We build a cover letter generator. So a review of it on our own blog has an obvious problem: you have no reason to believe us. The only way to write something useful here is to be specific about what the tool does, specific about what it does not do, and specific about the applications where you should close the tab and write the letter yourself.
That is what this is. If you want the short version: the generator is good at producing a structured, job-aware first draft in the language you need, fast. It is not good at knowing your story, and it will never be. The gap between those two sentences is the whole review.
What the tool is actually for
The generator exists to remove the blank page. Most people who stall on a cover letter are not stuck on writing ability. They are stuck on the first sentence, on which three things from a twelve-year career belong in 300 words, and on the question of how formal to sound for a company they have only read about.
It reads two inputs: your profile or CV data, and the job posting you are targeting. From that it produces a letter that maps your background against what the posting asks for, in a structure hiring managers recognise. You then edit it.
That last clause is not a disclaimer tacked on at the end. It is the design. The output is a draft. If you send it unedited, you will send something competent and forgettable, which in a competitive process is the same as not sending anything.
How it works, step by step
1. It pulls your background. Rather than asking you to retype your career into a form, it works from profile or CV data you have already given it. This matters more than it sounds. The quality of any generated letter is capped by the quality of the source material. A thin CV with three bullet points per role produces a thin letter. If your CV says "managed onboarding process" with no outcome attached, the letter cannot invent the outcome.
2. It parses the job posting. The posting is the second input, and it is doing most of the differentiating work. The tool reads what the role asks for, in what order, and with what emphasis, then decides which parts of your background to foreground. This is the mechanism behind tailored letters outperforming templates: the letter is assembled against a specific posting instead of being a fixed skeleton with the company name swapped in.
3. It drafts in your target language. Letters are produced in the language of the application, not translated from English afterwards. That distinction is real. A letter translated out of English into French carries English sentence rhythm and an English level of directness, both of which read as slightly off to a French recruiter. Drafting natively avoids that.
4. It gives you something to edit. You get a full letter, not a scaffold with blanks. From there your job is to replace the parts only you can write.
What it does well
Structure. The output follows a shape that works: an opening that connects you to this specific role, two or three body paragraphs that each carry one claim supported by evidence, a close that states what you want. Structure sounds like a low bar until you read thirty real cover letters, most of which are a chronological summary of the CV in paragraph form.
Relevance filtering. If you have had six roles and the posting is about data quality, the tool will foreground the two roles where you touched data quality and compress the rest. People are surprisingly bad at this about their own careers. We over-weight the job we are proudest of rather than the job that matches.
Volume without collapse. If you are applying to fifteen roles in a month, manual tailoring degrades. By letter nine you are copy-pasting paragraph two. The generator holds a consistent baseline across all fifteen so your effort goes into the parts that differ. This is the case where automation earns its place, and we have written about when an AI-drafted letter is and is not appropriate in more detail.
Language coverage. Applying across markets is where the manual approach breaks hardest, because writing a genuinely native letter in your third language is slow and the result is usually stiff.
Tone calibration. The register shifts depending on the sector signals in the posting. A letter for a public institution and a letter for a twelve-person startup should not sound the same, and most people only have one register.
Where it falls short
This is the part that decides whether the rest of this is worth reading.
It cannot supply your evidence. The generator can write "improved the support response time significantly." It cannot write "cut median first-response time from 14 hours to under 4 by restructuring the triage queue" unless that number is already in your CV. Specific numbers are the single strongest element in a cover letter, and they are exactly what the tool cannot invent. If you send the draft without adding them, you have sent a letter made of claims with no proof.
It does not know why you want this job. No system reading a job posting can tell a hiring manager why you left consulting, why this particular company matters to you, or what happened in 2023 that redirected your career. That is the material that makes a letter memorable, and it lives only in your head. We have written a full piece on the division of labour between what a machine can draft and what only you can write, and it applies to our own tool as directly as to anyone else's.
Openings are the weakest output. A strong opening usually depends on something contextual: a product decision the company made, a talk someone there gave, a problem in their market you happen to have solved. The generator writes a competent opening. Competent openings blend. This is why we tell you to rewrite the first paragraph every time, and why worked examples of strong hooks are worth more to you than any generated first line.
Unusual career situations are where it degrades most. A three-year gap for caregiving. A move from veterinary medicine into pharmaceutical sales. An internal application where the reader already knows you. A role you were made redundant from. In all of these, the correct letter is built around an explanation, and the generator does not know which explanation is true or how much of it to disclose. It will produce something reasonable and generic, and reasonable and generic is the wrong register for a situation that needs a direct human account.
It can smooth away your voice. Generated prose trends toward the middle: clean, even, slightly flat. If you write well, the draft may be worse than what you would have produced yourself. Some people should use the tool only for structure and then rewrite nearly all of the sentences.
If you need one letter, you probably do not need this. One application, an hour of attention, a job you actually care about: write it yourself. You will produce something better than any draft you would then have to heavily edit anyway. The tool earns its place at volume, or across languages, or when you are genuinely blocked.
Who it fits and who it does not
Good fit: you are applying to five or more roles a month; you are applying in a language you speak but do not write fluently for professional purposes; you have a solid CV with real outcomes in it; you freeze at the blank page but edit well once something exists.
Poor fit: you are making a single application to a role that matters enormously; your CV is thin and vague, so there is nothing substantial for the tool to work from; your situation needs explaining rather than presenting; you are a strong writer who finds editing someone else's prose slower than writing your own.
How to get a good result from it
The editing pass is not optional, and it is short. Fifteen minutes, in this order.
- Rewrite the opening completely. Every time. Replace the generated first paragraph with something only someone who looked at this company could write. One concrete reference is enough.
- Rewrite at least one body paragraph in full. Pick the one carrying your strongest claim and rebuild it around a real number, a named system, a specific outcome.
- Delete one paragraph. There is almost always one that says nothing. Cutting it makes the letter shorter and better.
- Add one thing the machine could not know. Why this company. What you want next. A connection to their market you happen to have.
- Read it aloud. Any sentence you would not say out loud, cut or rewrite. This catches the flatness quickly.
If you are weighing the tool against paid alternatives, the mechanisms matter more than the marketing, and we compared what paid generators actually give you for the money without special pleading for ourselves.
Honest verdict
The generator solves a real and narrow problem: producing a structured, posting-aware, natively written first draft quickly, repeatedly, in more than one language. Within that boundary it does its job well and saves a meaningful amount of time.
Outside that boundary it is the wrong tool. It cannot tell your story, cannot supply evidence you never wrote down, and cannot handle a career situation that needs explaining. Used as a finished product, it will make your applications more uniform and less persuasive, which is the failure mode we describe in our piece on generic versus tailored letters.
Use it as a first draft. Rewrite the opening and one body paragraph yourself, always. If you do that, it is genuinely useful. If you do not, you would be better off with a blank page and an hour.