ChatGPT and Claude Prompts That Get Malaysian Grads Hired

By SuperJobs Team
Quick Answer: Use ChatGPT or Claude in three passes: a gap check against the real job advertisement, a rewrite that keeps only the facts you fed it, and a mock interview you answer out loud. Both have free tiers as of August 2026, but model versions and usage limits change every few weeks, so check openai.com/chatgpt/pricing and claude.ai/pricing before you rely on any limit. The Department of Statistics Malaysia counted 5.98 million graduates in its Graduates Statistics 2024 release, so assume every applicant has the same scaffold and only your own numbers, project names and reasons for applying are scarce.
Jobstreet by SEEK's 2026 relevancy-gap campaign data put the Asia-wide rise in applications received per job at about 14% between the fourth quarter of 2025 and the first quarter of 2026, with total applications up 14.8%. That is an Asia-wide vendor figure, not a Malaysian one, and no Malaysia-only breakdown has been published. The pile is real anyway. A fresh grad in Puchong applying for a finance executive role competes with a stack of documents that all sound the same, because most were generated the same lazy way.
The Department of Statistics Malaysia's Graduates Statistics 2024, released on 31 October 2025, counted 5.98 million graduates in Malaysia, a graduate unemployment rate of 3.2%, and a median monthly salary of RM4,521 for graduates overall and RM5,724 for degree holders.
Every prompt below is meant to be copied straight into ChatGPT or Claude, with the square-bracket placeholders swapped for your real details. This is the prompt-level companion to our guide to how AI is changing job hunting in Malaysia, which covers the hiring side: how Malaysian employers screen with AI, how ATS parsing works, and which interview-practice tools cost what. Read that one for the picture. Use this one with the job ad open in another tab.
Where AI Actually Helps in a Job Hunt (and Where It Hurts)
Treat the model as a fast editor with no memory of your life. It is good at shape and coverage. It knows nothing about your final-year project at UTM or why you would rather work in supply chain than audit.
| Task | Use AI? | Why |
|---|---|---|
| Turning a vague CV bullet into an achievement line | Yes | You supply the number; it fixes the verb and the structure |
| Listing which job-ad requirements your CV never mentions | Yes | Gap-spotting is mechanical work |
| Generating 20 likely interview questions for a role | Yes | Breadth is what models are good at |
| Writing a cover letter from the job ad alone | No | Produces the exact generic letter recruiters reject |
| Inventing experience, tools or grades you lack | Never | It collapses the moment you are interviewed |
| Stuffing keywords or hidden text to trick an ATS | Never | Any human who opens the file can see it, and ManpowerGroup Malaysia says it is auditing for it |
Jobstreet's own career advice on AI cover letters puts it plainly: recruiters value authenticity, a generic cover letter may fail to make a lasting impression, and AI output does not always align with applicant tracking system requirements. Its responsible AI page adds that Jobstreet's services do not make the final decision to shortlist, interview or hire anyone. A human still reads you.
Work from real ads. Browse fresh graduate and entry-level jobs on SuperJobs and pick three kerja kosong you would genuinely accept.
Prompt Templates for Resume Rewriting
Start with raw material, not polish. For each CV bullet, write down what you did, what tool you used, and any number attached to it. Rough numbers work: 40 students, 12 weeks, RM8,000 budget.
If a file upload is refused on a free plan, paste your CV in as plain text instead.
Prompt 1: turn duties into achievements
You are a Malaysian recruitment consultant who screens fresh graduate CVs for
roles in Kuala Lumpur and Selangor.
Below is one line from my CV plus the raw facts behind it. Rewrite it as ONE
bullet of at most 22 words, starting with a strong verb, keeping every number
I gave and adding none of your own.
Original line: [e.g. "Helped with social media for university club"]
Raw facts: [e.g. ran Instagram for a 300-member club, 14 posts a month,
followers grew from 800 to 2,100 over 5 months, used Canva]
Target role: [e.g. Marketing Executive, FMCG, Petaling Jaya]
Rules:
- Do not invent metrics, tools, budgets or job titles.
