Writers already understand prompting better than they think. Every scene note, editorial memo, workshop question, query letter, and revision plan is a kind of prompt: a request shaped by intention.

The difference is that an AI system has no private understanding of your project. It does not know the emotional contract of your novel, the rhythm of your sentences, the standards of your genre, or what you meant to ask unless you put that information into the exchange. A vague prompt invites a vague answer. A well-built prompt gives the tool a narrow path to follow.

This lesson teaches a practical prompting method for writers: how to ask for help without surrendering taste, voice, or judgment.

What a Prompt Really Is

A prompt is the instruction you give an AI system. It can be a question, a task, a draft, a passage to analyze, a list of constraints, or all of those at once. Strong prompts usually do more than ask for an output. They define the assignment.

The writer’s prompt stack

PurposeWhat are you trying to accomplish?
ContextWhat does the AI need to know about the project?
RoleWhat kind of help should the AI provide: editor, reader, researcher, coach, or organizer?
MaterialWhat text, notes, facts, or examples should it use?
ConstraintsWhat should it avoid, preserve, or prioritize?
OutputWhat form should the answer take?

When a prompt is missing one of these pieces, the model fills the gap by guessing. That can be useful for brainstorming, but it is risky when you need precision. Prompting is the art of reducing unnecessary guessing.

The Basic Prompt Formula

Use this structure when you want dependable results:

You are helping me with [project type]. My goal is [specific outcome]. Use this context: [background, audience, genre, draft stage, constraints]. Here is the material: [paste text or notes]. Please produce [exact output format]. Preserve [voice, facts, names, structure, emotional intent]. Do not [invent facts, rewrite in another style, solve the wrong problem].

You do not need to use that formula every time. For quick work, a single sentence may be enough. But when the task matters, structure helps.

Weak Prompt, Stronger Prompt

The goal is not to sound technical. The goal is to be specific enough that the AI knows what kind of help you are asking for.

Weak

Make this better.

Stronger

Act as a developmental editor. Read the scene below and identify three places where the tension drops. Do not rewrite the scene yet. Give me concise notes with line-level examples and one revision option for each issue.

Weak

Give me ideas for my story.

Stronger

I am writing a contemporary literary mystery about two sisters cleaning out their late father’s house. Generate ten possible complications that reveal character rather than relying on violence, coincidence, or secret twins. Keep the tone grounded and emotionally restrained.

Weak

Rewrite this in a better voice.

Stronger

Analyze the passage for sentence length, diction, imagery, and emotional distance. Then suggest three revision strategies that would make the voice more intimate while preserving my original events and phrasing wherever possible.

Prompting for Writing Without Losing Your Voice

For writers, the most dangerous prompt is often the one that asks AI to “write it for me.” The result may be fluent, but it can also sand down the quirks, pressure, and specificity that make your work yours.

A better approach is to use AI as a reader and a drafting assistant, not as the final author. Ask it to diagnose, compare, outline, challenge, summarize, pressure-test, or generate options. Then make the final artistic decision yourself.

Useful rule: Ask AI for options, not authority. A good output should give you material to judge, not a decision to obey.

Seven Prompt Moves Writers Should Practice

1. Ask for diagnosis before revision

Instead of asking the model to rewrite a weak chapter, ask it to identify what is not working. This keeps you in the role of author and turns the AI into a critical reader.

2. Separate brainstorming from selection

Use one prompt to generate possibilities and another prompt to evaluate them against your goals. This prevents the first answer from becoming the plan by default.

3. Give the model your standards

Tell it what matters: psychological realism, narrative momentum, sensory specificity, subtext, scene causality, genre expectation, or line-level compression.

4. Specify what must stay unchanged

If a character name, plot fact, point of view, or tone is non-negotiable, say so. AI systems may otherwise “improve” the wrong thing.

5. Request a format you can use

A table, checklist, revision memo, beat sheet, reader report, or ranked list can be more useful than a long paragraph of advice.

6. Use examples when the pattern matters

If you want a certain kind of output, show the model a short example of the form. Examples are especially helpful for recurring tasks such as scene diagnostics, character summaries, newsletter blurbs, or revision checklists.

7. Iterate deliberately

Prompting is usually a conversation. After the first response, clarify what was useful, reject what was wrong, and narrow the next request.

A Reusable Prompt for Writers

Copy this framework whenever you want a controlled, craft-focused response:

Act as a careful writing coach, not a ghostwriter. I am working on: [project description]. My current goal is: [goal]. The intended reader experience is: [emotion, pace, genre expectation, effect]. Here is the passage or outline: [paste material]. Please do three things: 1. Identify what is already working. 2. Identify the biggest craft issue blocking the goal. 3. Give me three revision options, from light-touch to bold. Do not invent new plot facts. Do not imitate another living writer’s voice. Preserve my core premise and point of view.

What to Do When the Answer Is Wrong

Bad AI output is not always a sign that the tool is useless. Often it means the prompt did not give enough direction, the task was too broad, or the model guessed where it should have asked for more information.

Practice: The Three-Pass Prompt

Choose one paragraph, scene, poem, pitch, or essay draft. Then run three separate prompts:

Pass one: reader response

Read this as a first-time reader. What do you understand, what do you feel, and where do you become confused?

Pass two: craft diagnosis

Now read it as an editor. Identify the strongest craft element and the single most important revision opportunity.

Pass three: revision options

Give me three revision strategies. Do not rewrite the passage. Give me choices I can apply myself.

Notice the order. You are not asking AI to take over. You are asking it to help you see the work from different angles.

Key Takeaways

Coming next

Lesson 7: Explaining the Typical AI Interface

In the next lesson, we will slow down and look at the ordinary parts of an AI tool: the chat window, model picker, upload button, memory settings, image tools, voice mode, regeneration controls, and source panels. The goal is simple: make the interface feel less mysterious so writers can focus on the work.

Sources for this lesson

Next Lesson: 7 →
Continue to Week 7