Best ChatGPT Prompts for Resume Writing (2026)

Bullet points are where most resumes lose interviews. Hiring managers spend 6–10 seconds on an initial scan. Bullets that start with weak verbs, lack numbers, or bury the outcome get skipped. The goal is: strong verb → specific action → quantified result. The prompt below forces the model to produce that structure every time.

**Prompt — bullet point rewriting with metrics:** ``` You are a professional resume writer with 15 years of experience in [your target industry, e.g., software engineering / marketing / finance]. I am going to give you my current resume bullet points. Rewrite each one using this structure: [Strong action verb] + [specific action taken] + [quantified result or business impact]. Rules: - Every bullet must start with a past-tense action verb (no weak openers like 'Responsible for' or 'Helped with') - If I haven't provided a metric, ask me one clarifying question per bullet to get one before writing - Keep each bullet under 2 lines - Do not invent numbers — flag where I need to fill in a metric - Output format: numbered list matching the original order Here are my current bullets: [PASTE YOUR BULLETS HERE] ```

Run this prompt with GPT-5 or Claude Opus 4.8. Both models understand industry-specific impact language. If you don't have exact metrics, run the prompt and then ask the model 'What questions should I answer to add metrics to each of these?' as a follow-up — it will generate targeted prompts for each bullet. For deeper guidance on role-framing in prompts, see our guide on how to assign a role in a prompt.

Applicant Tracking Systems filter resumes before any human reads them. Most ATS tools score resumes by matching keywords from the job description. A strong resume can be filtered out simply because it uses 'revenue operations' where the JD says 'RevOps.' The fix is systematic keyword alignment — and a model handles this faster and more thoroughly than any manual approach.

**Prompt — ATS keyword optimization:** ``` You are an ATS optimization specialist. I will give you a job description and my current resume. Your job is to: 1. Extract the top 20 hard skills, soft skills, and job title keywords from the job description 2. Identify which of those keywords are missing from my resume entirely 3. Identify which keywords are present but phrased differently from the JD (and note the preferred phrasing) 4. Suggest specific sentences or bullet edits that naturally incorporate the missing/rephrased keywords — do not keyword-stuff or add anything I don't actually have experience with 5. Output a revised keyword-optimized version of my resume summary section Job description: [PASTE JOB DESCRIPTION] My current resume: [PASTE RESUME] ```

GPT-5.5 and Gemini 2.5 Pro are both strong for this task — both have large enough context windows to handle a full JD plus resume in one shot. OpenAI's documentation on long-context prompting notes that structuring the input with clear labeled sections improves extraction accuracy, which is exactly what this prompt does. Run the ATS check every time you apply to a different company — even a small keyword gap can drop your score below the filter threshold.

ATS keyword matching is necessary but not sufficient. A hiring manager reading a generic resume can tell instantly that it wasn't written for their role. Tailoring means reordering sections, emphasizing the experiences most relevant to that specific job, and mirroring the language the company uses for priorities, culture, and success metrics.

**Prompt — resume tailoring to a job description:** ``` You are a senior recruiter at [company name or 'a [industry] company']. I'm applying for the role of [job title]. I'm going to give you both my resume and the job description. Your task is: 1. Identify the top 5 qualifications the JD is prioritizing (explicit and implicit) 2. Match those to my resume — find the strongest evidence for each priority 3. Suggest specific edits: which experiences to move higher, which bullets to expand, which to cut or shorten 4. Rewrite my resume summary to directly address what this employer is looking for 5. Flag any gap between what the JD requires and what my resume shows Job description: [PASTE JOB DESCRIPTION] My resume: [PASTE RESUME] Output format: a numbered action list of edits, then the rewritten summary. ```

This prompt works especially well on Claude Opus 4.8, which handles nuanced inference about organizational priorities. The key is including both documents in full — models produce significantly better tailoring suggestions when they can see the complete JD rather than a summary you've paraphrased. For a broader look at how recruiters use AI for this kind of matching, see best ChatGPT prompts for HR 2026.

The resume summary is the highest-read section after your name and title. It sets the frame for everything below. A weak summary is two sentences of generic claims ('results-oriented professional with 8 years of experience'). A strong summary is a 3–4 sentence pitch that names your specialty, your most impressive quantified credential, and the problem you solve for employers.