- If a claim is not supported by my raw facts, leave it out.
- Give me three versions: factual, results-led, and skills-led.
Prompt 2: gap check against a real job ad
Here is a job advertisement and my CV. Both are pasted below.
JOB AD:
[paste the full ad from SuperJobs, Jobstreet or the company careers page]
MY CV:
[paste your CV as plain text]
Do three things:
1. List every requirement in the ad that my CV does not evidence.
2. For each gap, say whether it is a real gap or just missing wording, and ask
me a question that helps me answer it honestly.
3. List the exact phrases from the ad I could use IF they describe something I
have genuinely done.
Do not rewrite my CV yet. Just show me the gaps.
That second step is the whole trick, and the one most people skip. The model asks, you answer with real experience, and only then does anything get written.
One last pass: ask the model to cut the CV to one A4 page, keeping education, one line per role with its strongest number, and your tools, and cutting objective statements and SPM results if you hold a degree.
Run the result through the SuperJobs CV checker to see how it parses before an employer's system sees it. Our guide to writing a winning CV in Malaysia covers what Malaysian recruiters expect in each section.
Prompt Templates for Cover Letters
A 2025 Resume Now survey found 62% of hiring managers said AI-generated resumes without customisation were more likely to be rejected, while 78% said personalised details signalled genuine interest. That survey is US-based, but the mechanism is the same here: the reader is scanning for one sentence that could only have been written about their company. The prompt has to force you to supply it.
Act as a Malaysian hiring manager at [e.g. a mid-sized logistics firm in Shah
Alam]. I am applying for [e.g. Executive, Supply Chain] at [company name].
Before writing anything, ask me these four questions one at a time and wait
for my answer:
1. What specific thing about this company or role made you apply?
2. Which ONE experience of yours maps to the main duty in the ad?
3. What number can you attach to that experience?
4. What is the weakest part of your application, and your honest answer if I
press on it?
Then write a cover letter of 200-250 words in plain Malaysian business
English. No American idioms. No "I am writing to express my keen interest".
Open with my answer to question 1. Add no fact I did not give you.
Once you have a draft, strip the model's fingerprints:
Rewrite this cover letter in my voice. Here is something I wrote myself so you
can copy my rhythm and vocabulary: [paste a LinkedIn post, an email to a
lecturer, anything of yours].
Rules:
- Vary sentence length. At least two sentences under eight words.
- Remove every adjective that is not doing work.
- No "passionate about", no "dynamic", no "results-driven".
- Keep it under 250 words.
If the ad asks for a surat permohonan kerja in Bahasa Malaysia, add one line to the first prompt: write it in formal BM, same facts, nothing added. Then have a fluent friend read it aloud, because models slide into Indonesian phrasing.
Never mention a company value you have not read on the company's own site. That is the most common tell.
Prompt Templates for LinkedIn Profile Optimisation
Recruiters search LinkedIn by keyword and location, so your headline and About section are search fields, not a personal essay. Our wider AI guide does not cover LinkedIn at all.
Headline prompt
Write 8 LinkedIn headline options, each under 120 characters, for a Malaysian
fresh graduate.
Degree: [e.g. BSc Computer Science, Universiti Malaya, 2026]
Target roles: [e.g. Junior Software Engineer, Data Analyst]
Real skills with evidence: [e.g. Python, SQL, built a dashboard for a
final-year project on Selangor traffic data]
Location: [e.g. Klang Valley, open to Penang]
Every headline must contain a keyword a recruiter would type into LinkedIn
search. No slogans, no "aspiring", no emoji.
About section prompt
Write my LinkedIn About section in first person, 120-160 words, four short
paragraphs:
1. What I do and the job titles I want, named exactly.
2. One project or internship, including this number: [your number].
3. My tools and languages, listed plainly.
4. How to reach me and what I am open to (full time, contract, [state]).
Write it the way a 23-year-old Malaysian graduate speaks at work. No corporate
slogans. No sentence longer than 25 words.
