The short answer
An AI cover letter reads as AI-written because it is generic, not because a tool wrote it. It opens on the applicant, praises instead of proving, and could be sent to any employer. Give the AI one job listing, the employer’s current problem and your results with numbers, then set firm limits: under 200 words, open on the employer’s problem, match the listing’s top three requirements to three proven results, and cut every adjective that has no number behind it.
What a cover letter is for
A cover letter answers one question: why this candidate, for this job, at this company. The résumé lists what you have done; the letter connects the two or three most relevant results to what this employer needs now.
A letter that answers the question has four parts, in this order:
- The employer’s problem. One or two sentences on what the company is dealing with that this hire is meant to fix, with a fact that shows you looked.
- Three matches. The listing’s three most stressed requirements, each answered by one result of yours, with its number.
- The reason you are writing. A referral, if you have one, named in a sentence.
- One specific request. A short conversation about a named part of the job, not “I look forward to hearing from you”.
That fits in under 200 words. Many employers now receive the letter as the body of an email with the résumé attached, and a letter that fits on one screen is more likely to be read to the end.
Why AI-written letters get spotted
A hiring manager with a hundred applications notices AI letters because they are alike, not because software flags them. The same patterns come back when an AI tool is asked for “a cover letter for this job” with nothing else to work from:
- It opens on the applicant: “I am excited to apply for…”.
- It describes instead of proving: “results-driven”, “passionate”, “a proven track record”.
- It uses verbs chosen to sound important, such as “spearheaded” and “leveraged”, where “led” and “used” would do.
- It runs to 350 words or more and names the employer only in the greeting.
- Its numbers are missing, or worse, invented to fill the gap.
Every item on that list comes from the same cause: the tool had no specific material, so it filled the space with general language. The fix is in the input, not in a more elaborate request for “a human tone”.
Applicant tracking systems and the letter
An applicant tracking system (ATS) is the software employers use to store applications and rank them against the words in the job listing. The résumé is the document it reads most closely, so that is where the listing’s terms matter most.
The letter is mainly read by a person, but it should still use the listing’s own words for the skills it asks for. If the listing says “WMS” (warehouse management system) and your résumé says “system implementation”, both the software and the reader may miss that you have the experience they want.
How to write it with AI, step by step
1. Choose one listing and find its top three requirements
Paste the full listing and ask which three requirements it stresses most, in its own words. Those three, not all fifteen bullet points, are what the letter answers.
2. Find the employer’s current problem
Recent news, results, a product launch, a service failure, or the reason the role is open. One checkable fact is enough to open the letter.
3. Match each requirement to one result with a number
For each requirement, pick the single result that proves it: what changed, by how much, over what scale. If you have no result for a requirement, leave it out rather than claiming it.
4. Write under 200 words, with firm limits
Tell the tool what to leave out: no opening on yourself, no adjective without a number, no claim you have not supplied. Limits produce a better letter than style instructions.
5. Check every number yourself
Ask for a list of every figure in the letter and the result it came from, and confirm each one. An interviewer will ask about the numbers, and each must be one you can explain.
Worked example: Dana Okafor
The scenario
140 applications, one letter, no offers
Dana Okafor ran operations for two warehouses at Halden Freight in Columbus, Ohio, for eight years until her role was eliminated in March 2026. Over the next 11 weeks she sent the same résumé and the same 410-word cover letter to 140 operations jobs.
Her old letter named no employer’s problem and could have been sent anywhere. It was.
For Larkspur Home’s Senior Operations Manager role, the listing stressed three requirements: peak-season planning, a WMS rollout and cost control. The news Dana pasted showed that Larkspur’s holiday orders had shipped up to 18 days late the previous December. A former colleague, Jordan Blake, had offered to refer her. This is the letter that came back, at 156 words:
Dear Ms Reyes,
Larkspur’s customers waited up to 18 days for holiday orders last December. I ran Halden Freight’s two Columbus warehouses through the 2025 peak at 1.8 times normal daily orders, and order-to-ship time stayed under 36 hours. Jordan Blake, who worked with me at Halden, suggested I write to you about the Senior Operations Manager role.
Your listing asks for three things, and each is something I have done:
Peak-season planning. I rebuilt our pick routes and shift start times, cutting order-to-ship time from 52 to 31 hours across both sites, before volume nearly doubled.
