Writing creative copy, scripts, or educational materials with an artificial intelligence model requires a high degree of control over voice, style, and structure. If you simply ask a model to write creative content, the output can often sound formulaic or generic.
To solve this problem, prompt engineers use the ROSES framework. ROSES stands for Role, Objective, Style, Examples, and Steps. This structure is specifically optimized for creative consistency, ensuring the output aligns with a specific brand voice or artistic tone.
Exploring the ROSES Components
Let us break down each letter of the ROSES acronym to understand how to build creative prompts.
1. Role
The Role establishes the creative identity. It describes the background, personality, and expertise of the writer. For example, are you a playful social media manager or a formal speechwriter?
2. Objective
The Objective defines the ultimate goal of the content. What do you want the audience to feel, think, or do after reading? The objective guides the direction of the narrative.
3. Style
The Style outlines the tone, voice, and rules of the content. This section controls the aesthetics. You can specify constraints such as using active voice, avoiding corporate jargon, or keeping sentences short and punchy.
4. Examples
Examples are the secret weapon of creative prompting. By providing one or two samples of your past writing, you show the model exactly how to mimic your cadence, vocabulary, and formatting. This is known as few-shot prompting, and it is highly effective for style matching.
5. Steps
The Steps outline the precise workflow the model must follow to assemble the content. Instead of generating the entire piece in one pass, you guide the model through a logical sequence, such as writing the hook, drafting the body, and adding a call to action.
When to Use the ROSES Framework
The ROSES framework is designed for creative tasks where maintaining a specific voice is critical.
We recommend using ROSES for:
- Writing brand copy and advertising messages.
- Crafting social media posts for platforms like LinkedIn, Instagram, or X.
- Drafting video scripts, podcast outlines, and audio briefs.
- Designing lesson content, tutorials, and storytelling narratives.
- Brainstorming campaign messaging and catchphrases.
A Practical Demonstration: Naive vs. ROSES
Let us look at how the ROSES framework transforms a video script prompt.
The Naive Prompt
Write a short video script about how prompts work.The Result: The response will likely be a standard, academic lecture. It may start with a dry introduction and lack the engaging hook necessary for a modern short-form video.
The Structured ROSES Prompt
Role: Friendly and energetic educational content creator.
Objective: Explain the core concept of prompt engineering in under sixty seconds to a general audience.
Style: Fast-paced, conversational, using simple analogies. Avoid complex terminology.
Examples:
"Hey everyone! Ever wondered why AI sometimes acts like it did not hear you? It is not the AI, it is your instructions."
Steps:
1. Start with a hook highlighting a common user frustration.
2. Introduce the concept of a prompt as a map for the AI.
3. Provide one actionable tip for improvement.
4. End with a quick call to action to subscribe.The Result: The artificial intelligence will write a script that matches the conversational example, follows the step-by-step outline, and maintains the energetic tone required for a short-form video.
Conclusion
The ROSES framework provides deep control over the style and flow of creative content. By integrating examples (few-shot learning) and sequential steps, you guide the model to accurately emulate your unique voice and construct structured, engaging outputs.