**Prompt — resume summary writing:** ``` You are a resume writer specializing in [your industry]. Write a resume summary for me using the information below. The summary should: - Be 3–4 sentences, max 80 words - Open with my job title and years of experience in the most compelling framing - Include at least one specific quantified achievement (I'll provide my top one) - End with what I'm targeting next and why this is a natural progression - Sound like a human wrote it — no buzzword stacking, no 'passionate about' or 'dynamic professional' My background: - Job title(s): [e.g., Senior Product Manager, 7 years] - Top achievement: [e.g., led product that grew from 0 to 200k MAU in 18 months] - Skills/specialties: [e.g., B2B SaaS, growth loops, cross-functional leadership] - Target next role: [e.g., Director of Product at a Series B-C SaaS company] Write 3 different versions — formal, conversational, and achievement-first — so I can choose. ```

**Prompt — LinkedIn headline writing (bonus):** ``` Write 5 LinkedIn headline options for someone with this background: [PASTE 2–3 sentences about your role and top credential]. Each headline should be under 220 characters. Avoid 'passionate', 'innovative', and 'results-driven'. Include at least two options that lead with a specific outcome or number rather than a job title. ```

Both GPT-5 and Gemini 2.5 Pro produce strong summaries. Claude Opus 4.8 tends to write slightly more natural-sounding prose. Test all three and pick the output that sounds most like you — then edit from there rather than starting from scratch.

Most cover letters are either too generic ('I am writing to express my interest in the position') or too long. A cover letter's job is to answer three questions: Why this company? Why this role? Why me, specifically? If the letter can't answer all three with concrete evidence in under 350 words, it's not doing its job.

**Prompt — cover letter generation:** ``` You are a professional cover letter writer. Write a cover letter for me that is exactly 3 paragraphs and under 350 words total. Paragraph 1 (hook + why this company): Open with one specific thing about [company name] that genuinely drew me to this application — not generic praise. Reference something concrete like a recent product launch, their stated mission, or a market position they've taken. Paragraph 2 (why me — evidence): Pick the 2–3 most relevant experiences from my resume that directly address the top requirements of this job. Use specific numbers. Do not list everything — only the most relevant. Paragraph 3 (close): Confident, brief close. Express interest in talking. No begging, no filler. Tone: Professional but not stiff. First person. No passive voice. Do not include: 'I am a highly motivated', 'I would be a great fit', 'please find my resume attached', or any variation of these. Company: [COMPANY NAME] Role: [JOB TITLE] Key requirements from JD: [PASTE TOP 3–4 REQUIREMENTS] My most relevant experience: [PASTE 3–4 BULLET POINTS FROM YOUR RESUME] One specific thing I know/like about this company: [YOUR RESEARCH NOTE] ```

Claude Opus 4.8 and GPT-5.5 are both excellent for cover letters. The key variable is that last field — 'one specific thing I know/like about this company.' Models cannot research this for you; you have to provide it. A cover letter where the first sentence references something real about the company (a recent funding round, a product you've used, a blog post by their CEO) gets read. A generic opener gets skimmed.

LinkedIn About is read by recruiters, potential clients, and professional contacts. It's different from a resume summary — it can be longer (up to 2,600 characters), is indexed by LinkedIn's search algorithm, and benefits from a slightly more personal voice since readers expect it. The structure that works best: open with what you do and for whom, then your track record, then what you're looking for.

**Prompt — LinkedIn About section:** ``` You are a LinkedIn profile writer. Write a LinkedIn About section for me. Guidelines: - Length: 1,800–2,200 characters (LinkedIn shows the first 300 before 'see more', so the first 2–3 sentences must hook a reader) - Voice: First person, professional but human. Write like I'm talking to a smart colleague, not writing a formal bio. - Structure: (1) What I do and for whom in plain English — no job title jargon, (2) 2–3 specific accomplishments with numbers, (3) What I'm looking for or open to next, (4) One personal detail that makes me memorable (I'll provide this) - Include these keywords naturally (for LinkedIn search): [LIST 5–8 KEYWORDS FROM YOUR INDUSTRY] - Do not start with 'I am a' or list job titles like a resume Background: - Current role and company: [YOUR ROLE AND COMPANY] - Key accomplishments: [PASTE 3–5 BULLET POINTS] - Target opportunity or audience: [WHO YOU WANT TO FIND YOUR PROFILE] - One personal/memorable detail: [E.g., 'I build side projects every quarter' or 'former Division I athlete'] ```

LinkedIn's search algorithm gives weight to keywords in the About section, so including your target role title and core skills naturally in the text improves discoverability. Gemini 2.5 Pro's output for LinkedIn About tends to be slightly more natural-sounding for US professional contexts. Run the prompt, then read it aloud — if any sentence sounds like it was written by a committee, rewrite that sentence yourself.

