RISEN Framework
Five components that give AI a complete mission brief. RISEN goes beyond role and task to include explicit step sequencing, success criteria, and scope constraints — ensuring focused, well-bounded responses every time.
Origin: RISEN was developed by Kyle Balmer in 2023 as a community framework for structured prompting. The acronym stands for Role, Instructions, Steps, End Goal, and Narrowing. Its distinguishing feature is the “Narrowing” element — explicit scope constraints that prevent the AI from going off-topic. This addresses a common pain point where AI responses become too broad or unfocused, wandering into tangential territory instead of staying within the boundaries of the task.
Modern LLM Status: RISEN remains highly practical for complex, multi-step tasks where scope control is essential. While simpler frameworks like RACE cover the basics, RISEN’s Steps and Narrowing components address two problems that persist even with modern LLMs: tasks that require a specific sequence of operations, and outputs that need explicit boundaries. The framework is especially effective for project planning, process documentation, and any scenario where “do not include” is as important as “do include.”
Tell the AI What Not to Do
Most prompting frameworks focus on what you want. RISEN adds an equally important dimension: what you do not want. The “Narrowing” component explicitly defines boundaries, exclusions, and constraints — turning an open-ended task into a precisely scoped assignment.
Consider asking an AI to “write a marketing strategy.” Without narrowing, you might get a 3,000-word document covering social media, email, paid ads, partnerships, events, and influencer campaigns. With RISEN’s narrowing — “Focus only on organic social media. Exclude paid advertising, email marketing, and offline channels” — you get exactly the focused strategy you need.
The five components work as a complete mission brief: who you are (Role), what to do (Instructions), in what order (Steps), what success looks like (End Goal), and what to stay away from (Narrowing). Together, they create a prompt that is both directive and defensive.
Paradoxically, constraining the AI’s scope often produces better and more creative results. When the model knows what is off-limits, it invests its full capacity into the defined territory rather than spreading thin across every possible angle. Narrowing is not about limiting the AI — it is about focusing it. The best outputs come from prompts that are simultaneously ambitious in what they ask for and precise in what they exclude.
The RISEN Process
Five components that create a complete, bounded mission brief
Role
Establish the AI’s identity and expertise domain. The role determines vocabulary, depth, perspective, and professional judgment. A “startup growth advisor” thinks differently than a “Fortune 500 CFO” even when addressing the same business question.
“You are a senior DevOps engineer with expertise in cloud infrastructure and CI/CD pipeline optimization.”
Instructions
Define the core task with a clear, specific directive. This is the “what” — the primary action the AI should perform. Use precise verbs: “audit,” “redesign,” “diagnose,” “create a migration plan.” Vague instructions like “help with” or “look into” produce vague outputs.
“Audit our current deployment pipeline and identify bottlenecks that are causing our release cycle to exceed 48 hours.”
Steps
Specify the sequence of actions or phases the AI should follow. This is what separates RISEN from simpler frameworks — instead of leaving the process to the AI’s discretion, you define the order of operations. This is critical for tasks where sequence matters or where you need to control the analytical approach.
“Step 1: Map the current pipeline stages from commit to production. Step 2: Identify the three longest-running stages. Step 3: For each bottleneck, analyze root cause. Step 4: Propose specific optimizations with estimated time savings.”
End Goal
Define what success looks like. The End Goal is not the format of the output — it is the measurable outcome or decision the output should enable. This gives the AI a north star: every recommendation, analysis point, or content element should contribute to achieving this stated objective.
“The end goal is a prioritized action plan that our team can implement within two sprints to reduce our release cycle from 48 hours to under 12 hours.”
Narrowing
Set explicit boundaries and exclusions. Narrowing tells the AI what to leave out, what topics to avoid, what assumptions not to make, and what constraints to respect. This is RISEN’s signature component — the guardrail that keeps responses focused and prevents scope creep.
“Do not suggest changing our cloud provider (we are committed to AWS). Exclude recommendations requiring additional headcount. Focus only on tooling and process changes achievable with our current team of four engineers.”
See the Difference
How narrowing transforms unfocused outputs into actionable plans
Without Narrowing
You are a marketing strategist. Create a marketing plan for our new SaaS product launch.
A sprawling 2,500-word plan covering social media, email, paid search, display ads, content marketing, influencer partnerships, PR, events, webinars, referral programs, and affiliate marketing — most of which is irrelevant to the actual budget and team capacity.
RISEN Prompt
Role: B2B SaaS marketing strategist.
Instructions: Create a 90-day launch plan.
Steps: 1) Define target personas, 2) Select channels, 3) Create content calendar, 4) Set KPIs.
End Goal: Generate 500 qualified leads in the first quarter.
Narrowing: Budget is $5K/month. No paid advertising. No events. Team is one marketer plus a freelance writer. Focus on content marketing and LinkedIn organic only.
A tightly scoped 90-day plan focused exclusively on content marketing and LinkedIn organic, with realistic timelines for a two-person team, budget-aligned tactics, and measurable KPIs tied to the 500-lead goal. Always verify market assumptions and projected conversion rates before committing to a strategy.
Practice Responsible AI
Always verify AI-generated content before use. AI systems can produce confident but incorrect responses. When using AI professionally, transparent disclosure is both best practice and increasingly a legal requirement.
48 US states now require AI transparency in key areas. Critical thinking remains your strongest tool against misinformation.
RISEN in Action
See how five components create precisely scoped, actionable outputs
Role: You are a senior product manager at a fintech startup.
Instructions: Write a feature specification for adding two-factor authentication to our mobile banking app.
