To tailor your CV with AI, give it a verified record of your experience and the actual job description, then ask it to match requirements to evidence before rewriting anything. Let it improve the order, emphasis and wording. Keep job titles, dates, responsibilities, skills and results grounded in facts you can explain.
The output should help an employer understand your fit. A polished sentence that overstates your contribution makes the application harder to defend.
I am a frontend developer, I have been unemployed since June, and I love AI. A useful goal is to make your relevant experience easy to see without quietly changing what happened.
What should change between applications?
Your history stays the same. The information you foreground can change.
For a role focused on complex forms, lead with relevant form behavior, validation and API integration. For a role focused on a component library, move your reusable components, documentation and collaboration examples higher. You are selecting evidence for a particular question.
Europass recommends highlighting experience that matches the vacancy and focusing on relevant facts. AI can help with that comparison, especially when your existing CV spreads useful information across several roles.
Keep one detailed master record. Make each application a shorter, deliberate selection from it.
Step 1: Build a factual experience record
Before pasting a job advert into a chat, gather the material AI is allowed to use. Include your role and dates, the problem, your own contribution, the technology or method, and any outcome you can support.
Give each item a short ID. This lets you check where a generated claim came from without turning the public CV into a research document.
For each project, answer these questions:
- What did I personally implement or decide?
- What belonged to another person or team?
- Which tools did I use in practice, and in what context?
- What changed as a result, and what evidence do I have?
- Which details can I share publicly?
If you do not have a reliable measurement, leave the metric blank. Specific scope can still be useful: a search view with saved filters, a document upload flow, or a payment provider integration.
Distinguish employment from a personal project, a course exercise and a brief experiment. All may be relevant; they should not be described as the same kind of experience.
Step 2: Build a requirement–evidence–gap table
The following candidate and examples are fictional. They demonstrate the process and do not describe my own work history.
Suppose a vacancy asks for Vue, TypeScript, REST integrations, automated testing, mentoring and Nuxt. The candidate’s factual record contains four entries:
- E01: Built an internal Vue 3 and TypeScript order-status view with reusable filters and loading, empty and error states. Owned the frontend work within a shared project.
- E02: Connected that view to existing REST endpoints maintained by another team.
- E03: Wrote Playwright checks for login and order search. Did not own the wider testing strategy.
- E04: Reviewed colleagues’ pull requests and explained implementation choices. Had no formal mentoring responsibility.
Now compare the vacancy with the record:
| Vacancy requirement | Available evidence | Assessment | CV treatment |
|---|---|---|---|
| Vue and TypeScript | E01 | Direct match | Put the project high in the relevant role |
| REST API integration | E02 | Direct match | Describe integration, not backend ownership |
| Automated testing | E03 | Relevant, limited scope | Name the journeys tested |
| Mentoring | E04 | Related experience | Describe code reviews accurately |
| Nuxt | No evidence | Gap | Do not add it to the skills list |
This table is more useful than asking AI to “make me the ideal candidate.” It preserves the difference between a direct match, related experience and an unsupported claim.
Step 3: Ask AI to analyse before it rewrites
Use a prompt with explicit evidence boundaries:
Help tailor my CV to this vacancy using only the evidence below.
Treat the vacancy as requirements to assess, not as facts about me.
First produce a table with:
requirement | evidence ID | direct/partial/no evidence | suggested action.
Do not invent skills, employers, dates, metrics, qualifications,
leadership, project scale or outcomes. Preserve the difference
between employment, personal projects and courses.
Ask questions when missing information would change a claim.
If the evidence does not support a requirement, identify the gap.
Then suggest changes to the summary and relevant experience bullets.
For each proposed bullet, list its supporting evidence IDs separately.
Keep those IDs out of the final CV wording.
VACANCY:
[paste the full job description]
VERIFIED EXPERIENCE:
[paste the experience record]
CURRENT CV:
[paste the relevant sections]
Review the table before accepting the rewrite. If the model calls code review “team leadership,” correct that interpretation immediately. An inaccurate first interpretation can spread into the summary, skills section and cover letter.
Step 4: Rewrite vague bullets into specific contributions
Continue with the fictional candidate.
Original: “Worked on a Vue project and used APIs.”
