A cover letter draft built from your actual resume.
Most AI cover letters read like AI cover letters for one reason: the model was given a job description and nothing else to be specific about. Here is what changes when it has your real experience to work from.
Draft a cover letter →Free after an analysis · One free credit included · No subscription
The actual problem
Generic output is a missing-input problem, not a model problem.
Paste a job description into a general-purpose chat assistant and ask for a cover letter. What comes back is competent, well-organised, and completely interchangeable. It says you are excited about the opportunity, that you thrive in fast-paced environments, and that your skills align well with the requirements.
The model is not failing. It is doing the only thing available to it. Given no information about you, every specific claim it could make would be an invention, so it retreats to statements that are safe because they are true of anyone. Recruiters recognise that register instantly, and increasingly so, because they read a great many of these now.
Give the same model your resume alongside the posting and the constraint disappears. It can name the migration you led, the scale you operated at, the team you mentored. Specificity is what the letter was missing, and specificity has to come from somewhere real.
What grounded means
The draft is built from your matches, not from the posting alone.
In JobForte a cover letter is generated after a fit analysis, not instead of one. The analysis has already worked out which parts of your background genuinely answer this posting's requirements. The letter is then written from those specific matches.
Input
Your strongest genuine matches
The draft is built around the points where your experience actually meets a stated requirement, in the order the analysis ranked them. Not keyword overlap, and not a summary of the posting back at itself.
Structure
Three paragraphs, 120 to 180 words
An opening with one specific hook from the posting, a middle connecting two or three strengths to real requirements, and a closing that asks for the conversation. Short enough to be read in full.
Honesty
No invented specifics
The draft never manufactures a metric or an anecdote your resume does not support. Where one would genuinely help, it leaves a marked placeholder for you to fill instead.
Editability
Written to be rewritten
Each paragraph survives you swapping or rewriting any single sentence, so making it sound like you does not mean starting over.
The draft in the app
It tells you it is a draft, because it is one.
That notice sits above every draft on purpose. A tool that implies its output is ready to send is setting you up to be the fourth candidate that week to open with the same sentence.
Three drafts are available per job, free, once the job has been analysed. Re-analysing the job resets that allowance and costs one credit, the same one that produces the analysis.
Making it yours
Four edits that do most of the work.
- Replace the placeholders first. They mark exactly where a specific of yours raises the letter most. Leaving one in is worse than the sentence not existing.
- Read the opening aloud. If it is not a sentence you would say to a person, rewrite it. The first line decides whether the rest gets read.
- Cut one claim. Drafts tend to make one connection too many. The letter gets stronger when the two best points stand alone.
- Check every specific against your resume. Anything you cannot expand on for two minutes in an interview should not be in the letter.
If the analysis behind the draft turned up requirements you do not meet, that is worth knowing before you write anything. The resume checker covers what the fit analysis returns, and job description red flags covers what the posting's own wording tells you.
Questions
Why do AI cover letters sound generic?
Because the model is usually given a job description and nothing else. With no specifics about you, the only safe thing it can write is something that would be true of any competent applicant, which is exactly what reads as filler. The fix is not a better prompt or a more advanced model. It is giving the draft real material to be specific about: your actual experience, matched against the requirements of this particular role.
Should I send an AI-generated cover letter as it is?
No, and no honest tool should suggest otherwise. A generated draft solves the blank page, gets the structure right, and puts your strongest matches in a sensible order. What it cannot supply is your voice or the specific anecdote that makes a paragraph land. Read it aloud, replace anything you would not say, and fill in the specifics only you know. The draft is a starting point, and treating it as a finished letter is how people get caught out.
How long should a cover letter be?
Three short paragraphs, somewhere between 120 and 180 words in total. Long enough to make a specific case, short enough that a hiring manager reads all of it. A page of prose restating your resume is worse than nothing, because it signals you did not know what to cut. The opening earns the read, the middle connects two or three concrete strengths to real requirements, and the closing asks for the conversation.
What are the bracketed placeholders in the draft?
They mark the spots where a specific detail would strengthen a claim but only you have it, such as a metric, an outcome, or a short anecdote. The draft deliberately leaves a marked gap rather than inventing something plausible, because an invented specific is a fabrication you would have to defend in an interview. There are at most two, and both are meant to be replaced before you send anything.
Does the cover letter cost a credit?
No. Generating a cover letter is free once a job has been analysed, up to three drafts per job. Re-analysing the job resets that allowance and costs one credit, which is the same credit that produces the fit analysis itself. Every new account starts with one free credit.
Free after an analysis · One free credit included · No subscription