Search-coverage prompt
Here is my LinkedIn headline, About section and one experience entry:
[paste all three]
Act as a recruiter hiring for [role] in [city]. Tell me:
1. The ten search terms you would type into LinkedIn to find someone for this
role.
2. Which of those terms my profile already contains, word for word.
3. Which terms I should NOT add because nothing in my profile supports them.
Do not rewrite my profile. Just show me the coverage.
Point three matters most. A profile stuffed with terms you cannot defend gets you the screening call and then wastes it.
Paste the result into the SuperJobs LinkedIn optimizer to check keyword coverage, then apply to the roles you shortlisted on SuperJobs using the same wording across CV, LinkedIn and application form.
Using AI for Company Research Before Applying
Our AI job hunting guide already gives a basic company-research template, so this section covers only the part that goes wrong. Models state a revenue figure, a CEO name or an office location with total confidence and no source. Turn on web search, then force the citation:
Search the web and brief me on [company name] in Malaysia before my interview.
Give me:
1. What the company actually sells and who pays them.
2. Their Malaysian footprint: offices, headcount if published, main markets.
3. Anything announced in the last 12 months: funding, expansion, new products.
4. Three questions I could ask the interviewer, based only on points 1 to 3.
For every factual claim, give me the source URL. If you cannot find a source,
write "unverified" next to the claim. Do not guess.
Then verify the leads. Listed-company announcements sit on Bursa Malaysia's site, registration details with SSM, and salary bands against SuperJobs salary insights. If a number has no source anywhere, drop it rather than repeat it in an interview.
Using AI to Prep for Interviews (Mock Q&A)
For a comparison of dedicated practice tools and what they cost, our AI job hunting guide lists them. What follows is the prompt that makes an ordinary chatbot behave like a hard first-round interviewer, which is free and usually enough.
You are interviewing me for [role] at [type of Malaysian company]. Conduct a
realistic 30-minute first-round interview.
Rules:
- Ask ONE question at a time and wait for my answer.
- After each answer, rate it 1-5, name the weakness, then ask a follow-up that
presses on that weakness.
- Include two questions about my weakest area: [e.g. no internship experience
/ CGPA 2.9 / switching from engineering to marketing].
- Do not be encouraging. Be fair but hard.
- At the end, name the three answers I must rewrite.
My CV: [paste]
The job ad: [paste]
Salary is where a chatbot will hand you a number that gets you laughed out of the room. DOSM's RM4,521 median, and the RM5,724 figure for degree holders, covers all employed graduates at every experience level, not fresh graduates, and the release publishes no fresh-graduate or age-band breakdown. Use it as a ceiling reference only and build your range from comparable entry-level listings you can point to.
I have an interview for [role] in [city]. Comparable entry-level listings show
[paste 3 real ranges, with the job title and where you found each one].
Context you must respect: DOSM's Graduates Statistics 2024 median of RM4,521
a month (RM5,724 for degree holders) covers ALL employed graduates at every
experience level, not fresh graduates. Treat it as a ceiling reference only.
Build my range from the entry-level listings above.
Script a salary answer that gives a range rather than one number, justifies it
with these skills [list], handles "what is your current salary?" when I have
never been employed, and handles an offer RM500 below my range.
Write it as short spoken lines I can practise, not as a paragraph.
Practise out loud. The gap between what reads well and what you can say under pressure is where most fresh grads lose the room.
Avoiding the "AI Wrote This" Detection Problem
Detection is a real risk, but not the risk most people assume. AI text detectors are not reliable enough to bet a hiring decision on. Liang, Yuksekgonul, Mao, Wu and Zou, a Stanford team publishing in the journal Patterns in 2023, tested widely used GPT detectors and found they consistently misclassified non-native English writing as AI-generated while classifying native-speaker writing correctly. In Malaysia, where plenty of strong candidates write competent second-language English, a detector score says more about the detector than the applicant.
