Step 1: Cut the giveaway phrases
Certain phrases have become AI shorthand for "I have nothing specific to say." They are the fastest thing a recruiter skims past. Delete them on sight and replace each with a concrete fact:
| Sounds like AI | Sounds like you |
|---|---|
| "I have a proven track record of success" | "I grew trial-to-paid conversion from 4% to 7% in two quarters" |
| "I am passionate about this industry" | "I have shipped three side projects in this space on my own time" |
| "I am a great asset to any team" | "I was the person other engineers came to for release debugging" |
| "I am thrilled and excited to apply" | (delete it and open with your strongest concrete result) |
| "I look forward to leveraging my skills" | "I would spend my first month on your onboarding drop-off, which your reviews mention a lot" |
The pattern is simple: every vague enthusiasm phrase can be swapped for a specific fact about what you actually did. Numbers, tools, outcomes. That alone removes most of the AI smell.
Step 2: Pass the "could I send this to another company?" test
This is the single best filter. Read your letter and ask whether you could send it to a different employer by changing only the company name. If the answer is yes, it is generic, and generic is what reads as AI. Fix it by naming something real: the actual product you would work on, a specific thing the company does that you respect, a reason you want this role and not just a job. One genuine, un-copyable sentence does more than three paragraphs of polish.
Step 3: Break the rhythm
AI writes in a smooth, even cadence. Every sentence is a similar length, every paragraph a tidy three lines. Human writing is lumpier. Some sentences are long and work through an idea. Some are short. Go through your draft and deliberately vary it. Cut one sentence in half. Let another run. That unevenness is one of the strongest signals that a person wrote it, and it is the thing detectors and recruiters both pick up on.
Step 4: Read it out loud
Say the whole letter out loud, or paste it into a screen reader. Anything you would never actually say to a person is a line to rewrite. "I am eager to contribute to your dynamic organization" is not something a human says. "I want to work on the checkout flow because I have rebuilt one before and know where they break" is. Your ear catches AI phrasing faster than your eye does.
The faster path: start from a specific draft
Most of this editing exists because the first draft was generic. If you start from a draft that already pulls in the job description and your real experience, there is far less to strip out. That is what JobPilot does: it reads the posting and your background and writes a tailored draft in your browser, so the starting point is specific instead of boilerplate. You still add the one un-copyable reason and vary the rhythm, but you are editing a real letter rather than sanding down filler.
Write a tailored cover letter that starts specific
JobPilot reads the job posting and your experience and drafts a cover letter matched to the role, right in your browser, so you spend your time personalizing instead of deleting filler.
Add JobPilot to Chrome — Free →Free plan available · works on LinkedIn, Indeed, and any job posting
Frequently asked questions
What words make a cover letter sound like AI? "Thrilled to apply," "passionate about," "proven track record," "dynamic environment," "leverage my skills," "a great asset." They are generic enough to fit any job, so replace each with a concrete fact about your work.
Do recruiters care if I used AI? Mostly no. They care that it is generic. A specific, tailored letter written with AI help reads fine; a vague one reads as low effort no matter who wrote it.
Should I run it through an AI detector? As a final mirror, yes, but do not trust the score as a verdict. A flag usually means the writing is still generic, which is the real thing to fix. A free local detector is enough.