Most HR professionals type simple requests into ChatGPT and get generic results. Prompt engineering is the skill of crafting instructions that produce specific, high-quality outputs tailored to your exact needs.
The difference is dramatic. A basic prompt produces a generic job description. An engineered prompt produces a job description with inclusive language, specific competencies, correct salary benchmarks, and your company's tone of voice.
Tell the AI to assume a specific expert persona before answering.
Basic prompt: "Write interview questions for a marketing manager."
Engineered prompt:
You are a senior HR consultant with 15 years of experience in talent acquisition for technology companies in Southeast Asia. You specialise in competency-based interviewing. Generate 8 interview questions for a Marketing Manager role at a B2B SaaS company in Singapore. For each question, specify: the competency being assessed, what a strong answer looks like, and a red flag answer.
Set explicit boundaries on format, length, tone, and content.
Example for policy writing:
Draft a Remote Work Policy for a Malaysian financial services company with 200 employees. Constraints:
- Maximum 2 pages (A4)
- Plain language (avoid legal jargon)
- Must reference Malaysia Employment Act 1955 where relevant
- Include a table summarising eligibility by role type
- Tone: professional but approachable
- Include sections: Purpose, Eligibility, Equipment, Working Hours, Performance, Data Security
Break complex tasks into sequential reasoning steps.
Example for compensation benchmarking:
I need to assess whether our compensation package for a Senior Software Engineer in Singapore is competitive. Think through this step by step:
- First, identify the relevant market benchmarks for this role in Singapore (2025-2026 data)
- Then, compare our package: base S$8,500/month + 2 months bonus + stock options worth S$20,000/year
- Analyse each component separately (base, bonus, equity)
- Finally, recommend adjustments and explain your reasoning
Show the AI examples of the output quality you expect.
Example for job descriptions:
Write a job description for a Data Analyst following this format:
[Example — Senior Marketing Manager] Role: Senior Marketing Manager Department: Marketing | Reports to: CMO | Location: Singapore Impact: Lead demand generation strategy that drives 40% of company revenue Responsibilities: [3-5 bullet points starting with action verbs] Requirements: [4-6 requirements, each with specific benchmarks] Nice-to-haves: [2-3 items] Why join us: [2 sentences on culture and growth]
Now write one for: Data Analyst, Analytics team, reports to Head of Analytics, Singapore office.
Sourcing message:
Write a LinkedIn InMail to a passive candidate for a [Role] position. The candidate currently works at [Current Company] as [Current Role]. Our key differentiators are: [list 3]. Maximum 150 words. Tone: professional, not salesy. Include a specific compliment about their background.
Interview scorecard:
Create an interview scorecard for a [Role] position. Include 6 competencies: [list them]. For each competency, provide: the assessment question, a 1-5 rating scale with descriptions for each level, and space for interviewer notes. Output as a table.
Training programme design:
Design a 3-month leadership development programme for first-time managers in a [Industry] company. Think step by step:
- What are the top 5 skills new managers need?
- For each skill, design a learning module (objective, format, duration, assessment)
- Create a monthly schedule that balances learning with work
- Suggest 3 metrics to measure programme effectiveness Output the full programme in a structured table format.
Survey analysis:
Analyse these engagement survey results using the following framework:
- Group responses into themes (categorise by: leadership, compensation, growth, culture, workload)
- For each theme, calculate the sentiment (positive/neutral/negative)
- Identify the top 3 strengths and top 3 improvement areas
- For each improvement area, suggest 2 evidence-based interventions
- Prioritise interventions by impact (high/medium/low) and effort (high/medium/low)
Data: [paste anonymised survey data]
Policy comparison:
Compare these two versions of our AI Usage Policy. For each section, note:
- What changed
- Whether the change strengthens or weakens the policy
- Any gaps or missing considerations
- Recommended revisions Output as a comparison table with columns: Section, Version 1, Version 2, Assessment, Recommendation
Version 1: [paste] Version 2: [paste]
Bad: "Help me with hiring." Better: "Generate 6 competency-based interview questions for a Finance Manager role at a Singapore bank, focusing on regulatory compliance, team leadership, and stakeholder management."
Bad: "Write an employee handbook section on leave." Better: "Write the Annual Leave section of an employee handbook for a Malaysian company. Must comply with the Employment Act 1955. Company offers 14 days AL for < 2 years service, 18 days for 2-5 years, 22 days for 5+ years."
Bad: "Explain our new benefits package." Better: "Explain our new benefits package in an all-hands presentation script. Audience: 150 employees across 3 countries. Tone: enthusiastic but clear. Highlight the top 3 changes and what they mean for each employee level."
The first output is rarely perfect. Good prompt engineering involves refining:
Organise your best-performing prompts by category:
Store these in a shared document (Google Docs, Notion, SharePoint) so the entire HR team can use and improve them.
Prompt engineering for HR is the skill of writing structured, specific instructions for AI tools that produce high-quality outputs for HR tasks. Techniques include role prompting, constraint-based prompting, chain-of-thought reasoning, and few-shot examples. It transforms generic AI outputs into tailored, professional HR deliverables.
Four key improvements: (1) specify a role or persona for the AI, (2) set explicit constraints on format, length, and tone, (3) provide context like industry, country, and regulations, and (4) iterate on outputs rather than accepting the first result. Building a prompt library of proven templates accelerates improvement.
For recruitment: role prompting (act as an experienced recruiter), constraint-based prompting for job descriptions (format, length, inclusivity), chain-of-thought for candidate evaluation, and few-shot examples for consistent interview question quality. Always include industry, seniority, and location context.
