Optimize your resume with AI, step by step
You apply, and then silence. Most of the time, it's not you: a robot rejected your resume before a human ever saw it.
Good news: used well, AI beats that robot. Here's the complete method we use ourselves: free, with AI (ChatGPT, Claude, Gemini...), holding nothing back. Copy-paste prompts included.
This is the article our sales team would have preferred we never publish. We're going to show you, free and in full, how to build by hand what JobAlign does for you: turning your LinkedIn profile into a resume that passes the ATS filters, prompts included, with ChatGPT, Claude, or Gemini. The complete playbook, so you can do it all yourself and never pay a cent. We're just betting that once you've timed the hour it takes per job posting, our 3 minutes will suddenly look very reasonable.
What's inside
- 1. Why AI changes the game (and the ATS)
- 2. Step 1: Gather the raw material
- 3. Step 2: Decode the job posting
- 4. Step 3: Structure and prioritize
- 5. Step 4: The master prompt
- 6. Step 5: Refine section by section
- 7. Step 6: ATS formatting
- 8. Step 7: Export a clean PDF and test it
- 9. The mistakes that sink your resume
- 10. By hand or automated? The verdict
Why AI changes the game
Generative AI (ChatGPT, Claude, Gemini...) is exceptional at rewording, structuring, and aligning your resume to a posting, as long as you steer it well. What it can't do: guess your real numbers, and produce a PDF that an ATS can actually read. We'll get back to that. For now, the golden rule: AI rewords, it never invents.
Gather the raw material
"Garbage in, garbage out": the quality of your resume depends first on what you feed the AI.
β’ Your experience: dates, job titles, companies, concrete responsibilities.
β’ Your quantified results: %, $, volumes, timelines, team size. This is the AI's fuel: a resume without numbers stays vague.
β’ Your education, your skills, and certifications.
The easiest path: start from your LinkedIn profile. You can export it as a PDF and copy the text, or follow our LinkedIn to resume walkthrough. A well-kept profile saves a ton of time: which is exactly why it pays to optimize your LinkedIn profile first.
Decode the job posting
An optimized resume is a resume aligned to ONE specific posting. First step: extract what the posting really wants.
Keep this result handy: you'll reuse it as-is in Step 4, without pasting the whole posting again. To go deeper, see ATS resume keywords and where to place them.
Prompt #1: Analyze the posting
Turns the posting into an ATS scoring grid: you know exactly what to aim for.
You are a recruiting expert. Analyze the job posting below and return, as clear lists: 1. The exact job title. 2. The 10 to 15 most important keywords and skills (technical and behavioral). 3. The tools / software / certifications mentioned. 4. The experience level and degrees expected. 5. The 3 criteria that seem to be the priority (the ones repeated or emphasized). Job posting: [PASTE HERE THE FULL TEXT OF THE POSTING]
Structure and prioritize
The ATS reads in order. What matters most for the posting has to come first.
β’ Summary (2-3 lines) with the title of the role you're targeting.
β’ Experience: the most relevant entries for this posting first, not necessarily the most recent.
β’ Skills aligned to the posting, then education.
Cut anything that doesn't help for THIS role, no mercy. A targeted resume beats an exhaustive one. This is exactly the logic behind a resume tailored to each posting.
The master prompt
This is the heart of the method: a single prompt that rewrites your resume, aligned to the posting and readable by an ATS.
Don't stop at the first output: it's a draft. The next steps are there to refine it and verify every fact.
Prompt #2: The master prompt (full resume)
The heart of the method: it starts from the Step 2 analysis (not the raw posting) and your real data.
You are an expert in recruiting and in resume optimization for ATS systems. MY RAW INFORMATION: [PASTE HERE your background: experience (dates, titles, companies, responsibilities, quantified results), education, skills, certifications] POSTING ANALYSIS (result of Step 2): [PASTE HERE the list of keywords and requirements you got from the analysis prompt. If you're in the same conversation, just write: use the analysis above.] MISSION: based on this analysis, rewrite my resume so it's ATS-optimized and aligned to the posting. NON-NEGOTIABLE RULES: - Integrate the exact keywords and titles from the analysis when they match my real experience. - Turn each role into 3 to 5 bullets: action verb + task + quantified result. - Put the most relevant experience and skills first, based on the analysis. - Clean, machine-readable style: no tables, no columns, standard headings (Experience, Education, Skills). - DO NOT INVENT any information. If a detail is missing, flag it in [brackets] instead of creating it. - Language: English. Target length: 1 page (2 if more than 10 years of experience). RETURN: (1) a 2-3 line summary, (2) the rewritten resume section by section, (3) the list of keywords from the analysis that you integrated.
Refine section by section
The master prompt does the rough cut; the variations sharpen each part.
Prompt #3: The summary
The first 3 lines decide whether they keep reading. Three versions beat one.
Based on my background and the posting above, write 3 versions of a resume summary (2 to 3 lines each) for the [JOB TITLE] role.
- Results-focused, concrete, no empty superlatives ("passionate", "dynamic"...).
- Naturally weave in the keywords [KEYWORD 1, KEYWORD 2, KEYWORD 3] from the posting.
Give me a short version, an expertise-focused one, and an impact-focused one.
Prompt #4: Quantified bullets
The format ATS systems and recruiters love: action -> result -> number.
Reword the responsibilities below into 4 punchy bullets, in the format: action verb + what I did + measurable result. - If I didn't give a number, suggest in [brackets] the TYPE of metric to add (%, $, count, timeline): without inventing a value. - One idea per bullet, no useless jargon. Responsibilities: [PASTE HERE the responsibilities for one role]
Prompt #5: Skills section
Aligns your skills with the exact wording of the posting: without overdoing it.
