The 4-Part Context Framework: How to Get 80% Quality AI Output Every Time

The Problem Most People Have with AI

You've tried ChatGPT or Claude. You got...generic garbage.

"Write me a legal brief" → Gets you high school essay quality

"Create a patient education handout" → Gets you WebMD copywriting

"Draft an architecture proposal" → Gets you buzzword soup

So you conclude: "AI isn't ready for professional work."

Wrong conclusion. AI is ready. Your prompts aren't.

The real problem: You're asking AI to read your mind instead of giving it context.

The solution: The 4-Part Context Framework.

What it does: Turns AI from "sometimes helpful" to "delivers 80% quality first drafts that only need your expert 20% polish."

Time to learn: 15 minutes Time saved: 5-10 hours/week, forever

Let's fix your AI prompts.

The 4-Part Context Framework

Every professional AI prompt needs exactly four elements:

Role: Who you are (your expertise level and domain)
Audience: Who will receive this (their context and needs)
Goal: What you need this document to achieve
Constraints: Specific requirements, limitations, preferences

Miss any of these → Bad output Include all four → 80% quality first draft

Let me show you how.

Part 1: Role (Who You Are)

Why this matters: AI needs to know what expertise level to write from.

Bad prompt:

"Write a contract"

AI thinks: Am I writing for a lawyer? A business owner? A student? What kind of contract? What jurisdiction?

Good prompt with Role:

"I'm a corporate attorney with 10 years experience in M&A transactions"

AI now knows: Write at attorney-level sophistication, use proper legal language, assume M&A context.

The Role formula:

I'm a [profession] with [X years] experience in [specialty/domain]

Examples by profession:

Lawyer:

"I'm a litigation attorney with 12 years experience in commercial disputes, practicing in federal courts in California"

Doctor:

"I'm a primary care physician with 8 years experience in family medicine, treating diverse patient populations"

Engineer:

"I'm a senior software architect with 15 years experience in distributed systems and cloud infrastructure"

HVAC Contractor:

"I'm an HVAC contractor with 20 years experience in residential and light commercial installations in hot climates"

Consultant:

"I'm a management consultant specializing in operational efficiency for mid-market manufacturing companies"

Why years of experience matters: AI adjusts language sophistication, assumes domain knowledge, and includes appropriate technical depth.

Part 2: Audience (Who Receives This)

Why this matters: Same content, different audiences = completely different language.

Example (explaining same medical condition):

To patient: "You have high blood pressure, which means your heart is working harder than it should. We'll start you on a medication to help your blood vessels relax."

To medical student: "Diagnosis: Essential hypertension, likely multifactorial etiology. Initial management: ACE inhibitor, target BP <130/80, reassess in 4 weeks."

To insurance company: "ICD-10: I10 - Essential (primary) hypertension. Treatment: Lisinopril 10mg PO daily. Medical necessity: Patient meets JNC-8 guidelines for pharmacological intervention."

Same diagnosis. Three completely different documents.

The Audience formula:

This is for [specific person/group] who [their context, knowledge level, concerns]

Examples:

Lawyer writing to judge:

"This motion is for Judge Martinez (Northern District of California), who tends toward narrow interpretation of precedent and values concise, fact-based arguments"

Doctor writing patient education:

"This handout is for a 65-year-old patient with 8th-grade health literacy who is worried about medication side effects"

Engineer writing for executives:

"This proposal is for the C-suite (non-technical) who care about business impact, risk reduction, and ROI, not implementation details"

HVAC contractor writing estimate:

"This estimate is for a homeowner (non-technical) who is most concerned about comfort and energy costs, currently paying $400/month in summer cooling"

Why audience matters: AI adjusts technical depth, addresses specific concerns, and uses appropriate language level.

Part 3: Goal (What This Should Achieve)

Why this matters: AI needs to know the document's purpose, not just its format.

Bad prompt:

"Write a memo"

AI thinks: Informational memo? Persuasive memo? Legal memo? Decision memo? What's the point?

