Prompt engineering is the skill that separates people who get mediocre AI outputs from those who get genuinely impressive ones. The models are the same for everyone — the difference is entirely in how you communicate with them.
The Five Core Principles
- Specificity beats brevity — more context almost always produces better output
- Role assignment improves quality — "Act as a senior copywriter with 15 years B2B experience"
- Examples outperform instructions — showing beats telling, every time
- Format specification eliminates reformatting — tell the AI exactly how to structure output
- Constraints focus creativity — "in under 200 words, using no jargon" produces sharper output
The RICE Framework
RICE prompting framework
| Component | Description | Example |
|---|---|---|
| Role | Define who the AI is | "You are a conversion copywriter with 10 years of SaaS experience" |
| Intent | State the goal clearly | "Write a landing page headline for..." |
| Context | Provide relevant background | "Target audience is SMB CEOs aged 35-50" |
| Examples | Show 2-3 examples of good output | "Here are two examples I like: [examples]" |
Advanced Techniques
- Few-shot prompting: provide 3-5 examples of desired input-output pairs
- Self-critique: "Now identify 3 weaknesses in this response and fix them"
- Constraint stacking: add 3-4 specific constraints to focus output quality
- Output chaining: use output of one prompt as input to the next
🎯
The One-Sentence Rule
If you cannot describe your goal in one clear sentence, your prompt will produce unfocused output. Write the goal sentence first, then expand.
Prompt EngineeringChatGPTClaudeLLMAI Productivity
Tanvir Tuhin
AI consultant, digital marketer, and study abroad mentor based in Aberdeen, UK. Founder of JJAT Education.
Work with Tanvir