Using ChatGPT for a Cover Letter

Cover letters are the worst case for general-purpose AI drafting and the most common use of it. The reason is structural: a letter’s value is entirely in specifics, and specifics are what a model does not have.

The failure mode

Ask a chatbot for a cover letter given a job title and a company name, and it produces something like this.

I am writing to express my strong interest in the Operations Manager position at Vance Logistics. With a proven track record in operational excellence and a passion for driving efficiency, I am confident I would be a valuable addition to your team. My experience has equipped me with strong leadership, communication and problem-solving skills, and I am particularly drawn to Vance’s commitment to innovation in the logistics space.

Fluent, grammatical, and empty. It contains no fact, and it could be sent to any company for any operations role.

This is not a detection problem. A recruiter does not need a tool to notice that a letter says nothing; they simply move on. The cost is not being caught, it is being unremarkable.

Why letters are worse than resumes for this

Two reasons.

A resume has structure to fall back on. Even a generically phrased bullet carries an employer, a date and a task. A letter has no structure, so when the specifics are absent there is nothing left.

A letter is what gets quoted at you. Interviewers open with it. “You mentioned you cut turnaround by 40%, tell me about that.” An invented figure on a resume is a liability; the same figure in a letter is the first question you will be asked.

The prompt that works

The principle is the same as for resumes: supply everything, ask only for structuring, forbid invention explicitly.

Structure the material below into a cover letter of about 250 words, in four short paragraphs.

Rules: use only facts I provide. Do not add numbers, achievements, company praise, or claims about my skills that I have not given you. Do not include any sentence about enthusiasm, passion, or admiring the company. If a paragraph would be stronger with information I have not supplied, leave a bracketed note telling me what to add. Plain language, varied sentence length, no marketing adjectives.

The posting: [paste the full posting]

The role’s first stated requirement: [copy it out]

My example that answers it, in rough notes: [what the situation was, what you did, what changed, with real numbers, including the awkward parts]

Anything needing explanation: [gap, relocation, career change, or none]

The clauses that do the work are the ban on enthusiasm and the instruction to leave bracketed notes rather than invent. Without the first you will get a paragraph of admiration; without the second you will get numbers you never provided.

Write the notes first

This is the step that determines everything, and it is where the actual thinking happens.

Before opening any tool, write out one example in messy prose: what the situation was, what you did, what changed, what was harder than expected, and any real number. Include the awkward detail. In the letter about supplier consolidation, the sentence about persuading two staff members is what makes it credible, and no model can produce that because it does not know it happened.

A chatbot given rich notes produces a decent letter. Given a job title, it produces the paragraph above.

The three edits after it drafts

One: delete every unsourced claim. Any number, name or outcome you did not supply comes out, not softened. Check this first because it is the one with consequences.

Two: replace the generic sentences with a specific one. There will be at least one sentence listing qualities. Cut it and put a fact in its place, even if the letter gets shorter. Shorter and specific beats longer and general. See the cover letter length guide.

Three: read it aloud. This catches both the uniform rhythm and any phrasing you would not say in an interview. If a sentence would embarrass you when read back to you, it goes.

Then check the opening in isolation. Purdue’s cover letter guidance emphasises targeting the specific role, and the opening is where AI drafts fail that test most reliably. See how to start a cover letter.

What it is genuinely good for

Four uses that do not involve generating content about you.

Reading the posting. Paste it and ask what the employer appears to prioritise and in what order. A useful reading aid.

Cutting. Give it your own 400-word letter and ask it to reduce it to 250 without losing any specific fact. It does this well.

Structure checking. Ask whether each paragraph has exactly one topic. It is reliable at spotting the paragraph doing three jobs.

Tone flattening. If your draft is too eager or too stiff, asking for a plainer register works.

In all four you supply the content and it works on the form, which is the only safe division.

Two cautions

On what you paste. A cover letter plus a resume is a concentrated packet of personal information. Check what a tool retains and whether it is used for training before uploading it.

On tools promising results. There is a market of products offering AI cover letters with claimed response-rate improvements. Performance claims in advertising are expected to be substantiated, which is the subject of the FTC’s business guidance. Treat an unexplained multiplier as marketing.

Related

For the resume equivalent, see the ChatGPT resume guide. For structure, see the cover letter outline guide, and for graded examples the best cover letter examples. For a purpose-built alternative, see AI resume builders.

Or use the cover letter generator, which works from your own material, free at Kazifi.