How to use ChatGPT for a cover letter that still sounds like you

There are two ways to use ChatGPT for a cover letter, and they produce opposite results. The first is to type "write me a cover letter for this job," paste the posting, and send whatever comes back. The second is to use it as a fast, tireless drafting partner that helps you organise real material, and then to do the actual writing yourself. The first way produces the exact letter hiring managers have learned to skim and discard. The second way saves you an hour and produces something worth reading.

The reason the lazy way fails is not that AI writing is detectable by some tool. It is that a cover letter has one job, which is to make a specific person want to meet a specific candidate, and generic prose cannot do that job. When ChatGPT writes with no real detail to work from, it falls back on the safe, hedged, everyone-says-this register: passionate about, excited by the opportunity, proven track record, thrive in fast-paced environments. A reader who has seen four hundred letters recognises that register instantly, and it reads as effort not taken.

So the mental model matters before any prompt does. ChatGPT is very good at structure, at getting words onto the page, at reorganising and tightening. It is bad at being you, because it has never met you and is guessing at a plausible person. The workflow in this post plays to the first and refuses the second: let it draft, and keep the voice, the specifics, and the honesty firmly in your hands.

The single biggest lever on the output is what you put in. A one-line prompt asking for a cover letter forces the model to invent everything, and it invents the blandest plausible version, because bland is safe when it knows nothing. A rich prompt, one that includes the actual job description and three or four of your genuine, concrete achievements, gives it real material to arrange, and the draft that comes back is specific enough to be worth editing. Same tool, completely different result, entirely because of what you fed it.

This is why the goal is never to make ChatGPT sound impressive in the abstract. It is to give it enough truth that it does not have to make anything up. Before you prompt, spend five minutes gathering the raw material: the responsibilities the posting emphasises, two or three things you have actually done that map to them, and one real reason you want this particular role at this particular place. That preparation is the work; the prompt just arranges it.

A useful rule is that the model should never be the source of a fact. It arranges facts you supply. The moment you let it fill a gap with an invented metric or a responsibility you did not have, you have handed yourself a claim you cannot defend in the interview, and you have traded a small time saving for a real risk. Real material in is not just a quality tip; it is the thing that keeps the letter honest.

People worry about detection tools, but the real flag is not a software verdict; it is a reader noticing the letter could have been sent to any company on earth. The most common tell is uniform blandness: three paragraphs of competent, grammatical, utterly generic prose with no specific detail and no discernible voice. It is not that the writing is bad. It is that it is anonymous, and a hiring manager reads anonymous as a candidate who did not care enough to say something only they could say.

The second tell is over-enthusiasm with no anchor. Unedited AI reaches for superlatives it has not earned: thrilled, delighted, deeply passionate, the perfect fit. Real people making a real case tend to be more measured, because measured is what confidence sounds like. A letter that gushes without a single concrete reason behind the gushing reads as filler, and filler is exactly what the enthusiasm register is covering for. The third tell is structural sameness, the identical setup-body-closing shape with the identical transitions, which shows up when several applicants all lightly edit the same model output.

The fourth and most damaging tell is a claim that does not survive contact. AI will write "increased revenue by 40 percent" or "managed a team of twelve" if it thinks it fits, and if you leave it in and cannot defend it, the interview is where it comes apart, along with your credibility on everything else. The cure for all four tells is the same and it is not a tool: real material, an honest voice, and one specific detail that ties this letter to this job and no other. Do that editing pass and the question of whether a machine helped stops mattering, because what you send is unmistakably yours.

A cover letter does not travel alone. It points at a resume and, increasingly, at a portfolio or a site under your name, and its job is partly to make the reader want to open those. ChatGPT can help you draft the letter quickly, but the things it points to have to be real, current, and easy to reach. A great letter attached to a stale resume or a dead link wastes the hour you just saved. Get the letter drafting fast so you can spend the time you save keeping the assets behind it sharp.

It also helps to keep a master version of everything. A base resume you tailor per role, a portfolio that shows the work the letter claims, and one link that ties them together mean each new application is an edit rather than a rebuild. That is where the real time saving in a modern search comes from: not from letting a model write for you, but from having your material organised well enough that adapting it is quick and honest.

Folio is one place to keep that together: a resume with a deterministic, published ATS score, a portfolio site, and a contact inbox, all under one link you can point a cover letter at. On the free plan the site lives at portfolio.wrxstack.com/yourname with a small Made with Folio badge, the full theme gallery is on the paid tier, and the resume exports as PDF and DOCX with no watermark. Let ChatGPT get you past the blank page, do the editing pass that makes the letter yours, and point it at material that is genuinely worth opening.