Good prompt with Goal:

"This memo needs to persuade the client to settle rather than proceed to trial, emphasizing litigation risk and cost-benefit analysis"

The Goal formula:

The purpose is to [specific action or outcome you need]

Examples:

Lawyer (motion to dismiss):

"Goal: Convince the judge to dismiss the case for lack of standing based on 9th Circuit precedent"

Doctor (referral letter):

"Goal: Get the cardiologist to see the patient urgently (within 2 weeks) for suspected unstable angina"

Engineer (architecture doc):

"Goal: Get stakeholder approval for microservices migration by demonstrating reduced maintenance burden and improved scalability"

HVAC contractor (estimate):

"Goal: Win the job by showing value of recommended solution vs cheaper alternatives, emphasize comfort and energy savings"

Consultant (strategic recommendation):

"Goal: Persuade CEO to invest in process automation by showing 18-month ROI and competitive risk of not acting"

Why goal matters: AI knows whether to be persuasive, informational, technical, or educational. Same facts, different framing.

Part 4: Constraints (Specific Requirements)

Why this matters: This is where you get exactly what you need, not generic output.

Constraints include:

Format requirements
Length limits
Tone preferences
Specific elements to include/exclude
Deadlines or urgency
Style guides

The Constraints formula:

Requirements:
- Format: [specific structure]
- Length: [word/page count]
- Tone: [professional/aggressive/friendly/technical]
- Must include: [specific elements]
- Must avoid: [things to exclude]
- Style: [any specific preferences]

Examples:

Lawyer (legal brief):

Constraints:
- Format: Federal court brief format, proper bluebook citations
- Length: Maximum 15 pages
- Tone: Aggressive but professional, confident not arrogant
- Must include: Summary of argument, statement of facts, legal analysis, conclusion
- Must avoid: Emotional appeals, overstating precedent strength
- Cite only: 9th Circuit cases from last 10 years

Doctor (chart note):

Constraints:
- Format: SOAP note structure
- Length: Concise, focused on key clinical points
- Tone: Professional medical documentation
- Must include: Vital signs, chief complaint, assessment, plan, patient education provided
- Must avoid: Abbreviations not on approved list, subjective opinions
- Follow: Our clinic's documentation standards for billing compliance

Engineer (technical spec):

Constraints:
- Format: ADR (Architecture Decision Record) template
- Length: 2-3 pages maximum
- Tone: Technical but accessible to senior management
- Must include: Context, decision, consequences, alternatives considered
- Must avoid: Implementation details, code examples
- Focus: Business value and risk mitigation

HVAC contractor (estimate):

Constraints:
- Format: Professional estimate with itemized pricing
- Tone: Friendly but expert, no jargon
- Must include: Three pricing tiers (budget, recommended, premium), energy savings estimates, comfort improvements
- Must explain: Why recommended solution vs cheap fix
- Must avoid: Technical HVAC terms the homeowner won't understand
- Include placeholders: For photos I'll upload, for pricing I'll fill in

Why constraints matter: This is how you get AI output that's 80% ready instead of 40% ready.

Putting It All Together: Complete Examples

Example 1: Legal Brief for Motion to Dismiss

Bad prompt (most people start here):

"Write a motion to dismiss"

Result: Generic, unusable, needs complete rewrite

Good prompt using 4-Part Framework:

ROLE: I'm a commercial litigation attorney with 12 years experience, practicing in federal courts in Northern California

AUDIENCE: This motion is for Judge Sarah Martinez (ND Cal), who is known for narrow interpretation of precedent and prefers concise, fact-based arguments over emotional appeals

GOAL: Convince the judge to dismiss the case for lack of standing, based on plaintiff failing to demonstrate concrete injury per Spokeo v. Robins and subsequent 9th Circuit precedent

CONSTRAINTS:
- Format: Federal court motion format with bluebook citations
- Length: Maximum 15 pages
- Tone: Confident and professional, not aggressive
- Must include: Summary of argument, statement of facts, standing analysis under 9th Circuit precedent, conclusion
- Must avoid: Overstating our position, attacking opposing counsel personally
- Citation focus: 9th Circuit cases from 2020-2025, emphasize Supreme Court guidance from Spokeo and TransUnion
- Key argument: Plaintiff has no concrete injury, only statutory violation without harm

Based on the following facts: [brief case summary]

Draft the motion to dismiss.

Result: 80% quality brief that needs only your expert polish on precedent nuance and case-specific details.