The real detection is a human recruiter recognising the pattern in three seconds.
| Tell | Fix |
|---|---|
| Opens with "I am writing to express my keen interest" | Open with the specific reason you applied |
| Praises the company in words taken from nowhere | Name a product, a client, an announcement you read |
| Every sentence 18 to 25 words long | Put a five-word sentence in |
| Adjectives with no evidence: dynamic, passionate, results-driven | Replace each with a number |
| Same letter structure across five applications | Rewrite the opening paragraph for every one |
| Wording no Malaysian uses: "reach out", "circle back", "spearheaded" | Write how you would speak in a KL office |
| Three-item lists everywhere, each item the same length | Break the rhythm; use two items, or five |
| A closing paragraph that summarises the letter | Cut it and end on the ask |
Run this over your own draft before you send it:
Read this cover letter. List every sentence that could appear, unchanged, in
an application to a different company in a different industry. Quote them back
to me. Do not rewrite anything.
Whatever it quotes back is the part a recruiter skims past. Then read the letter aloud: any sentence you would feel embarrassed saying to a person reads as machine output, so cut it.
Then add one detail no model could have produced. The lecturer who supervised your project, the system you used at your internship, the reason you left engineering for marketing. Specificity is the only real proof of authorship.
Ethical Guardrails: Don't Cross These Lines
The test in our AI job hunting guide still holds: would you be comfortable telling the interviewer exactly what the tool did? Three things fail that test in ways specific to applying in Malaysia right now.
Hidden text is the first. ManpowerGroup Malaysia reported in October 2025 that around 10% of resumes scanned through its AI filters contained hidden-text tricks, white-on-white instructions aimed at the screening model. That is the firm's own screening data with no sample size or method published, so read it as a signal rather than a national rate. The same post advises Malaysian employers to audit their tools for hidden commands, which is the part that should worry you: the trick is on the industry's radar, and getting caught ends your candidacy with that recruiter.
Fabrication is the second. Never invent experience, tools, grades or employment dates. A fabricated line becomes a lie you defend in the interview, and employers routinely treat a fabricated qualification as a disciplinary matter once you are on the payroll.
Confidentiality is the third. Internship documents, client data and other people's personal details do not belong in a prompt, whichever tool you are using.
Two shorter rules. Do not use AI during a timed test, take-home or live coding round unless the employer says you may. And if you used AI in a portfolio piece, say so in one line rather than passing it off as unassisted work.
If an employer asks whether you used AI, answer honestly and specifically: you used it to structure and edit, and all facts and examples are yours. That answer is defensible. "No" is not, if it is untrue.
Frequently Asked Questions
Is it okay to use ChatGPT to write my resume in Malaysia?
Yes, as an editor rather than an author. Jobstreet's own career advice stresses authenticity and tailoring, and hiring manager surveys show it is generic, uncustomised AI output that gets rejected, not AI assistance itself. Supply your own achievements and numbers, let the tool tighten the wording, then rewrite anything that does not sound like you. Every claim must be defensible in an interview.
Which is better for job applications, ChatGPT or Claude?
Both work, and both have free tiers as of August 2026. Anthropic announced Claude Sonnet 5 on 30 June 2026 and states it is the default model on the Free and Pro plans. Free-tier model versions and usage limits change every few weeks on both products, so check openai.com/chatgpt/pricing and claude.ai/pricing before you rely on any limit. Many job seekers draft in one tool and edit in the other, because a second model spots the first one's filler wording.
Can Malaysian employers detect an AI-written cover letter?
AI detectors are unreliable. A 2023 Stanford study published in the journal Patterns found they consistently misclassify writing by non-native English speakers as machine-generated. The real detection is human. Recruiters reading dozens of applications recognise generic praise, uniform sentence length and vague adjectives immediately. The fix is not disguising the text but adding details only you could know: a named project, a real number, a specific reason you applied there.
