Prompt Engineering
A Practical Guide to Writing Effective AI Prompts
Version: 1.0
Audience: Government Officials, Business Professionals, Managers, Developers,
Business Analysts, Students
Table of Contents
1. Introduction
2. What is Prompt Engineering?
3. Why Prompt Engineering Matters
4. How Large Language Models Work
5. Anatomy of an Effective Prompt
6. Types of Prompts
7. Prompt Engineering Techniques
8. Prompt Templates
9. Business Use Cases
10. Government Use Cases
11. Common Mistakes
12. Best Practices
13. AI Governance and Ethics
14. Practical Exercises
15. Cheat Sheet
16. Conclusion
1. Introduction
Artificial Intelligence (AI) has transformed the way organisations create documents,
analyse information, automate repetitive tasks, and support decision-making. However,
the quality of AI-generated responses depends significantly on the quality of the
instructions provided.
Prompt Engineering is the practice of designing clear, structured, and effective prompts
that enable AI systems to produce accurate, relevant, and useful outputs.
This guide introduces prompt engineering principles and demonstrates how they can be
applied across business, government, education, and software development.
2. What is Prompt Engineering?
A prompt is an instruction, question, or request given to an AI model.
Prompt Engineering is the process of crafting prompts that maximise the quality,
relevance, and usefulness of AI-generated responses.
Think of prompts as instructions given to an expert consultant.
Poor instructions produce poor results.
Clear instructions produce high-quality outputs.
3. Why Prompt Engineering Matters
Good prompts help to:
• Produce more accurate responses
• Save time
• Improve productivity
• Generate higher-quality reports
• Reduce revisions
• Improve decision support
• Automate repetitive work
• Enhance creativity
4. How AI Understands Prompts
Large Language Models process prompts by identifying:
• Context
• Intent
• Instructions
• Constraints
• Examples
• Desired output
The more context provided, the better the response.
Example:
Poor Prompt
“Write a report.”
Improved Prompt
“Prepare a two-page executive report summarising the progress of the Department’s
Digital Transformation Programme. Include objectives, achievements, risks,
recommendations, and use a professional tone.”
5. Anatomy of an Effective Prompt
An effective prompt typically contains:
Objective
What do you want?
Example:
Prepare a project proposal.
Context
Provide background.
Example:
The proposal is for implementing an electronic document management system.
Audience
Who will read it?
Example:
Chief Director
Executive Committee
Minister
Board
Tone
Examples:
• Formal
• Professional
• Technical
• Friendly
• Executive
• Academic
Output Format
Examples:
• Table
• Report
• Presentation
• Email
• Bullet list
• Executive summary
• Action plan
Constraints
Examples:
Maximum two pages
Use South African legislation
Include references
Avoid technical jargon
6. Types of Prompts
Information Prompt
Example
Explain machine learning.
Summarisation Prompt
Summarise the attached report into one page.
Transformation Prompt
Rewrite the document in plain English.
Analytical Prompt
Compare cloud computing and on-premise infrastructure.
Creative Prompt
Generate five innovative digital transformation initiatives.
Decision Support Prompt
Evaluate three procurement options and recommend the best solution.
Coding Prompt
Develop a Python application that reads an Excel file and generates a dashboard.
7. Prompt Engineering Techniques
Zero-Shot Prompting
Ask without examples.
Example
Explain blockchain.
One-Shot Prompting
Provide one example.
Example
Example:
Input: Purchase Order
Output: Finance
Now classify:
Invoice
Few-Shot Prompting
Provide several examples before asking AI.
Step-by-Step Prompting
Ask AI to solve problems sequentially.
Example
Analyse the requirements.
Identify risks.
Recommend mitigation.
Produce implementation plan.
Role-Based Prompting
Assign AI a role.
Example
Act as a Chief Information Officer.
Prepare an ICT Strategy.
Persona Prompting
Act as
• Auditor
• HR Director
• Financial Analyst
• Medical Specialist
• Procurement Expert
Constraint Prompting
Examples
Limit to 500 words.
Avoid abbreviations.
