RISEN Framework: Role, Instructions, Steps, End Goal, Narrowing

R.I.S.E.N. Framework

The RISEN framework is a five-component prompt engineering method that structures AI prompts around Role, Instructions, Steps, End Goal, and Narrowing. Developed by Kyle Balmer as an evolution of the simpler RISE framework, RISEN adds a critical "Narrowing" component that gives you precise control over constraints, tone, format, and scope. This structured approach aligns with best practices documented in OpenAI's prompt engineering guide and Anthropic's prompt engineering documentation.

Introduction

The R.I.S.E.N. Framework (Role, Instructions, Steps, End Goal, Narrowing) is a structured approach to prompt engineering designed for tasks that require multi-layered, detailed outputs. Where simpler frameworks handle quick requests well, RISEN excels when you need the AI to follow a specific process, work toward a defined outcome, and stay within clear boundaries.

The framework builds on the foundation of RISE (Role, Instruction, Specifics, Examples) by replacing the "Specifics" and "Examples" components with three more actionable elements: Steps, End Goal, and Narrowing. This shift moves the framework from describing what you want to directing how the AI should work.

RISEN produces outputs that are:

• Role-Grounded - Anchored in domain expertise and professional perspective
• Task-Focused - Directed by a single, clear instruction
• Process-Driven - Guided through a numbered sequence of actions
• Outcome-Oriented - Aimed at a defined success criteria
• Precisely Scoped - Bounded by explicit constraints on format, tone, and length

The RISEN framework is particularly valuable for:

• Complex multi-step tasks like business plans, research reports, or technical documentation
• Projects where the AI needs to follow a specific methodology or workflow
• Situations requiring strict control over output format, length, or tone
• Collaborative tasks where outputs feed into larger workflows
• Any request where a vague prompt would produce an unusable result

Origin & Background

The RISEN framework was created by Kyle Balmer, who introduced it through his promptentrepreneur content on TikTok and LinkedIn. Balmer developed RISEN as a practical evolution of the RISE framework, recognizing that while RISE provided a solid foundation with Role, Instruction, Specifics, and Examples, it lacked an explicit mechanism for setting boundaries on AI output.

The problem RISEN solves:

Anyone who has used AI tools knows the frustration of getting a response that is technically correct but practically useless: too long, wrong tone, missing the point, or wandering off-topic. RISEN addresses this by separating the what (Instructions), the how (Steps), the why (End Goal), and the boundaries (Narrowing) into distinct components.

Why the five components work together:

• Role activates the right knowledge domain and professional vocabulary
• Instructions focus the AI on a single, clear deliverable
• Steps provide a roadmap that prevents the AI from skipping important phases
• End Goal keeps every step aligned with the ultimate objective
• Narrowing eliminates the most common failure modes: wrong length, wrong tone, off-topic content

The evolution from RISE to RISEN:

The original RISE framework asked you to provide examples of desired output, which works well for short tasks but becomes impractical for complex requests. RISEN replaces that with a process-oriented approach: instead of showing the AI what the output should look like, you tell it how to get there and what constraints to respect. This makes RISEN particularly effective for tasks where you cannot easily provide an example of the final product.

Practical adoption:

RISEN has gained traction among business professionals, content creators, and developers who need repeatable prompt structures for complex tasks. Its five components map naturally to how you would brief a human colleague: tell them who they are (Role), what to do (Instructions), how to do it (Steps), what success looks like (End Goal), and what to avoid (Narrowing).

How R.I.S.E.N. Compares to Other Frameworks

When to choose RISEN:

• Your task has multiple phases or requires a specific workflow
• You need tight control over output length, format, and scope
• The AI keeps producing off-topic or unfocused responses to simpler prompts
• You are building prompts for repeated use in business workflows
• The task is complex enough that a four-component framework feels insufficient

When to use something else:

• For quick, simple tasks where five components add unnecessary overhead, use A.P.E.
• When strict guardrails and compliance are the priority, use TAG
• When audience and tone are the most important factors, use CO-STAR
• For professional consultation where role depth matters most, use R.A.C.E.

