Day 15: Build a Reusable Prompt

The Concept

Most people who use AI regularly have noticed they keep writing the same kinds of prompts over and over. They ask for feedback on writing. They ask for plans. They ask for explanations of things they need to understand quickly. Each time, they write the prompt from scratch, adjusting it slightly, and get results that vary more than they should. The effort compounds invisibly — and so does the inconsistency.

A reusable prompt template solves this by separating what stays the same from what changes. The fixed structure — the instruction pattern that produces a useful result — stays constant. The variable parts, the details specific to each situation, are marked with labelled placeholders in square brackets. Filling in the brackets takes ten seconds. Writing a new prompt from scratch takes several minutes and rarely produces something as well-constructed as a template you have refined over time.

The shift from one-off prompts to saved templates is one of the most practical upgrades available to regular AI users. It is also one of the least discussed, because it sounds administrative rather than exciting. But a prompt library is a genuine productivity asset — a collection of tools that get better the more you use and refine them.

From one-time prompt to repeatable tool

The process of converting a prompt into a template starts with recognising what made the original prompt work. Good prompts usually have a structure: they establish context, give a specific instruction, and specify the form of the output. That structure is the fixed part. The details — the topic, the audience, the specific content — are the variable parts.

When you ask AI to do the conversion for you, as today's prompt does, it identifies this distinction automatically and rewrites the prompt with placeholders. The result is a clean, titled template you can drop into any notes system or document. Seeing the filled-in examples the prompt requests helps you verify that the template actually generalises — that it produces something useful for different situations, not just the one you had in mind.

Anatomy of a good template

A good prompt template has four qualities. It is specific enough that the output is consistently useful, but general enough that the placeholders cover a real range of situations. It includes clear instructions that do not need to be remembered separately — everything the AI needs to do is contained in the template itself. The placeholders are labelled clearly, so you always know what kind of information belongs in each bracket. And it has a short memorable title so it is easy to find and use without reading the whole thing each time.

Templates that fail usually have one of two problems. They are too vague — essentially just a category label rather than a real instruction set — so filling them in still requires significant effort and the output is generic. Or they are too specific to the original use case and the placeholders do not cover the actual variation in how the prompt needs to be used. The template conversion prompt handles both problems by showing you filled-in examples, which expose immediately whether the generalisation is working.

Building your prompt library

A prompt library does not need to be elaborate. A folder in Notion, a document in your notes app, or even a dedicated section of a simple text file all work equally well. What matters is that the library is accessible when you are in the middle of work — not buried in a folder you never open — and that it is easy to add to.

The habit to build is simple: any time a prompt produces a notably good result, run it through the template conversion process and save it. Over a few weeks, you will accumulate a set of tools that reflect your actual work and the specific outputs you care about. ChatGPT, Claude, and Gemini all work equally well for both building templates and using them — the templates you create are not tied to any single tool.