Effective prompting: Core principles

Principles of Effective Prompting

Principle 1: Clarity Above All

An AI model doesn't read your mind — it reads your words only. Be specific instead of vague.

Vague: "Improve this code" Specific: "Improve the performance of this Python function. Problem: slow with lists > 10,000 elements. Goal: reduce execution time by at least 50%"

Principle 2: Context Changes Everything

AI knows nothing about your situation unless you tell it. Provide sufficient context.

Without context: "Write an email" With context: "Write an email to a client who has delayed their payment for 3 weeks. Tone: professional but firm. Don't mention the invoice number explicitly"

Principle 3: Specify the Desired Format

Don't expect AI to know how you want the output. Tell it explicitly.

| Instead of | Say | |------------|-----| | "Explain" | "Explain in 3 numbered points" | | "Evaluate" | "Evaluate in a table with columns: Pros / Cons / Recommendation" | | "Summarize" | "Summarize in two sentences for the CEO" |

Principle 4: Role + Task = Better Results

Giving the model a specific role significantly improves output.

Without role: "Review this contract" With role: "You are a lawyer specializing in employment contracts in Saudi Arabia. Review this contract and identify any unusual clauses or potential risks for the employee"

Principle 5: Iteration and Improvement

The first prompt is rarely the perfect one. Treat it as a first draft:

1. Start with a basic prompt
2. Evaluate the output: what's missing?
3. Add more precise instructions
4. Repeat until you reach the goal

The Thumb Rule

If you were explaining the task to a colleague who knows nothing about your context, you'd give them the same details you should give to the model.