Career changers face a specific challenge: their most recent titles and company names signal the wrong thing. A nurse applying for a UX research role, or a sales director moving into product management, needs to reframe the same experience for a different audience. The goal is not to hide the past — it's to translate it so a hiring manager sees the transferable value immediately.

**Prompt — career-change resume framing:** ``` You are a career coach who specializes in helping professionals pivot industries or functions. I am changing careers from [CURRENT CAREER PATH] to [TARGET CAREER PATH]. I need help reframing my resume so it leads with transferable skills rather than reading as a misfit. Here is my situation: - What I've done: [2–3 sentences about current career] - What I'm targeting: [specific job title and industry] - Skills/experiences that transfer directly: [list what you think transfers] - My biggest 'translation' challenge: [e.g., 'all my titles say Sales but I want Product — how do I position that?'] Tasks: 1. Identify the 5 most transferable skills from my background to my target role 2. Suggest how to reframe my job titles, if at all (e.g., add a descriptor in parentheses) 3. Rewrite my resume summary to position me as a natural fit for the target, not an outlier 4. Suggest which experiences to expand and which to minimize 5. Flag any genuine skill gaps and suggest how to address them (coursework, side projects, certifications) ```

Claude Opus 4.8 is particularly strong on career-change framing because it reasons carefully about what skills actually transfer versus what merely sounds similar. For recruiters and HR professionals who screen career changers at scale, our sibling posts on best Claude prompts for recruiters 2026 and prompt engineering for recruiting show the other side of this equation — what screeners are actually looking for.

The most common resume problem isn't weak verbs — it's achievements that lack numbers. Most people know they should quantify but feel stuck because they don't have clean metrics memorized. The prompt below pulls the numbers out by asking targeted questions.

**Prompt — achievement quantification interview:** ``` You are a resume writer helping me quantify my professional achievements. I'll give you a list of things I did at each job. For each item, ask me the most important single question that would unlock a number. After I answer, rewrite it as a polished resume bullet with the metric included. Start with the first item. Ask only one question at a time. Keep each question specific and practical — not 'what was the impact?' but 'how many users/clients/dollars/hours did this affect?' Here are my raw achievements: [PASTE YOUR UNQUANTIFIED ACCOMPLISHMENTS, ONE PER LINE] ```

This prompt works as an interactive conversation — don't dump the whole thing and ask for output. Run it turn-by-turn. The model will ask one focused question, you answer, and it writes the bullet. This approach produces better results than asking the model to guess at metrics, which leads to hedged output ('helped drive significant growth') that's no better than what you started with.

**Prompt — finding numbers you've forgotten:** ``` I need to add metrics to my resume but I can't remember the exact numbers. For each of these job duties, suggest 3–5 questions I should ask myself (or look up in my old emails/reports/dashboards) to find a real number. Then suggest what a reasonable range might be for someone at my seniority level so I know if my memory is in the right ballpark. Job duties: [PASTE YOUR DUTIES] My seniority level: [e.g., mid-level marketing manager at a 200-person B2B SaaS company] ```

Once you've polished your resume, it becomes source material for interview prep. Recruiters ask behavioral questions that map directly to the experiences on your resume. A model can generate the likely questions, help you structure STAR-format answers, and flag the gaps where your resume promises something you haven't prepared to defend.

**Prompt — generate likely interview questions from resume:** ``` You are a senior recruiter at a [industry] company. I'm interviewing for the role of [JOB TITLE]. Here is my resume. Based on what I've listed: 1. Generate the 10 most likely behavioral interview questions this resume will prompt a recruiter to ask 2. For each question, explain in one sentence why the resume invites that question 3. Flag the top 3 areas where my resume makes a claim that a prepared interviewer would probe My resume: [PASTE RESUME] ```

**Prompt — write STAR answers from resume bullets:** ``` Help me prepare a STAR-format answer (Situation, Task, Action, Result) for the following resume bullet. Ask me follow-up questions if you need more detail to make the answer specific and credible. After I provide the details, write the full STAR answer in spoken-word style (as if I'm saying it in an interview, not reading a report). Keep it under 2 minutes when read aloud (roughly 300–350 words). Resume bullet: [PASTE ONE BULLET] ```

This works best run one bullet at a time rather than in bulk. Claude Opus 4.8 and GPT-5 both ask sharp follow-up questions that surface the details you'll need to answer confidently. For the full picture on how AI can support the recruiting side of hiring conversations, see role prompts for recruiters.