Time saved: 6 hours

Example 2: Medical Referral Letter

Bad prompt:

"Write a referral to cardiology"

Result: Generic template that doesn't convey urgency or specific concerns

Good prompt using 4-Part Framework:

ROLE: I'm a primary care physician with 10 years experience in family medicine

AUDIENCE: This referral is for Dr. James Chen, a cardiologist I've worked with for years who is typically booked 4-6 weeks out but will prioritize urgent cases

GOAL: Get Dr. Chen to see this patient within 2 weeks (not 6 weeks) because I'm concerned about unstable angina requiring urgent evaluation

CONSTRAINTS:
- Tone: Professional colleague-to-colleague, convey appropriate urgency without overstating
- Length: Concise (cardiologists are busy), but include all relevant clinical details
- Must include: Patient demographics, chief complaint, relevant cardiac history, recent workup results, specific question I need answered
- Must convey: I'm concerned about unstable angina pattern, patient has high risk factors, but I'm not calling 911-level emergency
- Format: Standard referral letter
- Request: "Would appreciate seeing within 2 weeks if possible given symptom pattern"

Based on this patient:
- 58-year-old male
- Chief complaint: Intermittent chest pressure, worse with exertion over past 3 weeks, now occurring at rest
- History: Hypertension, hyperlipidemia, family history of early MI (father at age 55)
- Exam: Unremarkable today, vitals stable
- Recent labs: Total cholesterol 240, LDL 165
- EKG: No acute changes, old inferior Q waves
- My concern: Pattern suggests possible unstable angina, needs stress test or cath

Draft the referral letter.

Result: Referral that gets patient seen in 2 weeks, not 6, because it conveys appropriate urgency with professional judgment.

Time saved: 15 minutes per referral, clearer communication

Example 3: HVAC Estimate (Trades/Contractor)

Bad prompt:

"Write an estimate for HVAC repair"

Result: Generic template that doesn't address customer concerns or win jobs

Good prompt using 4-Part Framework:

ROLE: I'm an HVAC contractor with 18 years experience in residential and light commercial systems in Arizona (hot climate, cooling-focused)

AUDIENCE: This estimate is for homeowners of a 2,400 sq ft ranch house who are frustrated with high energy bills ($400/month in summer) and uncomfortable upstairs bedrooms (5 degrees warmer than downstairs)

GOAL: Win this job by showing them the value of the recommended solution (properly sized AC + duct repair + insulation upgrade) vs the cheap fix (just replace old AC unit)

CONSTRAINTS:
- Format: Professional estimate with three pricing tiers
- Tone: Friendly expert, no HVAC jargon
- Must include:
  * Brief explanation of what's wrong in plain English
  * Three options: Budget fix, Recommended solution, Premium solution
  * Energy savings estimates for recommended vs budget
  * Comfort improvement explanation
  * Financing options
- Must explain: Why recommended solution matters (not just cheaper fix)
- Must emphasize: Comfort (family priority) and energy savings (financial priority)
- Include placeholders: [PRICING - I'll fill in], [PHOTOS - I'll upload]
- Warranty info: Our standard warranties for each tier

Based on my diagnosis:
- Current system: 15-year-old 2-ton AC (undersized for 2,400 sq ft, should be 3-ton minimum)
- Duct issues: Visible leaks in attic, estimated 30% efficiency loss
- Insulation: R-13 when it should be R-38 for Arizona climate
- Result: System runs constantly, can't keep up, upstairs always hot

Draft the estimate.

Result: Professional estimate that addresses customer concerns, explains value, and wins jobs at recommended pricing (not budget pricing).

Time saved: 2 hours per estimate, higher closing rate

Quick Reference: Context Framework Templates

For Lawyers

ROLE: [Type of law] attorney with [X years] experience in [specialty]
AUDIENCE: [Judge/client/opposing counsel] who [their tendencies/concerns]
GOAL: [What you need the document to achieve]
CONSTRAINTS:
- Format: [Court requirements]
- Length: [Page limit]
- Tone: [Aggressive/professional/neutral]
- Citations: [Jurisdiction, time period]
- Must include: [Required elements]
- Must avoid: [Things to exclude]

For Doctors

ROLE: [Specialty] physician with [X years] experience
AUDIENCE: [Patient/colleague/insurance] who [their context]
GOAL: [Clinical or administrative outcome needed]
CONSTRAINTS:
- Format: [SOAP/Referral/Patient education]
- Tone: [Professional/patient-friendly]
- Reading level: [If patient-facing]
- Must include: [Clinical elements]
- Must comply with: [HIPAA/Documentation standards]

For Engineers/Technical

ROLE: [Level] [specialty] with [X years] in [domain]
AUDIENCE: [Technical/non-technical stakeholders] who care about [their priorities]
GOAL: [Get approval/explain decision/document architecture]
CONSTRAINTS:
- Format: [ADR/RFC/Technical spec]
- Technical depth: [Match audience]
- Length: [Page/word limit]
- Must include: [Specific sections]
- Focus: [Business value/Technical details/Risk]

