Building Reusable Prompt Templates

The highest leverage move in prompt engineering isn't writing a better one-off prompt — it's systematising the prompts that work so you never have to recreate them from scratch.

Why Templates Are Worth Building

Every time you write a prompt from scratch for a recurring task, you're doing wasteful work. You're also introducing inconsistency — the prompt you write on a Friday afternoon when you're tired is worse than the one you wrote on a Tuesday morning when you were fresh.

• A template:
• Captures the version that works before you forget how you wrote it
• Produces consistent quality regardless of when you use it
• Can be shared with teammates so everyone benefits
• Gets refined over time as you discover improvements

The Anatomy of a Prompt Template

A good prompt template has three parts:

1. 1.The stable scaffold — the role, format, and constraints that stay the same every time
2. 2.The variable slots — the parts that change with each use, clearly marked
3. 3.Optional parameters — elements you sometimes include and sometimes omit

Example template for writing a client email:

You are a professional [ROLE].

Write an email to [AUDIENCE] about [TOPIC].

Context: [RELEVANT BACKGROUND]

The tone should be [TONE]. Length: [LENGTH]

Do not include: [ANYTHING TO AVOID] ```

Useful Template Categories to Build

• Content creation templates
• Blog post first draft (with audience, angle, length, tone slots)
• Social media post (with platform, goal, character limit slots)
• Email newsletter section

• Analysis templates
• Document summary (with output format: bullets / narrative / exec summary)
• Pros and cons analysis
• Stakeholder analysis

• Communication templates
• Client update email
• Internal team announcement
• Meeting agenda and follow-up email

• Coding templates
• Code review (review this function for: performance, readability, edge cases)
• Documentation writing

Where to Store Your Templates

• The tool doesn't matter as much as the habit. Options:
• Notion or Obsidian for categorised libraries with descriptions
• A simple text file — often more useful than it sounds; low friction
• Custom GPTs or Claude Projects — store system prompts that pre-load when you start a session

The key behaviour: when a prompt works well, capture it immediately. Don't rely on memory.

Few-Shot Prompting: Teaching by Example

One advanced templating technique: providing examples of the output you want within the prompt itself.

Instead of describing what you want, you show it:

"Write a one-sentence product description for each item in the format shown in these examples: > - Noise-cancelling headphones: Block the world out. Focus on what matters. > - Standing desk: Your body wasn't designed to sit all day. Now it doesn't have to. > Now write descriptions for: [LIST OF PRODUCTS]"

Examples communicate style, tone, and format more precisely than descriptions can.

Module Summary

Prompt engineering is a learnable, improvable skill:

• Strong prompts include task, context, format, and constraints
• Role assignment steers the model's knowledge domain and tone
• Chain-of-thought prompting improves reasoning accuracy
• Templates systematise what works and reduce rework

The single most impactful thing you can do after this module: pick one recurring AI task you do every week, and build a template for it this week. That one template will pay back in saved time within a month.