ROSIE Framework

Overview

The ROSIE framework is a structured prompting approach designed for general-purpose AI interactions and one-off tasks. It provides a systematic way to craft prompts that ensure the AI understands the context, role, and requirements clearly. ROSIE stands for Role, Oneself, Specifics, Instructions, and Examples.

Framework Structure

The ROSIE framework consists of five key components that build a comprehensive prompt:

R - Role

Define the AI's role or persona for the task. This sets the context for how the AI should approach the problem and what expertise it should draw upon.

Example: "You are a customer service specialist with 5 years of experience in e-commerce support."

O - Oneself

Explain your business, organization, or personal context. This helps the AI understand your unique situation, industry, and operational environment.

Example: "I run an online boutique that specializes in sustainable fashion and eco-friendly home goods."

S - Specifics

Provide the specific data, context, and details about the current task. Include any relevant information, documents, or background that the AI needs to work with.

Example: Include the incoming email, customer details, or specific requirements for the task.

I - Instructions

Give clear, specific directions for what you want the AI to do. Be explicit about the desired output format, length, tone, and any constraints.

Example: "Craft a helpful email response that addresses the customer's concern, provides relevant information, and suggests next steps. Provide two response options."

E - Examples

Show the AI exactly what you want the output to look like. Provide sample inputs and corresponding outputs to demonstrate the format, style, and quality you're seeking.

Example: Include a sample email response showing the desired tone and structure.

Example Implementation

Here's a complete ROSIE framework prompt for crafting customer service email responses:

You are my customer service email specialist.

I run an online boutique that specializes in sustainable fashion and eco-friendly home goods, focusing on quality products that align with environmentally conscious values.

Here is what I was sent:
<Email>
Hi, I recently purchased a bamboo towel set from your store but noticed some quality issues. The fabric seems thinner than expected and there are some loose threads. Can you help?
</Email>

I want you to craft a helpful, empathetic email response in my voice that acknowledges the issue, offers a solution (like exchange or refund), and provides additional assistance. Respond with two options.

For example:
<ExampleResponse>
Thank you for reaching out and for your recent purchase. I'm sorry to hear you're experiencing quality issues with your bamboo towel set. We take product quality very seriously and want to make this right. Would you prefer an exchange for a new set or a full refund? Please let me know your preference and I'll process this immediately.
</ExampleResponse>

Another Example: Product Update Announcement

You are a product manager communicating with software users.

I lead product development at a SaaS company that provides project management tools for creative teams, emphasizing user-friendly design and powerful collaboration features.

Product update details:
<UpdateInfo>
New Feature: Advanced Timeline View
Release Date: Next Tuesday
Key Benefits: Better project visualization, drag-and-drop scheduling, milestone tracking
Potential Impact: May require brief loading time during first use
</UpdateInfo>

Write a professional product update email to our user base announcing this new feature. Highlight the benefits, explain any temporary impacts, and encourage feedback. Keep it concise and exciting.

Example format:
<ProductUpdate>
Subject: Exciting New Feature Coming to Your Workspace!

Dear [User],

We're thrilled to announce [feature name] launching [date]!

[Brief description and key benefits]

[Address any potential concerns]

[Call to action for feedback]

Best regards,
[Your Name]
Product Team
[Company Name]
</ProductUpdate>

Why It Works for AI

The ROSIE framework works effectively with AI because it addresses the core challenges of AI prompting:

Structured Context Building

  • Role setting gives the AI a clear persona and expertise level to emulate
  • Oneself explanation provides domain-specific context that shapes how the AI interprets information
  • Specifics ensure all relevant data is included, reducing ambiguity

Clear Expectations

  • Instructions eliminate guesswork by being explicit about desired outcomes
  • Examples provide concrete demonstrations of quality and format expectations

Reduced Hallucination Risk

By providing comprehensive context upfront, ROSIE minimizes the AI's tendency to make assumptions or generate incorrect information. The structured approach ensures the AI stays grounded in your specific business reality.

Consistent Quality

The framework creates a repeatable process that produces consistent, high-quality outputs across different users and scenarios within your organization.

When to Use ROSIE

ROSIE is ideal for:

  • One-off tasks that don't require frequent repetition
  • General AI interactions where you need high-quality responses
  • Complex tasks requiring detailed context and specific formatting
  • Business communication like emails, reports, or client updates
  • Situations where precision and personalization matter

Use ROSIE when you need to ensure the AI fully understands your business context and produces tailored, professional outputs for important communications or decisions.