The Anatomy of a Good Prompt

L2 Lesson 1 of 5 — Intentional Prompting

1

2

3

4

5

Level 2

Intentional Prompting

Structure, techniques, and iteration — how to move from vague questions to precise instructions.

From Questions to Instructions

In L1, you learned how to have your first conversations with AI. Now comes the next paradigm shift: A good prompt is not a question — it’s an instruction. And like any good instruction, it has a clear structure.

Most people type a sentence and hope for the best. Professional users build their prompts deliberately — from five building blocks that together make the difference between “meh” and “exactly what I needed.”

The Five Building Blocks

Every strong prompt consists of up to five components. Not every prompt needs all five — but the more you use intentionally, the better the result.

Building Block	Question It Answers	Example
Role	Who should the AI be?	”You are a senior project manager.”
Context	What’s the situation?	”Our team is rolling out a new CRM system.”
Task	What exactly should it do?	”Create a risk assessment with the five biggest risks.”
Format	What should the output look like?	”As a table with columns: Risk, Likelihood, Mitigation.”
Constraints	What should it not do?	”No technical jargon, one page max, written in plain English.”

Role — Setting the Perspective

Role Prompting A technique where you assign the AI a specific role or expertise to shape the quality and perspective of its response. gives the AI a frame of reference. When you say “You are an experienced financial advisor”, it responds differently than without that framing — more domain-specific, more structured, with different vocabulary.

Context — Explaining the Situation

Without context, the AI guesses. With context, it understands what you actually need. This includes: background, target audience, prior work, or specific constraints of your situation.

Task — Naming the Job

The task is the core. It answers the question: What exactly should the AI do? The more precise the task, the less the AI has to guess.

Format — Specifying the Shape

Prose, table, bullet list, email, slide outline? If you don’t specify the format, the AI picks one — and it’s rarely the one you need.

Constraints — Setting the Boundaries

Constraints are what most people forget. They tell the AI what it should not do: no jargon, 200 words max, no speculation, only based on the data provided.

Before and After

Without structure:

Write something about risks when rolling out a CRM.

Result: A generic essay about CRM risks — long, broad, not tailored to your team.

With all five building blocks:

You are a senior project manager specializing in enterprise software rollouts. (Role)

Our 20-person sales team is switching next month from spreadsheet-based tracking to HubSpot. Most team members have little to no experience with CRM systems. (Context)

Create a risk assessment covering the five biggest risks for this transition. (Task)

Format: a table with the columns Risk, Likelihood (high/medium/low), Impact, and Recommended Mitigation. (Format)

Keep the language simple — no IT jargon. One page max. (Constraints)

Result: A ready-to-use risk table, tailored to your team and your situation.

The difference is not the length of the prompt — it’s the structure.

What Are System Prompt and User Prompt?

You may have come across these terms: A System Prompt A hidden instruction that defines the AI's behavior for an entire conversation — set before the actual exchange and typically not visible to the user. is an instruction set before the conversation that governs the AI’s baseline behavior. The User Prompt The visible input you send directly to the AI — your question, task, or instruction in the chat. is what you type into the chat.

In most tools, you only write the user prompt. But when you use the five building blocks, you’re essentially doing what a system prompt does: giving the AI a clear frame before it starts generating.

Try It

1. Pick a task from your workday — for example, drafting an email, writing a summary, or outlining a presentation.
2. Write a simple prompt first, without any structure. Note the result.
3. Rewrite the same prompt using all five building blocks (Role, Context, Task, Format, Constraints). Compare the results.
4. Experiment: What happens when you leave out the Role? What changes when you skip the Constraints?

Think Further

A good prompt has structure — not just content. The five building blocks are not a rigid template but a thinking tool. Not every prompt needs all five. But once you know what the building blocks are, you can make deliberate choices about which ones to include.

In the next lesson, you’ll learn specific prompting techniques that go beyond basic structure — from Zero-Shot to Chain-of-Thought.