For Contractors/Trades

ROLE: [Trade] contractor with [X years] experience in [specialty]
AUDIENCE: [Homeowner/Commercial client] concerned about [their top priority]
GOAL: [Win job/Explain value/Get approval]
CONSTRAINTS:
- Tone: [Friendly expert, no jargon]
- Must include: [Pricing tiers/Warranties/Timeline]
- Must explain: [Why recommended vs cheap fix]
- Emphasize: [Customer's stated concerns]
- Include placeholders for: [Pricing/Photos]

Common Mistakes (And How to Fix Them)

Mistake 1: Vague Role

Bad: "I'm a lawyer" Good: "I'm a commercial litigation attorney with 12 years experience in contract disputes, practicing in New York state courts"

Why it matters: "Lawyer" could be criminal defense, family law, tax law. AI needs specificity to match expertise level and domain.

Mistake 2: Ignoring Audience

Bad: "Write a memo about the diagnosis" Good: "Write a patient-friendly handout about diabetes for a 60-year-old with 8th-grade health literacy who is worried about insulin injections"

Why it matters: The same medical information requires completely different language for patients vs medical students vs insurance companies.

Mistake 3: Unclear Goal

Bad: "Draft a proposal" Good: "Draft a proposal that convinces the CFO to approve $500K investment by demonstrating 18-month ROI and mitigating her concerns about implementation risk"

Why it matters: AI needs to know whether to inform, persuade, explain, or request.

Mistake 4: No Constraints

Bad: Just giving Role/Audience/Goal Good: Adding specific format, tone, length, must-include elements

Why it matters: This is how you get from 50% quality to 80% quality. Constraints turn generic into professional.

How to Build Your Personal Prompt Library

Don't start from scratch every time.

Step 1: Create Templates for Your Repetitive Tasks

Identify the 3-5 documents you create most often:

Lawyer: Motions, client memos, contract reviews Doctor: Chart notes, referrals, patient education Engineer: Architecture docs, technical specs, code reviews Contractor: Estimates, scope of work, change orders

Step 2: Build a 4-Part Template for Each

Example (Lawyer - Motion Template):

ROLE: I'm a [practice area] attorney with [years] experience in [specialty]
AUDIENCE: This motion is for Judge [name] ([court]), who [tendencies]
GOAL: [What you need the motion to achieve]
CONSTRAINTS:
- Format: [Court requirements]
- Length: [Page limit]
- Tone: [Your style]
- Must include: [Standard elements]
- Citations: [Jurisdiction preferences]

[Space for case-specific facts]

Save this template. Next time you need a motion, fill in the blanks.

Step 3: Refine Based on Results

After AI generates output:

If quality is 60%, ask: What context was missing?
If quality is 90%, ask: What made this work so well?
Update your template with what you learned

Over time: Your templates get better, AI output gets more consistent, time saved increases.

The Bottom Line

Bad prompts: "Write me a [thing]" → 30% quality, needs total rewrite

Good prompts using 4-Part Framework:

Role: Who you are
Audience: Who receives this
Goal: What it should achieve
Constraints: Specific requirements

Result: 80% quality first draft, needs only your 20% expert polish

Time saved: 5-10 hours/week on documentation

The difference between "AI is useless for professional work" and "AI just saved me 6 hours" is context.

Professionals who master the 4-Part Framework get AI to deliver consistently good first drafts.

Professionals who don't waste time fighting with generic AI output.

Which one will you be?

Your Next Steps

This week:

✅ Pick ONE repetitive task you do
✅ Build a 4-Part Framework template for it
✅ Test with AI, compare to your normal workflow

This month: 4. ✅ Refine template until you hit 80% quality 5. ✅ Build templates for your other 2-3 common tasks 6. ✅ Track time saved

Related guides:

The 4th Way to Scale Expertise - Why context matters
AI for Lawyers - Legal-specific workflows
AI for Doctors - Medical documentation
Voice-to-Brief Workflow - Combine with voice input

The framework works. Start using it today.

Method & Sources

Framework credit: The 4-part structure (Role, Audience, Goal, Constraints) is adapted from best practices in AI prompt engineering and Nate B Jones's emphasis on context as a multiplier in "The AI Expertise Bottleneck."

Examples tested: All example prompts were tested with ChatGPT-4 and Claude to verify 80%+ quality output.

Last updated: November 22, 2025