How to Write Better AI Prompts for Any Profession: A Practical Guide
Learn the CRTSE framework for writing profession-specific AI prompts that produce accurate, useful output every time.
Most professionals who try AI for the first time get mediocre results and assume the technology is not ready. The problem is almost never the AI — it is the prompt. A vague prompt produces a vague answer. A structured prompt produces professional-grade output that actually saves time.
This guide introduces the CRTSE framework, a repeatable method for writing prompts that work across any profession. Whether you are drafting a SOAP note, a sales email, a lesson plan, or a legal memo, the same five elements produce consistently better results.
The CRTSE Framework
CRTSE stands for Context, Role, Task, Standards, and Examples. Each element gives the AI a critical piece of information it needs to generate useful output.
Context tells the AI about the situation. Who is the patient, client, or student? What has happened so far? What constraints exist? Without context, the AI fills in blanks with generic assumptions.
Role defines who the AI should write as. A pharmacist writes differently than a personal trainer. Specifying the role activates domain-specific vocabulary, formatting conventions, and professional standards.
Task is the specific output you need. Not "help me with documentation" but "write a prior authorization letter for this medication." The more specific the task, the more useful the output.
Standards are the rules the output must follow. This includes formatting requirements, regulatory standards, reading level targets, word count limits, and any compliance considerations.
Examples show the AI what good output looks like. Even a partial example dramatically improves output quality by giving the AI a concrete target to match.
CRTSE in Practice: Four Professions
Pharmacist SOAP Note
A pharmacist using the SOAP Note Generator might structure their input like this:
- Context: 62-year-old male, type 2 diabetes, current A1C 8.4%, on metformin 1000mg BID, adding empagliflozin 10mg daily
- Role: Clinical pharmacist in ambulatory care setting
- Task: Generate a SOAP note for this medication therapy management encounter
- Standards: Follow DAP format, include ICD-10 codes, meet CMS documentation requirements for MTM billing
- Examples: "Assessment should list each drug therapy problem with clinical justification"
The result is a complete, billable SOAP note rather than a generic paragraph about diabetes management.
Sales Rep Email
A sales representative using the Sales Email Generator applies the same framework:
- Context: Prospect is a mid-market SaaS company, 200 employees, currently using a competitor product, showed interest at a trade show last week
- Role: Enterprise software sales representative
- Task: Write a follow-up email referencing our trade show conversation and proposing a demo
- Standards: Under 150 words, one clear CTA, professional but not stiff, no jargon
- Examples: "Subject lines that have worked: 'Quick follow-up from [Event Name]'"
Teacher Lesson Plan
A teacher using the Lesson Plan Generator might prompt:
- Context: 8th grade science class, 28 students, mixed ability levels, currently in the ecosystems unit, students struggled with food web concepts last week
- Role: Middle school science teacher aligned to NGSS standards
- Task: Create a 50-minute lesson plan on energy flow in ecosystems
- Standards: Include warm-up, direct instruction, group activity, formative assessment. Align to NGSS MS-LS2-3. Differentiate for three levels.
- Examples: "Formative assessment should be an exit ticket with 3 questions"
Attorney Legal Memo
An attorney using the Legal Research Memo Tool structures their input:
- Context: Client is a small business owner in California who signed a non-compete with a former employer, wants to start a competing business
- Role: Employment law attorney
- Task: Draft an internal research memo on the enforceability of this non-compete under current California law
- Standards: IRAC format, cite specific statutes (California Business and Professions Code Section 16600), include recent case law, 2-3 pages
- Examples: "Conclusion should give a clear recommendation on likelihood of enforcement"
Common Prompt Mistakes
Being Too Vague
"Write a note about my patient" gives the AI nothing to work with. Specify the patient context, the type of note, and the documentation standard. Every missing detail is a detail the AI will guess wrong.
Skipping Format Specification
If you need bullet points, say so. If you need SOAP format, say so. If you need a specific word count, say so. AI defaults to paragraph form and tends to be verbose. Tell it exactly what structure you want.
No Audience Specification
A drug interaction summary for a physician reads very differently from one for a patient. A financial report for a client reads differently from one for an internal audit. Always specify who will read the output.
Ignoring Professional Standards
Every profession has documentation standards, compliance requirements, and formatting conventions. Referencing these by name — CMS guidelines, NGSS standards, ABA formatting rules, HIPAA requirements — helps the AI produce output that meets real-world expectations.
One-Shot Prompting
Do not try to get everything perfect in a single prompt. Start with the CRTSE framework, review the output, then refine. Add constraints you forgot. Ask the AI to adjust the tone, shorten a section, or add missing elements. Iteration is faster than perfection on the first try.
Building Your Prompt Library
Once you find a prompt structure that works, save it. The most efficient professionals build a personal library of prompt templates for their recurring tasks. A pharmacist might have templates for prior auths, MTM notes, and patient counseling. An attorney might have templates for demand letters, research memos, and client communications. A teacher might have templates for lesson plans, rubrics, and parent emails.
Our tools catalog includes pre-built prompt structures for over 47 tasks across 26 professions. Each tool applies the CRTSE framework automatically, so you can focus on inputting your specific details rather than engineering the prompt from scratch.
Start Here
Pick one task you do repeatedly — the one that eats the most time. Apply the CRTSE framework to it. Compare the AI output to what you would have written manually. Refine the prompt until the output requires minimal editing. Then move to the next task.
The professionals saving 10-15 hours per week with AI are not using magic prompts. They are using structured, specific prompts built on a consistent framework. CRTSE gives you that framework.
Save hours every week with the AI Career Lab — All 7 Claude Vaults
All seven profession-specific Claude Vaults — 315 skills total, with setup wizards and guard skills.
Related Guides
Designers After AI: A 30-Day Reskilling and Repositioning Playbook (2026)
Claude Design launched April 17. A concrete 30-day plan for designers: diagnose what's at risk, reskill on AI-augmented workflows, reposition, and apply.
The Agentic-AI Job Guide: 8 New Roles, What They Pay, and How to Break In
Agentic AI Engineer, AI Agent Architect, AI Trainer — eight new job titles that didn't exist three years ago, with 2026 salary bands, the skills you need, and the realistic on-ramp from each adjacent role.
The 2026 AI-Augmented Resume: How to Show AI Fluency Without Sounding Like ChatGPT Wrote It
ATS systems in 2026 use semantic matching and actively flag AI-generated language. Here's how to write a resume that proves AI fluency without tripping the detectors — and the four sentence patterns to delete from your resume today.