This is Hamamoto from TIMEWELL.
AI is available to nearly everyone now. The gap between people who get impressive results and those who get mediocre ones is not primarily a difference in which tool they're using — it's a difference in how clearly they communicate with it.
This article covers the fundamentals of prompt design: why structure matters, how to use Markdown formatting to improve AI comprehension, how to work iteratively toward the output you actually want, and how to build a personal prompt generator that makes strong prompting the default rather than the exception.
Why Prompt Structure Changes Results
The Instruction Problem
AI systems are designed to follow instructions — but vague instructions produce vague results. This is not different from the dynamic in human workplaces. If a manager says "write something good for the meeting," the person receiving that instruction doesn't have enough information to produce what's actually wanted. They produce something reasonable, and the manager is disappointed.
The same thing happens with AI. A prompt like "create a product proposal" will produce a generic response. A prompt that specifies the role, the context, the format, the audience, and the objective will produce something substantially more useful.
The key insight is that you are not just asking AI a question — you are designing the parameters within which it will generate a response. The quality of what comes back reflects the quality of those parameters.
Markdown as a Communication Tool
One of the most underutilized techniques for structuring AI prompts is Markdown notation — the same formatting used in documentation and note-taking applications.
Role assignment with #. Opening a prompt with a hash symbol creates a heading that tells the AI what role it should occupy. For example: # You are a skilled product manager at a beverage company. This anchors the AI's response to a specific perspective and knowledge base.
Hierarchical structure with ## and ###. Subheadings organize different sections of the prompt — background information, specific objectives, constraints, output format. When the AI can see the structure of what you're asking, it can address each element more precisely.
Priority marking with -. A hyphen before an item creates a list. Using this to enumerate the factors that matter most — or the components that must appear in the output — gives the AI a clear checklist rather than a vague direction.
Code fence and asterisk for precision. Backticks (```) can mark exactly where an instruction begins and ends. Double asterisks (**) signal emphasis on the most critical requirements. These tools reduce the risk of the AI misreading which part of a prompt is instruction versus context.
A Practical Example: Product Proposal
Consider a request to create a ten-slide product proposal for a new highball product, to be presented at an executive meeting.
A weak prompt: "Create a product proposal for a new highball."
A stronger prompt structure:
# Role: You are an experienced product manager at a beverage company## Background: The company is facing declining market share in the traditional highball category. We are launching a product targeting the 25–35 demographic.## Objective: Create a 10-slide executive presentation with slides covering market context, target customer, product concept, competitive differentiation, go-to-market strategy, financial projections, and risk assessment## Format requirements: Use data tables for financial projections. Each slide title should be an action-oriented sentence.## Supplementary information: [Insert any market data or customer research you have]
The output from the second prompt is not marginally better — it is categorically different. The AI has been given a specific perspective, a specific objective, a specific format, and specific constraints. It can work within those parameters rather than invent them.
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
Iterative Refinement: Prompting Is a Dialogue
First Drafts Are Starting Points
A common mistake is treating the first AI response as a complete answer. In most real work contexts, the first output from a prompt is a starting point for refinement — not a finished product.
This is not a limitation of AI. It is the same dynamic as any human drafting process. A consultant's first draft of a client report gets reviewed, commented on, and revised. An AI prompt should be treated the same way.
Example: News article summary
First prompt: "Summarize this news article in 300 characters."
AI produces a summary, but it captures the basic facts without addressing the broader social implications.
Second prompt: "Please redo the summary with particular emphasis on the social impact implications."
The result now includes the context that was missing. The second response was only possible because the first response revealed what needed to be added.
This iterative cycle — prompt, review, refine — is the actual workflow of effective AI use. Expecting a single prompt to produce a final output is like expecting a first draft to be publication-ready.
Building a Personal Prompt Generator
Once you've used AI regularly enough to recognize that structuring prompts takes time and cognitive load, the next step is to offload that work to AI itself.
The approach: use AI to build a reusable prompt structure for the types of work you do repeatedly.
In practice, this looks like: open a chat session and enter something like "I want to build a rough prompt maker. Give me a prompt template for achieving [your objective]." AI will generate a structured Markdown template — role, context, objective, format, constraints — that you can then save and reuse.
The output of this process is a personal library of prompt templates. When you need to write a product proposal, or create a market analysis, or generate a customer email, you're not starting from a blank page — you're starting from a structure that's already worked before, with small adjustments for the specific task.
Tools like Google Gemini have similar functionality built in, accessible from the "Run" tab in the interface.
This shift — from writing prompts from scratch every time to maintaining and improving a library of templates — changes the relationship between you and the AI from reactive to systematic.
The Mindset That Makes Prompt Design Work
You Are the Variable
AI tools are available to everyone with internet access. The quality of outputs varies enormously between users — not because the tool behaves differently for different people, but because different people are giving it different instructions.
This means the most important variable in your AI-assisted work is not which tool you're using but how clearly you think through what you want before you begin. Prompt design is fundamentally a thinking skill: clarifying your objective, organizing the information that's relevant, and specifying the format that will be useful.
Users who see disappointing results from AI and attribute it to the tool's limitations are typically missing this point. The more productive response to a disappointing output is to ask: what information was I not giving it? What constraint did I leave implicit that should have been explicit?
Improvement Is the Work
No prompt is final. The right mental model is iteration: each version of a prompt teaches you something about how to make the next version better. Over time, you develop intuition for what level of detail produces useful results, which types of instructions are too vague, and what kinds of tasks the AI handles reliably versus where it needs significant additional guidance.
This iterative investment pays returns in proportion to how much you use AI. For professionals whose daily work involves document production, analysis, or communication, the compounding benefit of improved prompting is significant.
Summary
Effective AI prompting is a learnable, improvable skill. The core techniques:
- Assign a role using a
#heading — this anchors the AI's perspective and knowledge - Provide background context explicitly — don't assume the AI knows your situation
- Structure your prompt with Markdown hierarchy (headings, bullet points) to separate objective, constraints, format requirements, and supplementary information
- Use backticks and asterisks to signal exactly where instructions begin and where critical requirements are
- Treat the first output as a draft — refine iteratively rather than accepting the first response as final
- Build a prompt library — use AI to generate reusable templates for the work you do repeatedly
The shift from expecting AI to read your mind to giving it clear, structured instructions produces results that are not incrementally better — they are qualitatively different. The investment required is mainly the discipline to specify what you actually want.
Reference: https://www.youtube.com/watch?v=3ZJJF_7UNSM
TIMEWELL AI Consulting
TIMEWELL is a professional team supporting business transformation in the AI agent era.
Services
- AI agent implementation: Business automation using GPT-5.2, Claude Opus 4.5, and Gemini 3
- GEO strategy consulting: Content marketing strategy for the AI search era
- DX acceleration and new business development: Business model transformation through AI
In 2026, AI is shifting from "something you use" to "something you work with." Let's think through your organization's AI strategy together.
Related Articles
- The Reality of a Part-Time Employee Who Worked Full-Time, Took Two Maternity Leaves, and Changed Her View of Work | TIMEWELL
- Before Paternity Leave — What You Absolutely Must Do to Take Leave Even During a Busy Period
- Pursuing a Hands-On Architecture Firm: Finding My Own Way as the 5th Generation of a Construction Company | Fujita Construction