Compare my skills list to the posting and return 3 lists: (a) skills to keep as-is, (b) skills to reword using the EXACT wording from the posting, (c) skills the posting expects that I have but didn't list (to add only if it's true). My skills: [PASTE] Posting: [PASTE]
Prompt #6: Critical proofread
The safety net: zero invention, zero keyword stuffing, readable sections.
Proofread this resume critically and return your notes: 1. Flag any inflated or unverifiable claim. 2. Spot keyword stuffing (keywords repeated artificially) and suggest a natural version. 3. Check that every section has a standard heading an ATS can read and that nothing important is missing for the posting. Resume: [PASTE HERE your resume]
ATS formatting
This is where most AI-generated resumes fail. Your AI assistant gives you TEXT, not a layout.
β’ A single column (columns scramble everything for a machine).
β’ Standard section headings: Experience, Education, Skills.
β’ No tables, no images, no icons, no text boxes.
β’ Classic fonts, simple date format, standard bullets.
Paste the text the AI produced into a clean template, not a busy graphic one. For bases already built for ATS, see our ATS-optimized resume templates.
Export a clean PDF and test it
The home stretch: produce a PDF the machine can read, and check it.
Open your PDF, hit select all, then copy-paste into a blank text editor. If the text comes out scrambled, with lost characters or mixed-up columns, the ATS will see the same mush. Rework the formatting until the copy-paste comes out clean and in order.
Finally, run through the checklist below before you send it.
The mistakes that sink your resume
β Letting the AI invent
Fake roles, numbers, or skills: it gets caught in the interview and wrecks your credibility. AI rewords, you verify.
β Keyword stuffing
Repeating "project management" 12 times no longer fools modern ATS systems and makes the resume unreadable. Natural first.
β The same resume for every posting
This is mistake #1. AI makes tailoring fast: use that to personalize for every application.
β A graphic layout
Columns, icons, text boxes, photos: pretty to the eye, unreadable to the machine. Clean = safe.
β Not testing the PDF
If the copy-paste from your PDF comes out scrambled, so does the ATS. Test before you send.
By hand or automated?
You now have all the keys. The method works, and it's free. Let's be honest about what it costs.
That's exactly what we automated. JobAlign does Steps 1 through 7 in 3 minutes: LinkedIn import in 1 click, optimized prompts built in, posting analysis, guaranteed ATS format (87% pass rate), and a clean PDF. The LinkedIn resume is free and unlimited; only the tailored resumes built for a specific posting are paid, from β¬1.25, no subscription.
If redoing all this by hand doesn't bother you, go for it: we'll be glad we helped. If your time is worth more than that, you know where to find us.
AI by hand vs JobAlign
The same method, two ways to run it.
| Step | AI by hand | JobAlign |
|---|---|---|
| LinkedIn data import | Manual copy-paste | 1 click (extension) |
| Posting analysis | Prompt + manual reading | Automatic |
| Optimization prompts | Master them yourself | Optimized, built in |
| ATS formatting | Do it by hand (risky) | Guaranteed (87%) |
| ATS-readable PDF export | Not guaranteed | Clean PDF, ready to use |
| Tailor to a new posting | Start over | A few seconds |
| Time per resume | 45-60 minutes | Under 3 minutes |
| Cost | Free (your time) | LinkedIn resume free; tailored from β¬1.25 |
The checklist before you send
- β The key keywords from the posting appear naturally in the resume
- β Each role has "action + result + number" bullets
- β No invented information: everything is verifiable
- β Single column, standard section headings, zero tables/images
- β The copy-paste from the PDF comes out clean and in order
- β The most relevant experience for the posting is at the top
- β The file is named First-Last-Resume.pdf
- β It all fits on 1 page (2 max if 10+ years)
Frequently asked questions
Can ChatGPT really build a resume that passes the ATS?
For the text and the keywords, yes, very well. Its limit is the formatting: ChatGPT, like Claude or Gemini, produces text, not a PDF guaranteed to be readable by an ATS. Steps 6 and 7 are what make the difference, and that's exactly where most AI-generated resumes fail.
What's the best prompt to tailor your resume to a posting?
The "master prompt" (Prompt #2 in our library): it takes your real data and the posting, then rewrites the resume with the right keywords, inventing nothing. Round it out with the per-section prompts to refine.
Could the AI invent things on my resume?
Yes, if you don't keep it in check. All our prompts forbid it from inventing and ask it to flag gaps in [brackets]. Absolute rule: you verify every line before you send.
Is it free?
The manual method described here is 100% free (you just need access to ChatGPT or an equivalent). On the JobAlign side, the resume optimized from your LinkedIn profile is free and unlimited; only resumes tailored to a posting are paid, from β¬1.25, no subscription.
How long does it take by hand?
Count on 45 to 60 minutes per posting the first time, a bit less after that. The painful part: you have to start over for every application, whereas a dedicated tool reuses your profile.
Keep the method. Win back the time.
You now know how to beat the ATS by hand. If you'd rather take the 3 minutes than the 60, JobAlign does all of it for you: the LinkedIn resume is free, tailored resumes from β¬1.25.
Create my optimized resumeRelated guides
ChatGPT vs JobAlign
Detailed ChatGPT vs JobAlign comparison for creating resumes. Why a specialized tool withβ¦
Tailor Resume to Job Posting
Complete guide to tailor your resume to each job posting and pass ATS filters. Concrete stepsβ¦
Resume Keywords ATS
Practical guide to identify the right ATS keywords in a job posting and integrate them intoβ¦
ATS Guide
Complete guide on ATS (Applicant Tracking Systems): how they work, limitations, and bestβ¦