Use UK English.
Format Prompting
Return results as
• Markdown table
• Excel-ready table
• JSON
• PowerPoint outline
• Project plan
8. Prompt Templates
Email
Write a professional email to [recipient].
Purpose:
Tone:
Maximum length:
Report
Prepare a report containing
Introduction
Objectives
Current Situation
Analysis
Risks
Recommendations
Conclusion
Presentation
Prepare a 15-slide executive presentation.
Audience:
Objectives:
Speaker notes included.
Meeting Minutes
Generate meeting minutes using
Attendees
Discussion
Decisions
Actions
Due dates
Business Proposal
Prepare a proposal including
Executive Summary
Background
Scope
Benefits
Timeline
Budget
Risk Assessment
Recommendations
9. Business Use Cases
Human Resources
• Job descriptions
• Performance reviews
• Training material
Finance
• Budget reports
• Financial analysis
• Cost reduction strategies
Marketing
• Campaigns
• Social media
• Product descriptions
IT
• System documentation
• Test cases
• API documentation
• User manuals
Legal
• Contract summaries
• Compliance analysis
• Risk assessments
Operations
• SOPs
• Workflow documentation
• Process optimisation
10. Government Use Cases
Artificial Intelligence can assist with
Policy drafting
Cabinet memorandums
Parliamentary questions
Annual Performance Plans
Strategic Plans
Procurement documentation
Tender specifications
Risk registers
Audit responses
Monitoring and Evaluation reports
Budget motivation
Grant applications
Service delivery improvement
Citizen engagement
Batho Pele implementation
11. Common Prompt Mistakes
Too vague
Example
Write something about healthcare.
Better
Prepare a two-page report on improving patient waiting times in provincial hospitals.
No audience specified
Specify who will read the document.
No format
Always specify
PowerPoint
Report
Email
Table
No context
Explain why you need the information.
Too many requests
Break large requests into smaller prompts.
12. Best Practices
Be specific.
Provide context.
Define the audience.
Specify the format.
Provide examples.
Review outputs.
Validate important information.
Iterate and refine prompts.
Never rely solely on AI for critical decisions.
13. AI Governance and Ethics
Organisations should establish governance covering:
Privacy
Security
Transparency
Human oversight
Bias mitigation
Accountability
Responsible AI
Compliance with applicable legislation
For South African organisations, AI use should align with the Protection of Personal
Information Act (POPIA), the Promotion of Access to Information Act (PAIA), cybersecurity
policies, procurement regulations where applicable, and organisational governance
frameworks.
14. Practical Exercises
Exercise 1
Rewrite a vague prompt into a detailed prompt.
Exercise 2
Generate an executive report from meeting notes.
Exercise 3
Draft a procurement specification using AI.
Exercise 4
Summarise a 20-page policy into two pages.
Exercise 5
Develop an implementation roadmap using AI.
15. Prompt Cheat Sheet
Instead of Use
Explain Explain with examples
Write Prepare a professional
report
List Categorise into a table
Compare Compare advantages,
disadvantages, risks and
recommendations
Summarise Summarise for
executives
Improve Rewrite using
professional business
language
Analyse Analyse and recommend
actions
Create Produce a structured
implementation plan
Example Prompt
“You are a Chief Information Officer in a South African National Department.
Prepare a five-year Digital Transformation Strategy aligned with the National Development
Plan, departmental strategic objectives, POPIA, PFMA, and cybersecurity requirements.
Include the current environment, strategic objectives, implementation roadmap,
governance, risk management, budget estimates, key performance indicators, expected
benefits, and recommendations.
Present the response as a professional executive report suitable for presentation to the
Director-General.”
16. Conclusion
Prompt Engineering is becoming an essential professional skill in the age of Artificial
Intelligence. Effective prompts improve the quality of AI outputs, reduce effort, increase
productivity, and enable organisations to use AI responsibly. By applying structured
prompting techniques, professionals can leverage AI to support planning, analysis,
decision-making, reporting, and service delivery while maintaining appropriate human
oversight and governance.