Three Wishes, Clearly Numbered

Opening Scene

A wisher who says “make things better” leaves everything to chance. A wisher who says, clearly and separately, “first, heal my grandmother’s illness; second, keep my family financially secure; third, do both without any unintended side effects” has done something genuinely different: broken one vague hope into distinct, addressable components. That same structural discipline, applied to a prompt, is one of the most reliable ways to reduce the ambiguity covered in Article 4.

In Plain English

Prompt structure separates a request into distinct, clearly labeled components — typically the actual instruction (what to do), relevant context (background information the model needs), and the desired output format (how the response should be structured). Rather than blending everything into one dense paragraph, structuring a prompt this way makes each component easier for the model to parse correctly, and easier for you to check for the ambiguity covered in Article 4.

The Old Way

Before structured prompting became standard practice, prompts were often written as single, unstructured blocks of text:

Early prompts frequently blended instruction, context, and desired format into one undifferentiated paragraph, making it harder for a model to reliably distinguish what was actually being asked from supporting background detail.
Users often repeated or buried the actual instruction within a long stretch of context, risking the model weighting the wrong part of the prompt most heavily.
There was no widely shared convention for how to organize a prompt clearly, leaving structure entirely to individual habit and intuition.

Structured prompting emerged as a genuinely practical response to the ambiguity and inconsistency that unstructured prompts often produced.

What’s Changing (and Why AI Is the Reason)

Widely shared conventions have emerged for structuring prompts clearly — often using headers, delimiters, or explicit labels to separate instruction from context from format — making structure a genuine, teachable skill.
Many production LLM APIs now offer explicit system, user, and assistant message roles, connecting directly to the system prompt concept covered in Article 8, providing built-in structural support beyond a single unstructured text block.
As prompts have grown longer and more complex, particularly for the multi-step tasks covered in Article 9, clear structure has become essential simply to keep a prompt manageable and debuggable.

The Metaphor, Fully Extended

The Genie’s Lamp	Prompt Structure Concept
One vague, blended hope for “things to get better”	One dense, unstructured paragraph blending instruction and context
Three wishes, clearly separated and individually addressable	A prompt clearly separated into instruction, context, and format
A genie who can address each distinct wish precisely	A model that can parse each distinct prompt component precisely
A wisher who’s learned to organize hopes into addressable requests	A practitioner who’s learned to organize prompts into clear components

For Beginners: What to Actually Do

Practice rewriting a dense, unstructured prompt into clearly separated instruction, context, and format sections.
Learn to use simple structural conventions — headers, bullet points, delimiters — to make a prompt’s components visually distinct.
Get comfortable using an LLM API’s system and user message roles where available, rather than blending everything into one message.

For Practitioners and Leaders: The Deeper Layer

Establish a standard structural convention for prompts across your team or organization, making prompts easier to review, debug, and maintain.
Recognize prompt structure as directly connected to prompt maintainability, particularly for prompts that will be revised or reused over time.
Invest in structural conventions especially for the more complex, multi-step prompts covered in Article 9, where clarity matters even more.

Quick Recap

Prompt structure separates a request into distinct components: instruction, context, and desired format.
Structuring a prompt this way reduces ambiguity and makes each component easier for the model to parse correctly.
Widely shared conventions and built-in API roles now support this practice directly.
Clear structure becomes increasingly essential as prompts grow longer and more complex.

Where This Fits in the Series

Article 5 covered organizing a wish into clear, addressable components. Article 6 covers a technique for making a wish’s intent unmistakable: showing the genie an example.