Tailor your CV to a vacancy with ChatGPT

To tailor your CV to a specific vacancy with ChatGPT, provide the job requirements and your actual experience as separate inputs. Ask for an evidence-matching table first, then rewrite the relevant CV section using only supported facts. Do not turn an unknown language level, an unused tool, or an unmeasured result into an achievement.

This method is useful if you are looking for your first job, changing fields, or adapting an existing CV to one position. The goal is to help a reader quickly see your relevant experience and understand what you actually did.

Gather your own facts first

Record your projects, tasks, tools, and the work you personally completed. Include the context: a learning project, volunteer work, or paid employment.

“I worked with data” needs clarification. “In a learning project, I used Excel filters to organize an orders table” already identifies an action, a tool, and the setting. If you have the file or report, check that the description matches your work.

A draft does not need a personal identification number, full address, or someone else’s contact details. Replace real names with temporary labels and use only material you are permitted to process.

OpenAI’s job-search example also relies on verified experience and checking individual claims (source 1). The vacancy and candidate notes below are fictional material created for this article.

Example: separate the vacancy from the experience

Imagine a junior operations assistant vacancy with these requirements:

Prepare the candidate’s facts separately. The source notes for this exercise are:

F01: In a learning project, I organized a table of fictional orders in Excel; I used filters and SUM.
F02: For the same learning project, I wrote a one-page report in Georgian.
F03: I have not used SQL yet.
F04: My English level has not been assessed.

This is neither a real candidate’s CV nor a transcript of an actual ChatGPT response. The IDs let you trace each claim below back to a source fact.

Preserve the distinction between required and desirable qualifications. R4 is desirable; the lack of SQL experience does not change that. R3 is required, but F04 does not establish the candidate’s actual English level.

First prompt: show the evidence match in a table

Provide ChatGPT with the vacancy table and fact notes, then ask:

Compare the job requirements with the candidate facts.
Do not write a CV at this stage.

Return a table:
requirement ID | required or desirable |
fact ID and exact quote | status | clarification question.

Use these statuses:
- "Supported by an example" when the relevant action appears in the notes;
- "Needs clarification" when there is insufficient information;
- "Experience missing" when the notes explicitly state that the tool
  has not been used yet.

Review each requirement once. Do not treat missing information as a
negative answer. Do not present a learning project as employment.
Do not add a language level, a new tool, a job title, or a result metric.
Do not calculate an overall match percentage.

OpenAI’s prompting guide describes specifying the goal, context, output, and boundaries (source 2). Here, the table format helps you inspect the basis for each conclusion.

How to check the evidence-matching table

This is a manually prepared reference for the exercise:

“Supported by an example” means the notes contain the relevant action. One learning spreadsheet does not establish knowledge of every Excel feature, and writing a report does not by itself assess its quality. You still need to show the work and explain your contribution.

“The candidate has B2 English” is unsupported for R3. “The candidate does not have B2 English” would also go too far: the level is simply unknown. This fictional vacancy does not ask for a certificate, so the assistant should not add that requirement either.

The follow-up for R4 is straightforward: if you have learned SQL but omitted that experience from the source notes, describe the actual experience and update the input first. If F03 stays unchanged, SQL cannot appear in the skills list.

Second prompt: write two supported statements

After checking the table, move on to drafting:

Using only F01 and F02, give me two short CV bullets in English
for a section called "Learning project".

In the first, emphasize the work done using Excel filters.
In the second, emphasize preparing the Georgian report.
Preserve the learning context and the description of fictional data.
Do not add a company, client, paid employment, business impact,
language level, or unsupported number.

After the text, separately identify the source fact for each bullet.
Prepare a draft only; do not submit an application or contact anyone.

A possible draft, written manually for this article:

- In a learning project, I used Excel filters to organize a table of fictional orders.
- For the same learning project, I prepared a one-page report in Georgian.

The first bullet is supported by F01 and the second by F02. The first emphasizes filters, which matter for the vacancy; omitting SUM from the source information does not change the fact. The second replaces “wrote” with “prepared,” preserving the type of work, language, and context.

Neither bullet establishes that the candidate meets every required qualification. The English-level question remains open. A polished draft does not establish complete suitability for the role.

Check an overstated sentence against the facts

For example, reject this sentence: “I improved the company’s order management and made the workflow 30% faster.” The exercise sources contain no company, process comparison, or such measurement.

Request a specific correction: “Remove the company and the 30% result. Keep the learning project, Excel filters, and table organization.” That improves the wording while preserving its source.

If you have measured a real result, state what you counted, what you compared, and which part you performed yourself. A number alone does not establish that your work was the sole cause of a change.

Keep the same facts in an English CV

For an English version, provide the checked Georgian text and ask the assistant not to add a new skill, job title, or period of experience. You can adapt the wording naturally; the facts should match across versions.

The guide to translating with ChatGPT and checking meaning can help with that step. Producing an English CV with a tool does not itself establish B2 English proficiency. F04 remains unresolved in this example after translation too.

Give the final file a clear name and keep the relevant vacancy alongside it. This helps you avoid accidentally using a draft adapted for a different position.

Trace the text back to your experience before sending

For every bullet, ask whether you can show the work or explain how you did it. Do the tool, your role, project setting, and every number match your notes? Separately check names, dates, contact details, and links.

If you have only a few experience points, writing directly by hand may be simpler. ChatGPT can help choose emphasis and improve sentences when you control the source facts. This method helps prepare the text; it cannot promise an employment outcome.

Common questions about tailoring a CV

What should I include if I have no paid experience?

Describe a learning, personal, or volunteer project in which you performed actual work. State the context explicitly. You do not need to turn a project into employment to make the work clear.

Does an achievement have to include a number?

If you lack a reliable measurement, describe the specific action and artifact you produced. Instead of adding an unsupported percentage, show what you prepared and what your contribution was.

Should I copy every term from the vacancy into my CV?

Use terms that describe your actual experience. Seeing SQL in a vacancy does not give you SQL skills; the requirement may be a learning goal rather than a skill you already have.

Full source addresses

To open a source, copy its full address into your browser.

1. OpenAI, Track job applications: https://learn.chatgpt.com/use-cases/track-job-applications
2. OpenAI, Prompting: https://learn.chatgpt.com/docs/prompting