the model a role so its answer matches the right expertise and voice.
Template Pattern
Force a fixed structure.
Give the model an output layout with placeholders it must fill in exactly.
Recipe Pattern
Provide ordered steps.
Hand the model a numbered sequence of steps to follow in order.
Output Automater Pattern
Machine-readable results.
Request structured output such as JSON, CSV, or a table for downstream use.
Flipped Interaction Pattern
Let the model ask you.
Have the model ask you clarifying questions before it answers.
Fact-Check List Pattern
Surface the assumptions.
Ask the model to list the facts and assumptions its answer relied on.
Patterns in Action
Here's how three of the most useful patterns look as real prompts.
Template Pattern
Generate a meeting agenda using EXACTLY this template:
Title: [meeting title]
Date & Time: [date and time]
Attendees: [list of attendees]
Objectives:
- [objective 1]
- [objective 2]
Discussion Items:
1. [item] — [owner] — [time]
Action Items: [to be filled after the meeting]
Fill every placeholder. Do not add extra sections.Flipped Interaction Pattern
I want you to help me plan a study schedule.
Before you give me the plan, ask me one question at a time
to gather what you need (subjects, exam date, hours per day,
weak areas). Once you have enough, produce the schedule.Output Automater Pattern
{
"task": "Extract action items from the meeting notes below.",
"output_format": "Return a JSON array where each item has: task, owner, due_date",
"notes": "Ravi will send the report by Friday. Priya to book the venue next week."
}Why Use Prompt Patterns
- Save time on tasks you repeat often
- Get consistent, predictable results every time
- Combine patterns to handle complex workflows
- Build a personal template library you can reuse and share
- Lower the mental effort of prompting — just pick the right pattern
Pattern Quick Reference
| Pattern | Use It When You Want To… |
|---|---|
| Persona | Set the expertise, perspective, and tone of the answer |
| Template | Force the output into a fixed, predictable structure |
| Recipe | Make the model follow a defined sequence of steps |
| Output Automater | Get machine-readable output (JSON, CSV, table) |
| Flipped Interaction | Have the model gather requirements before answering |
| Fact-Check List | See the assumptions and facts behind an answer |
Common Pitfalls to Watch Out For
Key Takeaway
Prompt patterns turn one-off prompts into a reusable toolkit. Learn the six core patterns, combine them for complex jobs, and build a personal template library — you'll prompt faster and get consistent results every time.