The interview prep prompt pair — question generation and STAR answer drafting — is most valuable the night before an interview. Block an hour, run both prompts against each job you've applied for, and you'll walk in having rehearsed the actual questions they're likely to ask, not generic behavioral questions from a job board.

Different models have different strengths for resume work. GPT-5 is the best generalist for bullet rewriting and ATS optimization — it's fast, follows formatting instructions precisely, and handles domain-specific language well across industries. GPT-5.5 adds deeper reasoning that helps with cover letter hooks and career-change framing. Claude Opus 4.8 writes the most natural-sounding prose and handles nuanced tone calibration, making it the best choice when you need a cover letter or LinkedIn About that sounds unmistakably human. Gemini 2.5 Pro has the largest default context window and handles very long job descriptions and resumes without truncation.

Prompt length matters: for resume tasks, longer prompts with explicit constraints consistently outperform short prompts. A prompt that specifies output length, voice, structure, and things to avoid produces tighter output than 'rewrite this resume bullet.' The prompts in this guide are intentionally verbose because that specificity is what separates usable output from something you need to heavily edit. For the underlying principles, see the anatomy of a great prompt — the same structural rules that improve coding prompts improve resume prompts.

One practical note on model costs: if you're running these prompts repeatedly across 20–30 bullets and multiple job applications, token costs can add up. Our AI Prompt Cost Calculator lets you estimate the exact API cost across GPT-5, Claude Opus 4.8, and Gemini 2.5 Pro before you run a large batch — useful if you're a career coach or recruiter using these prompts at volume.

Most people who are unhappy with AI resume output made one of four prompt mistakes. Knowing them saves you from running the same bad prompt three times hoping for a different result.

**Mistake 1 — asking without constraints.** 'Improve my resume' gives the model no target to optimize toward. Every prompt should specify what 'better' means: stronger verbs, quantified metrics, ATS keywords, shorter length, more natural voice. Pick one goal per prompt.

**Mistake 2 — pasting unformatted text.** If your resume is a Word doc or PDF, copy-paste often strips formatting into a wall of text. Label sections explicitly before pasting: 'WORK EXPERIENCE:', 'EDUCATION:', 'SKILLS:'. Models produce significantly better output when they can identify section boundaries.

**Mistake 3 — asking for everything at once.** 'Rewrite my resume and make it ATS-friendly and write a cover letter and give me interview questions' produces mediocre output on all four tasks. Run separate prompts for each task. The token budget the model spends on task-switching comes out of the quality budget for any single output.

**Mistake 4 — accepting the first output.** Every prompt in this guide is a starting point, not a final output. After the first response, ask: 'Make the first bullet punchier — the current verb is weak,' or 'The summary is too formal for a startup role — make it less corporate.' Models respond well to specific revision instructions. Two rounds of targeted feedback almost always produces better output than one round of a more complex prompt.

For a systematic approach to prompt iteration, the how to write better prompts — 15 rules guide covers the feedback loop pattern in detail.

The prompts in this guide are most powerful when run in sequence rather than in isolation. Here is the order that produces the best results with the least rework: start with the quantification prompts (Section 8) to ensure every bullet has a real metric before any rewrites — this is the foundation. Then run the ATS keyword prompt (Section 2) against the job description you're targeting to know what language to use. Then run the bullet rewriting prompt (Section 1) with your quantified bullets. Then run the tailoring prompt (Section 3) to align emphasis. Last, write the summary (Section 4) and cover letter (Section 5) — in that order, because the summary informs the cover letter's framing.

This workflow takes roughly 90–120 minutes the first time, faster on subsequent applications once you have a base resume to work from. The output is a resume that is keyword-optimized for a specific role, has quantified bullets written in the format recruiters scan fastest, and is backed by a cover letter that answers the three questions hiring managers actually need answered.

HR teams and recruiters who screen these applications at scale are increasingly using AI themselves — see best ChatGPT prompts for HR 2026 for the other side of the table. Understanding how screeners use prompts to evaluate candidates gives you an advantage when writing your own application materials: you can frame your experience in the exact terms their AI and human reviewers are trained to recognize as strong signals.