R.I.S.E.N. Framework Structure

1. Role

Define the persona, expertise, and perspective the AI should adopt

The Role component tells the AI who it should be. A well-defined role activates relevant domain knowledge, professional vocabulary, and methodological thinking. The more specific the role, the more specialized and useful the output.

Good examples:

• Senior data engineer with 10 years of experience building ETL pipelines for healthcare companies
• Certified financial planner specializing in retirement strategies for self-employed professionals
• Technical content strategist who has managed developer documentation for Fortune 500 companies

Bad examples:

• Expert (no domain specified)
• Writer (no specialization or context)
• Business person (vague and unhelpful)

2. Instructions

State the main task clearly and directly

Instructions should be a single, focused directive that tells the AI exactly what to produce. Think of this as the headline of your request. Keep it specific enough that there is no ambiguity about the deliverable.

Good examples:

• Create a 90-day onboarding plan for new engineering managers
• Write a competitive analysis comparing our pricing model to three direct competitors
• Draft a technical RFC for migrating our authentication system from session-based to JWT

Bad examples:

• Help me with onboarding (no specificity about what to produce)
• Do something about pricing (undefined deliverable)
• Fix our auth system (action without clear output format)

3. Steps

Provide a numbered sequence of actions for the AI to follow

Steps are the unique strength of the RISEN framework. By providing an explicit sequence, you control the AI's workflow and ensure it covers every important phase. Number your steps in logical order, and write each one as a clear action.

Good examples:

• 1. Research the top five competitors and list their key features. 2. Identify gaps in their offerings that align with our strengths. 3. Draft positioning statements for each gap. 4. Compile findings into a one-page executive summary.
• 1. Outline the three main sections of the document. 2. Draft each section with supporting data points. 3. Add a summary table comparing all options. 4. Write an executive recommendation paragraph.

Bad examples:

• Do the research and write it up (no sequence or specificity)
• Make it comprehensive (not actionable steps)
• Cover everything important (undefined scope)

4. End Goal

Define the ultimate objective and success criteria

The End Goal tells the AI what success looks like. This goes beyond the immediate deliverable to explain the purpose behind the request. When the AI understands the intended outcome, it makes better decisions at every step.

Good examples:

• The final document should enable our sales team to handle pricing objections with data-backed responses in real time
• This onboarding plan should reduce new-manager ramp-up time from 6 months to 3 months by giving them a clear weekly checklist
• The analysis should give our CTO enough information to make a go/no-go decision on the migration this quarter

Bad examples:

• Make it useful (subjective and vague)
• It should be good quality (undefined standard)
• Help us improve things (no measurable criteria)

5. Narrowing

Set constraints, limitations, tone, format, or scope boundaries

Narrowing is what separates RISEN from simpler frameworks. This component lets you eliminate the most common AI output problems by setting explicit boundaries. Think of it as a "do not" list combined with formatting requirements.

Good examples:

• Keep the document under 1,500 words. Use bullet points instead of paragraphs. Write in a direct, professional tone without marketing jargon. Focus only on North American markets.
• Limit the analysis to publicly available data. Format as a comparison table followed by a recommendation paragraph. Do not include speculative projections.
• Write at a 10th-grade reading level. Avoid acronyms unless defined. Include a glossary of technical terms. Output in markdown format.