Steps: 1) Describe the user problem and security need. 2) Define the feature requirements (functional and non-functional). 3) Outline the user flow from enrollment to daily use. 4) List acceptance criteria for QA. 5) Identify dependencies and risks.
End Goal: A spec document clear enough for our engineering team to begin implementation in the next sprint without additional clarification meetings.
Narrowing: Support only TOTP-based authentication (no SMS or email codes). Do not cover biometric authentication — that is a separate feature. Assume the user already has an account. Do not include cost estimates or timeline projections.
The Steps define exactly what sections the spec must include — preventing the AI from producing an incomplete document. The End Goal sets a quality bar: the output must be detailed enough to eliminate clarification meetings. The Narrowing prevents scope creep by excluding SMS, biometrics, cost, and timeline — keeping the spec laser-focused on TOTP implementation. Always have your engineering team review AI-generated specifications for technical accuracy before committing to development.
Role: You are a site reliability engineer leading an incident review.
Instructions: Write a blameless post-mortem for a 4-hour production outage caused by a database connection pool exhaustion.
Steps: 1) Summarize the incident timeline with timestamps. 2) Identify the root cause and contributing factors. 3) Detail the resolution steps taken. 4) Assess customer impact. 5) Propose preventive action items with owners and deadlines.
End Goal: A post-mortem that our VP of Engineering can share with stakeholders and that produces actionable follow-up tasks assigned to specific teams.
Narrowing: Maintain blameless tone throughout — no individual names or finger-pointing. Do not speculate about unconfirmed contributing factors. Exclude pricing or financial impact calculations. Focus on technical root cause and process improvements only.
The Narrowing enforces blameless culture — a critical organizational value that AI might violate without explicit instruction. The Steps ensure the post-mortem follows the standard format stakeholders expect. The End Goal ensures the output is executive-appropriate and action-oriented, not just a technical dump. Always verify incident details, timestamps, and technical root causes with your team before publishing a post-mortem.
Role: You are an instructional designer specializing in corporate onboarding programs.
Instructions: Design a two-week onboarding curriculum for new software engineers joining our platform team.
Steps: 1) Define learning objectives for each day. 2) Structure content into morning theory sessions and afternoon hands-on labs. 3) Include mentor check-in points. 4) Design an end-of-week assessment for each week. 5) Create a “30-day confidence checklist” for post-onboarding self-assessment.
End Goal: New engineers should be able to independently deploy a feature to our staging environment by the end of week two.
Narrowing: Assume engineers have 2+ years of experience (not entry-level). Do not cover HR policies, benefits enrollment, or company culture sessions. Focus exclusively on technical onboarding. Our stack is Python, Django, PostgreSQL, and Kubernetes — do not include content for other technologies.
Every RISEN component contributes to a tightly scoped output. The Steps define the curriculum structure (theory + labs + check-ins + assessments). The End Goal provides a measurable competency target. The Narrowing eliminates everything non-technical and locks the content to the actual tech stack — preventing generic onboarding content that wastes experienced engineers’ time. Always review AI-designed curricula with your technical leads to ensure accuracy and alignment with your actual codebase and processes.
When to Use RISEN
Best for complex tasks requiring explicit scope control and step sequencing
Perfect For
Tasks that require a specific sequence of operations — project plans, audit procedures, workflow designs, and any task where order matters.
When you need the AI to stay within explicit boundaries — budget constraints, technology limitations, team size restrictions, or topic exclusions.
Specifications, post-mortems, proposals, and reports where completeness and focus are both critical — every section must be present and nothing extraneous should intrude.
When every element of the output should contribute to a specific, measurable outcome — the End Goal component keeps the analysis purposeful rather than exploratory.
Skip It When
Writing a quick email or summarizing an article does not need five components — RACE or even a plain prompt is faster and equally effective.
Brainstorming, ideation, and creative exploration benefit from fewer constraints. RISEN’s narrowing can be counterproductive when you want the AI to range freely.
RISEN lacks explicit tone and audience dimensions. When the emotional register and reader profile matter most, CO-STAR is the better choice.
Use Cases
Where RISEN delivers the most value
Project Planning
Create scoped project plans with defined phases, success criteria, and explicit resource constraints that prevent unrealistic recommendations.
Technical Specifications
Write feature specs, architecture documents, and system designs with clear steps, measurable acceptance criteria, and technology constraints.
Compliance Audits
Structure audit procedures with defined scope, step-by-step assessment criteria, compliance targets, and explicit exclusions for out-of-scope areas.
Process Improvement
Diagnose workflow bottlenecks with structured analysis steps, targeted improvement goals, and narrowing that respects unchangeable system constraints.
Training Programs
Design curricula with sequenced learning objectives, competency milestones, and scope boundaries that match the audience’s actual skill level.
Incident Response
Create structured post-mortems and incident reports with defined investigation steps, resolution timelines, and blameless scope constraints.
Where RISEN Fits
RISEN bridges basic role-based prompting and full workflow orchestration
RISEN’s “Narrowing” component is its secret weapon. Most prompting frameworks tell the AI what to do. RISEN also tells it what not to do. In practice, this is often more valuable: a prompt that says “do not discuss pricing, competitors, or implementation timelines” produces a fundamentally different output than one that simply lists what to include. If your AI outputs keep going off-track or including unwanted content, adding narrowing constraints is often the fastest fix — even outside the RISEN framework.
Related Techniques & Frameworks
Explore complementary structured prompting approaches
Build Your RISEN Prompt
Structure your next prompt with all five RISEN components or find the right framework for your specific task.