Rewrite your resume with AI without sounding like AI

You can rewrite your resume with AI and have it read like you wrote it, if you do three things: give the model a list of facts it is allowed to use and nothing else, aim it at one specific job posting, and strip the phrases that give machine writing away before you send it. The failure mode is the opposite: paste the old resume, say "make it better", and get a page of "spearheaded cross-functional initiatives" that no interviewer believes.

Here is the process, the prompts, and the list of tells to remove.

Step 1: build the fact list first

Before any rewriting, extract every fact from your current resume into a plain list: employer, dates, title, tools, team size, numbers. Add facts that are missing from the resume but true (a project you forgot, a number you can back up). This list is the only source the model gets.

Prompt: "Here is a list of facts about my work history. Do not add any fact, number, tool or responsibility that is not in this list. If a bullet needs a number and there is none, write it without a number."

That last sentence is the one that stops the model estimating "improved efficiency by 30 percent".

Step 2: aim at one posting

Paste the job posting. Ask for its keywords: skills, tools, responsibilities, the words it repeats. Then: "Mark which of these keywords my fact list actually supports. Use only those in the rewrite." Keywords the facts do not support go nowhere; this is how you stay honest and still pass a keyword screen.

A resume aimed at one posting is shorter and sharper than a general one. Keep the general one as the source; generate a targeted version per application.

Step 3: rewrite in the standard shape

• Summary: two to three lines that name the role you are applying for, your years and the two things the posting cares most about.
• Experience: reverse chronological. Each bullet is action, what, result. Results carry numbers only from the fact list.
• Skills: grouped (languages, tools, methods), matched to the posting's wording where your facts support it.
• Education and certifications, short.

Ask for one to two pages and plain headings (Summary, Experience, Skills, Education). Fancy section names confuse applicant-tracking systems and people.

Step 4: remove the machine tells

Read the draft and cut these:

• Verbs that appear in every AI resume: spearheaded, leveraged, orchestrated, championed, drove (when nothing was driven)
• "Cross-functional", "stakeholders", "synergies", "dynamic", "results-driven", "passionate"
• Bullets that start with the same verb three times in a row
• Any adjective describing you rather than the work ("detail-oriented")
• Rounded numbers that feel invented ("over 100 clients") unless the fact list says so
• Perfect parallel structure in every bullet; real resumes vary

Replace with the plain verb for what you did: built, wrote, fixed, shipped, ran, cut, moved, hired, taught.

Step 5: read it as the interviewer

Every bullet should survive the question "tell me more about that". If you cannot talk for two minutes about a bullet, it is either inflated or belongs to someone else's resume. Cut it.

Then read it aloud once. Sentences you would never say are the ones that still sound generated.

Step 6: layout the screening software can read

• Single column, no tables, no text boxes, no images, no headers or footers with content in them
• A standard font at 10.5 to 11 point, plain round bullets
• Save as DOCX and as PDF; test the PDF by copying all its text into a plain editor and checking it reads in order
• File name: Firstname-Lastname-Resume.pdf

Prompts in one place

1. "Extract every fact from this resume into a list: employer, dates, title, tools, numbers. Add nothing."
2. "From this job posting, list the skills, tools and responsibilities it names. Mark which ones my fact list supports."
3. "Rewrite the resume for this posting using only supported facts. Summary of two to three lines, reverse chronological experience with action, what, result bullets, grouped skills. Numbers only from the fact list; otherwise no number. One to two pages, standard headings."
4. "List every phrase in the draft that reads as generic or machine-written, and suggest a plainer replacement for each."

What to keep from the old resume

Your voice. If the old resume had a bullet that a colleague would recognise as yours, keep it word for word. A rewrite that erases every trace of the person is the thing recruiters have learned to spot.

Related reading: How to use AI to find jobs that match your resume for building the list of postings to aim at, and Prompting patterns that save hours: brief, facts, checklist for the fact-list pattern used elsewhere.

FAQ

Can recruiters tell when a resume was written by AI? They notice the symptoms: inflated verbs, generic phrases, identical bullet structure, numbers that feel invented. Remove those and the question does not come up.

Should I let AI add numbers to my resume? No. Give it a fact list and instruct it to write bullets without numbers where none exist. An invented metric is the fastest way to fail an interview.

Is it OK to use AI to write a resume? Yes, as a drafting and editing tool, with your facts as the only source. The content is still yours to stand behind in the interview.

How do I make an ATS-friendly resume with AI? Ask for single column, standard headings, plain bullets, no tables or images, and export as DOCX and PDF. Test the PDF by copying its text out.

Should I tailor my resume to each job? Yes. Keep one master fact list and generate a targeted one-to-two-page version per posting, using only keywords your facts support.

What words should I remove from an AI-written resume? Spearheaded, leveraged, orchestrated, cross-functional, stakeholders, results-driven, passionate, detail-oriented, and any bullet you cannot talk about for two minutes.

If you want the rewrite done with these rules for one target role, the resume enhancement gig on my store is $9: https://rmgo.sites.tab.bot/resume?s=blog