A WMS rollout. I led the move to a new warehouse management system at two sites in 2024. It went live on schedule, and pick accuracy rose from 97.1% to 99.4%.
Cost control. Moving both sites to demand-based shift planning cut overtime cost by $310K a year.
I would welcome 30 minutes to discuss how I would plan Larkspur’s next peak season.
Yours sincerely,
Dana Okafor
Set side by side, the two letters differ on every line that matters:
| The old letter | The Larkspur letter | |
|---|---|---|
| Length | 410 words | 156 words |
| The employer’s problem | Not named | The opening line: Larkspur’s 18-day delays |
| Written for | Any operations job | Larkspur’s listing and its top three requirements |
| Sent to | 140 employers | One |
The letter was one part of a rebuilt search. Her 26 tailored applications, each with its own letter and a résumé rewritten around measured results, led to 7 first interviews: 27%, against 2.1% before. The search ended with a signed offer from Larkspur at $109,000 base and a $5,000 signing bonus.
Try it: a shortened Prompt 05
This is a shortened version of Prompt 05 from the Career & Job Search pack, set up to run on its own. Paste it into ChatGPT, Claude or Gemini and fill in the brackets.
You are a former head of talent acquisition who has read thousands of cover letters for operations, finance and technology roles. A cover letter answers one question: why this candidate, for this job, at this company. It opens with the employer's problem, not the applicant, and every claim carries a number the applicant can defend in an interview. THE JOB: - The listing, in full: [paste] - What you know about the employer's current situation: [news, results, or the problem this hire is meant to fix] - Who referred you, if anyone: [name and how you know them] YOU: - Your target role and your current or last job: [title, employer, years] - Your results, with numbers: [3-6 results, each with what changed and by how much] Produce: 1. THE TOP THREE REQUIREMENTS. The three requirements the listing stresses most, in its own words. 2. THE MATCH. For each requirement, the one result of mine that proves it, with its number. If I have no result for a requirement, say so; do not invent one. 3. THE LETTER. Under 200 words, written as an email body: a subject line that states a result, an opening on the employer's problem, the three matches, the referral if there is one, and one specific request. No "I am excited to apply", no adjective without a number behind it, no claim I have not supplied. 4. THE CHECK. Every number in the letter, listed with the result it came from, so I can confirm each one is accurate. Write in plain first-person sentences. If my results are thin, say which requirement is unproven rather than padding the letter.
The full Prompt 05 draws on the Search Brief built by prompts 01 to 04, the target role, the market research and the rewritten résumé, so the letter uses the same results and the same listing terms. In Claude it also builds the letter as a Word document to match the résumé.
Checks before you send
- Every number is true, and you can explain how it was measured.
- The employer’s name, the role title and the hiring manager’s name are correct.
- The letter is under 200 words and fits on one screen.
- No adjective appears without a result behind it.
- The subject line states a result, not “Application for…”.
- Read aloud, it sounds like you speaking, not like a template.
Questions
Do recruiters still read cover letters?
Not always. Many screen on the résumé first. But when a listing asks for a letter, when the letter is the email body, or when you have been referred, it is often the first thing a hiring manager reads. A short, specific letter costs little and can decide between two similar résumés.
Can employers tell a letter was written with AI?
AI-detection tools are unreliable, and few hiring managers depend on them. What they notice is a generic letter: one that could have gone to any company. A letter built from your own results and this employer’s problem does not read that way, whichever tool drafted it.
Should I say I used AI?
That is your choice. What matters is that every claim in the letter is yours and true, and that you can discuss each result in an interview.
How long should a cover letter be?
Under 200 words for most roles, especially when it is sent as an email body. Longer letters usually add general statements, not evidence.
Does this work in ChatGPT, Claude and Gemini?
Yes. The prompt is plain text and works in all three. In Claude, the full pack version can also produce the letter as a Word document.
Written by a former Gartner Managing Partner and investment banking SVP
The guides and the prompt templates on this site come from a career spent building these documents: board decks, forecasts, business cases and hiring decisions, as Managing Partner at Gartner, SVP in investment banking and Country Manager at international subsidiaries. The worked examples are published in full on each product page. Browse the Prompt Library.
Dana Okafor, Halden Freight, Larkspur Home and the people named are fictional, and the figures are illustrative.
Hiring practice differs by employer and country; nothing here is legal advice. Output from any AI tool should be reviewed before use.