Should I tell an employer that I used AI in my application?
Only if asked, and then answer honestly. Say you used it to structure and edit, and that all experience and figures are your own. Most employers accept AI assistance in drafting. What they reject is fabricated experience, or a candidate whose interview answers do not match the written application.
What is the biggest mistake fresh graduates make with AI job applications?
Sending the same AI-generated letter to twenty companies with only the company name changed. Jobstreet by SEEK's 2026 relevancy-gap campaign data put the Asia-wide rise in applications received per job at about 14% from the fourth quarter of 2025 to the first quarter of 2026, so a generic letter sits in a bigger pile of identical ones. Rewriting the opening paragraph for each company takes ten minutes.
Can AI help me prepare for a Malaysian job interview?
Yes, and this is its strongest use. Ask the model to act as a hard interviewer, answer one question at a time out loud, and have it press on your weakest area. For salary questions, build your range from comparable entry-level listings rather than DOSM's published graduate medians, which cover all employed graduates at every experience level rather than fresh graduates only.
Take the Next Step
- Shortlist three real roles before you open any chatbot. Browse fresh graduate and entry-level jobs on SuperJobs.
- Check how your rewritten CV parses with the SuperJobs CV checker.
- Tidy your profile with the SuperJobs LinkedIn optimizer.
- Set your salary range using SuperJobs salary insights.
- Read how AI is changing job hunting in Malaysia for the employer-side picture, and the winning CV guide for Malaysia for the structure these prompts assume.
?Frequently Asked Questions
Is it okay to use ChatGPT to write my resume in Malaysia?
Yes, as an editor rather than an author. Jobstreet's own career advice stresses authenticity and tailoring, and hiring manager surveys show it is generic, uncustomised AI output that gets rejected, not AI assistance itself. Supply your own achievements and numbers, let the tool tighten the wording, then rewrite anything that does not sound like you. Every claim must be defensible in an interview.
Which is better for job applications, ChatGPT or Claude?
Both work, and both have free tiers as of August 2026. Anthropic announced Claude Sonnet 5 on 30 June 2026 and states it is the default model on the Free and Pro plans. Free-tier model versions and usage limits change every few weeks on both products, so check openai.com/chatgpt/pricing and claude.ai/pricing before you rely on any limit. Many job seekers draft in one tool and edit in the other, because a second model spots the first one's filler wording.
Can Malaysian employers detect an AI-written cover letter?
AI detectors are unreliable. A 2023 Stanford study published in the journal Patterns found they consistently misclassify writing by non-native English speakers as machine-generated. The real detection is human. Recruiters reading dozens of applications recognise generic praise, uniform sentence length and vague adjectives immediately. The fix is not disguising the text but adding details only you could know: a named project, a real number, a specific reason you applied there.
Should I tell an employer that I used AI in my application?
Only if asked, and then answer honestly. Say you used it to structure and edit, and that all experience and figures are your own. Most employers accept AI assistance in drafting. What they reject is fabricated experience, or a candidate whose interview answers do not match the written application.
What is the biggest mistake fresh graduates make with AI job applications?
Sending the same AI-generated letter to twenty companies with only the company name changed. Jobstreet by SEEK's 2026 relevancy-gap campaign data put the Asia-wide rise in applications received per job at about 14% from the fourth quarter of 2025 to the first quarter of 2026, so a generic letter sits in a bigger pile of identical ones. Rewriting the opening paragraph for each company takes ten minutes.
Can AI help me prepare for a Malaysian job interview?
Yes, and this is its strongest use. Ask the model to act as a hard interviewer, answer one question at a time out loud, and have it press on your weakest area. For salary questions, build your range from comparable entry-level listings rather than DOSM's published graduate medians, which cover all employed graduates at every experience level rather than fresh graduates only.