Bad examples:

• Keep it short (undefined length)
• Make it professional (subjective constraint)
• Don't make it bad (not actionable)

Example Prompts Using the R.I.S.E.N. Framework

Example 1: Business Strategy

Prompt:

Example 2: Technical Development

Prompt:

Example 3: Content Creation

Prompt:

Best Use Cases for the R.I.S.E.N. Framework

1. Complex Business Documents

• Market analysis and competitive research
• Strategic planning and go-to-market strategies
• Business proposals and investment memos
• Quarterly reviews and board presentations

2. Technical Architecture and Planning

• System design documents and architecture decision records
• Migration plans and technical roadmaps
• Code review guidelines and engineering standards
• API documentation and developer guides

3. Content Production at Scale

• Blog posts and thought leadership articles
• Email sequences and marketing campaigns
• Training materials and course outlines
• Social media content calendars

4. Research and Analysis

• Market research reports
• Data analysis and interpretation
• Literature reviews and summaries
• Competitive intelligence gathering

5. Process Documentation

• Standard operating procedures
• Employee onboarding guides
• Customer success playbooks
• Incident response runbooks

When NOT to Use R.I.S.E.N.

RISEN is a powerful framework, but it adds overhead that not every task requires.

Skip RISEN for quick, simple requests. If you just need a definition, a code snippet, or a one-paragraph summary, the five components are overkill. Use A.P.E. (Action, Purpose, Expectation) for tasks that take one sentence to describe.Skip RISEN when strict compliance guardrails are the priority. If your main concern is preventing the AI from producing harmful, off-brand, or non-compliant content, TAG (Task, Action, Guardrails) puts guardrails front and center in a simpler structure.Skip RISEN for creative brainstorming. When you want the AI to explore freely and generate unexpected ideas, too many steps and constraints can limit creativity. Use a lighter framework or open-ended prompting instead.

Common Mistakes to Avoid

1. Writing Steps That Are Too Vague

Problem: Steps like "research the topic" or "analyze the data" give the AI no direction on depth, method, or scope.Why it matters: Vague steps produce vague outputs. The AI has no way to know what "analyze" means to you, so it defaults to a surface-level treatment.How to fix: Each step should specify what to do, how deeply, and what the output of that step looks like. Instead of "research competitors," write "identify the top five competitors by market share and list their pricing tiers, target audience, and key feature differentiators."

2. Skipping the End Goal

Problem: Providing Role, Instructions, and Steps but leaving out the End Goal, assuming the AI will figure out the purpose.Why it matters: Without an End Goal, the AI optimizes for completing the steps rather than achieving an outcome. You get a technically complete but strategically misaligned response.How to fix: Always state what the output should enable. "This plan should let the marketing team launch within 30 days" is far more useful than just listing what the plan should contain.

3. Under-Using the Narrowing Component

Problem: Treating Narrowing as optional or adding only one constraint like "keep it short."Why it matters: Narrowing is RISEN's biggest advantage over simpler frameworks. Skipping it means you lose the precision that makes RISEN worth the extra effort.How to fix: Include at least three constraints in your Narrowing component. Cover format (bullet points vs. paragraphs), length (word count or page count), tone (formal vs. conversational), scope (what to include and exclude), and audience (who will read the output).

Copy-Paste R.I.S.E.N. Template

Use this template as a starting point for any RISEN prompt. Fill in each section with your specific details.

Conclusion

The RISEN framework gives you a structured, repeatable method for getting high-quality outputs from AI on complex tasks. Its five components, Role, Instructions, Steps, End Goal, and Narrowing, map directly to how you would brief a skilled colleague: tell them who to be, what to do, how to do it, what success looks like, and what boundaries to respect.

Why RISEN works:

• The Steps component prevents the AI from skipping important phases of a task
• The End Goal keeps every step aligned with a meaningful outcome
• The Narrowing component eliminates the most common AI output failures
• The full framework produces outputs that are immediately usable in professional contexts

Choosing the right framework:

RISEN sits in the sweet spot between simpler frameworks like RISE and RACE (which work well for shorter tasks) and heavier frameworks like CO-STAR (which add audience and tone as separate components). If your task has multiple steps and you need tight control over the output, RISEN is the right choice.

For a deeper walkthrough with seven real-world examples and model-specific tips, read the complete RISEN framework guide. For a side-by-side comparison of all major frameworks, see 9 Best AI Prompt Frameworks in 2026.

Start with the copy-paste template above, adapt it to your first task, and refine your prompts from there. The more specific you are in each component, the better your results will be.