How to Use ChatGPT to Tailor a Resume to a Job Description
Compare before you rewrite. This process uses ChatGPT to map job requirements to your resume, then limits edits to what is already true.
By JRNEY Editorial TeamUpdated October 8, 20265 min read22 views
JRNEY guides are written to help job seekers make resumes easier for ATS systems and recruiters to evaluate. Read our resume audit methodology and editorial standards.
To tailor a resume with ChatGPT, have it compare the job description with your resume and build a table of requirements and evidence first. Rewrite only after you have decided, row by row, what your resume already proves. If you ask for a tailored resume in one step, the model can make your experience sound closer to the posting than it is. The steps below keep you in control of every claim.
Step 1: Set up the inputs
You need three things in separate, labeled blocks: the job description, your current resume (with contact details removed), and a short list of true facts that are not on the resume yet. Paste the guardrail block from the ChatGPT resume prompts library at the top of each request. It tells the model to use only your facts and to leave placeholders instead of guessing.
MIT's career office advises building a resume that targets a specific position by identifying the skills a posting asks for and choosing relevant experience that shows them (MIT CAPD). That is the job here: select and arrange, not invent.
Step 2: Build the evidence map and decide what each row means
Send the guardrail block plus this request:
From JOB DESCRIPTION, list the requirements, split into "must have" and "nice to have" as the posting states them. For each, quote the line in RESUME that shows it, or write "no evidence". Return a table: requirement | resume evidence | strength (strong, partial, none).
Check the table against the posting yourself. Look for rows where "partial" evidence was rated "strong", and fix any row you disagree with.
Then decide what each row means:
| Strength | What it means | What to do |
|---|---|---|
| Strong | The resume already proves it | Move it higher or state it more plainly |
| Partial | Related work, not the same thing | Describe what you did accurately, in the posting's vocabulary only where it is the same work |
| None | You have not done it | Leave it out. If it is a must-have, decide whether to apply anyway |
Leaving a requirement uncovered is an honest outcome. Some roles will not be a match, and a mapping is a fast way to see that.
Illustrative example (not a real person)
A fragment of a job description mapped to a fragment of a resume. Square brackets are placeholders for facts the real person must supply.
| Requirement in posting | Resume evidence (before) | Edit (after) |
|---|---|---|
| Own onboarding for new accounts | "Ran onboarding calls for new accounts" in the fourth bullet | Move to the first bullet of the role. Wording stays as is. Add [X] accounts only if you can verify the count. |
| Coordinate with support and product | "Weekly meeting with support team" | "Held a weekly meeting with the support team." Do not add product unless you worked with them. |
| Keep account records in a CRM | "Logged account notes in [CRM tool]" | Move up in the skills list. Name the tool as it appears in your record. |
| SQL reporting | No evidence | Leave out. Mention a course or project only if it is real and recent. |
Every edit in the last column is a move, a clearer verb, or a placeholder. None adds a tool, number or team the before column lacks. The SQL row stays empty on purpose.
Step 3: Edit in a fixed order
Work through the resume in this order and send one section per request, including the guardrail block:
- Summary or headline. Ask for 2 to 3 lines built only from the strong rows.
- Skills. Ask it to reorder your existing skills for this posting, with nothing added.
- Experience bullets. Reorder within each role so the strongest evidence leads, then rewrite bullets one role at a time. Ask it to insert [X] where a number would help, and fill it in only with true figures.
The prompt library has the bullet-rewrite and summary prompts, so they are not repeated here.
Step 4: Check for copied language and invented facts
Read your result next to the job description. If a bullet reads like the posting with your name on it, rewrite it to describe what you did. For example, "Responsible for driving customer-centric outcomes" is a phrase from a posting, while "Ran onboarding calls for new accounts" is work. For a deeper look at this balance, see tailoring without keyword stuffing.
Then run the fabrication check (prompt 6 in the library) in a new chat: give it the original and the revised resume and ask it to list anything that appears only in the revised one. Google notes that generative AI output can contain inaccuracies, so the final fact-check is yours (Google Search Central).
Last, check the employer's rules. NIST's resume tips tell applicants to address each specialized experience requirement with clear, specific examples and not to use AI-generated, plagiarized or recycled content (NIST resume tips). Employers differ, so read the posting and application instructions.
When a tool is easier than prompts
If you apply to many roles, repeating this by hand adds up. JRNEY's job-description tailoring maps requirements to evidence and rewrites the relevant sections, and its optimization rewrites, condenses and reorders without adding claims, credentials, metrics, dates, employers or skills. You still decide what goes on the page.
FAQ
Should I tailor for every application?
Tailor when the role is a real match. Make the edits above on a copy so your master resume stays accurate for everything else.
Can I paste the job description and resume into ChatGPT?
You can, but remove contact details and anything private first, and check how the tool you use handles pasted text.
Sources
Job match
TargetedTailor the resume to one real job
Map the job description to your real evidence, then adjust the summary, skills, and strongest bullets without keyword stuffing.
Start tailoring free