RISE Framework: The Complete Guide with Examples

The RISE framework is one of the most widely-used prompt engineering frameworks, and for good reason. It's simple to remember, easy to apply, and dramatically improves results across almost any type of prompt.

What is RISE?

RISE is an acronym that stands for:

• Role - Who should the AI be?
• Instructions - What should it do?
• Steps - How should it proceed?
• Expectations - What should the output look like?

Each component addresses a common weakness in everyday prompts, ensuring you communicate clearly and completely with the AI.

Breaking Down Each Component

Role: Setting the Stage

The Role defines the persona, expertise, and perspective the AI should adopt. This is arguably the most impactful single element you can add to any prompt.

Why it matters: Different roles activate different "modes" in the AI. A marketing expert writes differently than a technical writer, even when given the same topic.

Strong role examples:

• "You are a senior UX researcher with 12 years of experience at FAANG companies"
• "You are a strict but encouraging high school calculus teacher"
• "You are a skeptical venture capitalist evaluating startup pitches"

Weak role examples:

• "You are an expert" (too vague)
• "You are helpful" (that's already implied)
• "You are a person" (adds nothing)

Instructions: Crystal Clear Tasks

Instructions tell the AI exactly what you want it to do. Vague instructions lead to vague outputs.

Strong instruction examples:

• "Write a 500-word blog post introduction that hooks readers with a surprising statistic"
• "Review this code for security vulnerabilities and rank them by severity"
• "Create a weekly meal plan with recipes under 30 minutes and under $50 total"

Weak instruction examples:

• "Write something good" (what is "good"?)
• "Help me with this" (with what specifically?)
• "Do the thing we discussed" (AI has no context)

Steps: The Roadmap

Steps break down complex tasks into manageable phases. This prevents the AI from jumping to conclusions and ensures thoroughness.

When to use steps:

• Multi-part deliverables
• Analytical tasks requiring research before conclusions
• Creative tasks that benefit from iteration

Example with steps:

Steps:
1. First, analyze the current market landscape
2. Then, identify our three main competitors
3. Next, compare their pricing strategies
4. Finally, recommend our optimal pricing position

Expectations: Quality Control

Expectations define what success looks like. Think of this as your acceptance criteria.

Expectation elements:

• Format (bullet points, paragraphs, tables)
• Length (word count, number of items)
• Tone (professional, casual, technical)
• Specific inclusions/exclusions
• Quality standards

Example:

Expectations:
- Format as a table with columns: Feature, Priority, Effort, Impact
- Include exactly 10 features
- Prioritize based on customer feedback data provided
- Exclude any features requiring new hires
- Use technical language appropriate for a dev team

Complete RISE Examples

Example 1: Business Strategy

ROLE: You are a McKinsey-trained business strategist with expertise in SaaS growth strategies.

INSTRUCTIONS: Analyze our customer churn problem and provide actionable recommendations.

STEPS:
1. Review the churn data I'll provide and identify patterns
2. Hypothesize the top 3 root causes based on the patterns
3. Validate each hypothesis with the data
4. Recommend specific interventions for each validated cause
5. Prioritize interventions by expected impact and effort

EXPECTATIONS:
- Start with an executive summary (3-4 sentences)
- Use data to support every claim
- Format recommendations as a prioritized table
- Include estimated impact for each recommendation
- Keep total length under 800 words
- Suggest specific metrics to track success

DATA:
[Insert your churn data here]

Example 2: Content Creation

ROLE: You are a content marketing specialist who has written for TechCrunch, Wired, and Forbes, specializing in making complex technology accessible.

INSTRUCTIONS: Write a blog post explaining how vector databases work for a non-technical audience.

STEPS:
1. Open with a relatable analogy that explains the core concept
2. Explain why traditional databases struggle with similarity search
3. Describe how vector databases solve this problem
4. Give 3 concrete use cases readers might encounter daily
5. End with a simple takeaway they can share with others

EXPECTATIONS:
- Target length: 800-1000 words
- Reading level: high school graduate (no jargon without explanation)
- Tone: curious and enthusiastic, like explaining to a smart friend
- Include one visual analogy per section
- Add a "TL;DR" summary at the end
- Avoid: code examples, mathematical notation, assumed prior knowledge

Example 3: Code Review

ROLE: You are a senior software engineer with expertise in Python security best practices and a background in security auditing.

INSTRUCTIONS: Review this Python code for security vulnerabilities and provide fixes.

STEPS:
1. Read through the entire code to understand its purpose
2. Identify all potential security issues (OWASP Top 10 focus)
3. Rank issues by severity (Critical, High, Medium, Low)
4. Provide fixed code for each issue
5. Explain why each fix is necessary

EXPECTATIONS:
- Format each issue as: Location → Problem → Severity → Fix
- Include the original vulnerable code snippet
- Include the corrected code snippet
- Explain the attack vector for Critical/High issues
- Suggest additional security improvements beyond the obvious
- Note any security trade-offs with the fixes

CODE:
[Insert code here]

Common RISE Mistakes

Mistake 1: Vague Roles

Bad: "You are an expert" Good: "You are a senior data scientist with 8 years of experience in NLP at Google"

Mistake 2: Multiple Instructions

Bad: "Write a blog post and also create social media content and design an email sequence" Good: Focus on one task, or explicitly break into sub-tasks within Steps

Mistake 3: Missing Steps for Complex Tasks

Bad: "Analyze my business and give recommendations" Good: Break analysis into: data review → pattern identification → hypothesis → validation → recommendations

Mistake 4: No Format Expectations

Bad: "Make it good" Good: "Format as H2 headers with bullet points, 800 words, professional tone"

When to Use RISE

RISE works best for:

• ✅ Task-based prompts with clear deliverables
• ✅ Professional and business contexts
• ✅ Content creation with specific requirements
• ✅ Analysis and recommendation tasks
• ✅ Code generation and review

Consider other frameworks when:

• ❌ The task requires complex reasoning (use Chain-of-Thought)
• ❌ You need to explore multiple solutions (use Tree-of-Thought)
• ❌ Context is more important than structure (use RACE)

Practice Template

Copy this template and fill in the blanks for your next prompt:

ROLE: You are a [specific role] with [relevant experience/expertise].

INSTRUCTIONS: [Clear, specific task statement]

STEPS:
1. First, [initial action]
2. Then, [next action]
3. Next, [following action]
4. Finally, [concluding action]

EXPECTATIONS:
- Format: [how to structure the output]
- Length: [word count or item count]
- Tone: [professional/casual/technical]
- Include: [must-haves]
- Exclude: [must-not-haves]

Next Steps

Now that you've mastered RISE, explore how it compares to other frameworks in our Complete Guide to Prompting Frameworks, or try the RACE Framework for context-heavy content creation.

Ready to apply RISE automatically? Try our Prompt Optimizer which can enhance any prompt using RISE principles.

Marcus Johnson is a Developer Advocate at PromptWizz, helping developers and professionals master prompt engineering.