Which prompt to choose
The CV, the cover letter and the online profile (LinkedIn and the like) have different rules. The CV has to get past the automatic filter and then convince a human in a few seconds; the letter has to explain the why; the profile has to get you found by whoever is searching. Choose where to start.
- Adapt the CV to a specific ad: first prompt, which aligns your words to those of the position.
- Write the cover letter: second prompt, which starts from the company's problem, not from you.
- Fix up the LinkedIn profile: third prompt, which makes it findable and readable.
How to do it
- Open the AI assistant.
- Give it two raw materials: the text of the ad (copy all of it) and your real experience. The AI rephrases, it doesn't invent: you have to provide the facts.
- Copy the prompt and send.
- Reread and cut anything you couldn't defend in an interview. If you can't really do it, it comes off the CV.
The operational syntax to adapt the CV to an ad:
Act as a recruiter expert in the field of this position.
Here is the job ad: "[paste the complete ad]".
Here is my current experience: "[paste the entries from your CV]".
Do two things:
1. Extract from the ad the 15-20 keywords and skills the selection software will look for.
2. Rewrite my experience using those words WHERE they truly match what I did.
Constraints: use exclusively skills I truly possess. If a keyword from the ad finds no match in my experience, flag it for me instead of inventing.
The operational syntax for the cover letter:
Act as a sharp candidate writing a short, targeted letter.
Position: "[role and company]". What the company does: "[two lines]".
My 3 most relevant experiences: "[list]".
Write a letter of 200 words maximum that:
- Opens from the company's problem or objective, not from "my name is and I'm looking for a job".
- Connects 2 of my concrete experiences to what they need.
- Closes with an invitation to talk, without flattery.
Constraints: no clichés like "dynamic team" or "stimulating challenge". Direct and professional tone.
The operational syntax for the LinkedIn profile:
Help me improve my LinkedIn profile to be found by whoever is searching for my role: "[role]".
Here is my current summary: "[paste]".
Give me:
1. A profile headline that says what I do and for whom, with the words recruiters search for.
2. A summary rewritten in the first person, 3-4 sentences, that reads in 10 seconds.
3. 10 key skills to add in the dedicated section.
Constraints: no empty buzzwords, no "passionate about innovation". Concreteness.
After rewriting the CV, do the decisive check: for each entry, could you tell a real example in an interview? If not, that entry is inflated and should be scaled back. You get past the automatic filter with words; the human after, you lose with lies.
A concrete example
Marco, a warehouse worker, wanted to move into a department supervisor role. His CV said "warehouse management". The ad asked for "team coordination", "stock management", "flow optimization". He pasted both into the first prompt.
The AI found that Marco already did those things, he just called them by different words: "I organize my colleagues' shifts" is team coordination, "I keep stock levels under control" is stock management. It rewrote the entries with the ad's terms, all true. One keyword, "specific management software", found no match: the AI flagged it instead of inventing it, and Marco decided to take a course. The CV went from the automatic trash bin to the interview.
When it does NOT work (and how to fix it)
If the AI invents skills you don't have
It's the most serious risk here: a lie on the CV collapses at the first interview. Lock down the prompt with "use exclusively the skills I provide; flag the ad's keywords that find no match, don't fill them in". And reread every line yourself, asking whether you can prove it.
If the CV becomes an unreadable list of keywords
Stuffing it with terms for the software makes the text indigestible for the human who reads it afterwards. Recent systems, moreover, understand the meaning and no longer reward accumulation. Ask "integrate the keywords into natural sentences that describe results, not into a list". A concrete result is worth more than ten scattered keywords.
If you can't find where to add skills on LinkedIn
The menus change often. Look for the "Skills" section from the profile edit; if you can't find it under that name, ask the AI "explain step by step where to add skills in the current version of LinkedIn" and follow the clue, not the exact button number.
A tip from someone who actually uses it
Don't have just one CV: keep a long version with everything you've done, and for each application use the AI to tailor a short version calibrated to that ad. Five minutes per application get you past the filters far more than a single CV sent out en masse. Whoever adapts gets through, whoever spams ends up in the pile.
Frequently asked questions
Do recruiters tell whether the CV is written with AI?
What they notice isn't the AI, it's the genericness. A CV full of clichés and devoid of concrete results reeks of laziness, written by hand or not. Use the AI for the form and the right words, but put in the numbers and the real examples yourself: those are what set you apart.
Do I have to use the exact words of the ad?
Where they match what you did, yes: the software looks for those. But only if they're true. Copying a skill you don't have to get past the filter gets you to an interview you'll fail. The goal is to get found for what you can really do, not to sneak in.
Does putting in lots of keywords increase my chances?
No, and it's the misunderstanding to debunk. The 2026 selection software no longer counts repetitions the way it used to: it reads the meaning, weighs the context, discards obvious accumulation. Stuffing the CV with terms achieves the opposite effect, and repels the human who reads after the machine. Fifteen keywords well integrated into real results beat fifty scattered at random.