Which tool to choose
You don't need to try them all. Choose based on where you're stuck.
- You already have a CV and a posting and want a draft in five minutes: free ChatGPT. Free users can search the web, upload files, and use GPTs: for a letter that's more than enough. The real limit is the cap on messages every few hours, after which you drop to a reduced model. For a single letter you won't hit it.
- You're worried about the automatic application filter (ATS): a dedicated tool like Kickresume. ATS is the software that pre-screens CVs by keyword and rewards those who use the exact terms from the posting. Kickresume builds the letter around those words: you paste in the job description and it aligns the text with the terms the software is looking for.
- You already have Microsoft 365: Copilot inside Word writes the letter without making you switch programs, from scratch or starting from an ATS-compatible template.
For most people the answer is free ChatGPT: no paid account, no learning curve. The rest of this guide works with that.
How to do it
The secret isn't "ask for a letter." It's giving the AI the two ingredients it's missing: who you are and what that company is looking for. ChatGPT writes cover letters, but the result is only as good as the prompt.
- Open ChatGPT (chatgpt.com) or the app. The path is the same whether from a browser or the app.
- Have two pieces of text ready to paste: the full job posting and your CV (or even just a bulleted list of your experience).
- Paste the prompt below, replacing only the two indicated sections. Leave everything else as it is.
- Read the draft. If it sounds generic, reply in the same chat: "Rewrite it more directly, cut the empty adjectives, and start from a concrete result."
The operational syntax:
You are a career advisor. Write me a cover letter
in English, in the first person, for the position described below.
Rules:
- Maximum 250 words, three paragraphs.
- Open with a specific sentence about the company or the role, not with
"My name is... and I'm applying for the position of...".
- In the middle paragraph use a concrete result of mine with a number
or a real example taken from my CV.
- Professional but human tone, short sentences, no expressions
like "dynamic", "proactive", "team player".
- Close with a clear request for an interview.
=== POSTING ===
(paste the text of the job posting here)
=== MY CV ===
(paste your CV or a list of your experience here)
- Copy the text, read it aloud, and make two changes by hand: change the first sentence into your own words and add a detail the AI couldn't have known (why you want that exact company, a project you follow, a member of the team you admire).
Feedback: if by the end of the read you yourself don't recognize your voice, the recruiter won't either. Those two manual touch-ups separate an application that gets read from one that gets discarded.
A concrete example
Marta, 29, is applying for a marketing role at a small cosmetics company. She pastes in the posting (which asked for "social media management and campaign data analysis") and her CV, which said "managed the Instagram profile for a local bakery." ChatGPT returns a middle paragraph that reads roughly: "Over the past year I managed the Instagram profile of a local bakery, growing followers from 800 to 3,200 and increasing online bookings by 40%."
That "40%" was neither in the posting nor in the CV: the AI made it up. Marta corrects it with the real figure (bookings had grown "noticeably, especially on weekends") and rewrites the opening, citing a product from the brand she actually uses. Result: a letter delivered in twenty minutes instead of two hours, with a hook no generic model would have given her.
When it does NOT work (and how to fix it)
If the AI invents numbers and results
This is the most serious risk: the model fills the gaps with plausible but false statistics, like the "40%" in the example. Never sign a letter with data you can't back up in an interview. Reread every figure and replace it with the truth, even if it's less flashy. A real result described in words beats a fake number.
If the letter sounds like a thousand others
This is the number one problem according to recruiters: too many letters arrive recognizably generic or AI-spat-out without a single edit. The way out is step 5: change the opening and the closing, insert a personal anecdote. Ten minutes of manual editing take you out of the pile.
If you run out of free messages
This happens if you regenerate the letter many times. The free plan has a cap on messages per time window. Fix: instead of asking for ten versions, write a single message with all the corrections together ("shorten it, change the opening, and make the second paragraph more concrete"). Or switch to another free assistant like Gemini or Claude to finish the job.
If the text is in overly formal or robotic English
The models tend toward a stiff tone. Add to the prompt: "Write as I would speak to a colleague I respect, not like a company memo." It works better than any request for a "professional tone."
A tip from someone who actually uses it
Don't have the AI write the letter: have it structure it. First ask for a bulleted skeleton ("just give me the three key messages I should communicate to this company, based on my CV"), then write the paragraphs yourself and use the AI only to polish the form. The application stays yours, and you can defend every sentence in an interview. It counts: most hiring managers consider the letter an important part of the decision.
Frequently asked questions
Should I attach the letter even if the posting doesn't require it?
Almost always yes. The majority of recruiters prefer candidates who send it even when it's optional. Omitting it when it is required is worse: very few decision-makers will consider an application that lacks the required letter.
Is the free version of ChatGPT enough?
For a letter, yes. Free users can search the web, upload files, and use GPTs: it covers personal use. The paid plan only makes sense if you're generating dozens of applications a day.
What's the difference between a cover letter and a motivational letter?
In practice they're the same thing. It's a short document that accompanies the CV: the candidate writes it to round out the resume with skills, experience, and motivations for the role. The pure "motivational" letter (for university or internships) leans a bit more on why you want that particular path.
Do recruiters notice I used AI, and do they penalize me?
They notice when you used AI and nothing else: the raw draft is recognizable and ends up discarded. The problem isn't the tool, it's handing in the first draft without touching it up. A letter built with AI but rewritten in your voice, with your real data and a personal detail, is indistinguishable from one written by hand from scratch, because you made it yours. AI saves you from the blank page, not from the thinking.