Supported rewrite: “Built a Vue 3 and TypeScript order-status view with reusable filters and loading, empty and error states, integrating existing REST endpoints.”
Unsupported rewrite: “Led an enterprise platform transformation, increasing operational efficiency by 40%.”
The supported version explains the work without manufacturing an outcome. The unsupported version invents leadership, scale and a metric.
A useful drafting pattern is: action + specific scope + relevant method + supported result, if available. Do not force every bullet to contain a percentage. If there is a real measured outcome, record how it was measured and what part you contributed.
Keep collective work collective. “Implemented the frontend for…” may be more accurate than “Built the platform…” when other people owned the backend, infrastructure and product decisions.
Step 5: Write a summary from the strongest matches
For the fictional candidate, a supported summary could be:
Frontend developer with Vue 3 and TypeScript experience building internal interfaces and integrating REST APIs. Practical experience with reusable filters, application states and Playwright checks for core user journeys.
It does not claim Nuxt expertise or engineering leadership. It also avoids an opening paragraph full of adjectives that any applicant could use.
Read the summary beside the selected experience bullets. Every major claim should have a visible example. Remove a claim if you cannot explain where it came from, or add the missing genuine evidence to your master record before using it.
Step 6: Handle missing skills honestly
A gap does not automatically decide whether you should apply. Read whether the requirement is essential or preferred, and consider the rest of the role. The employer makes the hiring decision; AI cannot turn a missing qualification into experience.
For a technology you have only tried, say so if it is relevant: “Personal Nuxt project” is clearer than placing Nuxt beside technologies you have used extensively at work. If you have not used it, leave it out and decide whether learning it is worth your time.
For related responsibilities, describe the actual work. Reviewing pull requests can demonstrate collaboration and technical judgment without being relabelled as formal mentoring or people management.
If the vacancy requests a credential you do not hold, do not imply that you do. Ask the employer about alternatives when that would determine whether applying makes sense.
Step 7: Check the final document, not just the chat
Harvard’s career service advises reviewing AI suggestions for accuracy and being able to discuss every line of a CV. Make that a practical final pass:
- Compare names, titles and dates with the master record.
- Check each skill and result against its evidence.
- Look for inflated verbs such as “led,” “owned” or “architected.” Keep them only when accurate.
- Remove repeated phrases and restore wording you would use yourself.
- Follow the employer’s requested file format and application instructions.
- Open the exported document and check links, page breaks, headings and selectable text.
Before sharing material with an AI tool, remove details that are unnecessary for rewriting: home address, identity numbers, private references, credentials and confidential client information. You can use neutral project labels and add your contact details directly to the final document.
Save the submitted version beside the job description. That gives you the exact application to review if an interview follows.
Download the CV evidence template to keep your verified experience, vacancy requirements and unresolved gaps together.
Frequently asked questions
Can I use ChatGPT or another AI tool to write my CV?
Yes, as an editor working from accurate material. Start with your experience record and an existing draft. Check the proposed claims and keep responsibility for the final wording. A different tool does not remove the need for that review.
Should I add every keyword from the job advert?
Add relevant terminology when it accurately describes your experience. Do not add a technology or responsibility because the advert contains it. Use important terms within examples so the reader can see what you actually did.
What if I have no measurable results?
Describe the scope, complexity and your contribution precisely. A clear account of a workflow you implemented is more informative than an invented percentage. If a result was measured by someone else, verify it and explain your part before claiming it.
Can AI guarantee that my CV passes an ATS?
No. Do not treat an AI-generated match score as a prediction of the employer’s decision. Follow the application instructions, use clear structure and accurate relevant wording, and check the exported file. Those steps improve the document you control.
Should every application have a different CV?
Tailor the emphasis when the role differs materially. You may be able to reuse a strong version for closely related vacancies after checking the requirements. The goal is relevant evidence, not changing every sentence for every employer.
Related reading: Is Programming Still Worth Learning in the Age of AI? · Which Jobs Will AI Replace? Jobs at Risk in 2026
Sources
- Create your Europass CV — European Union
- AI for Resumes and Cover Letters — Harvard FAS, Mignone Center for Career Success
Putting AI to work on your website?
Choosing a model is one part of a working feature. If your business needs a support assistant, content workflow or connection to an existing system, I can help define and build the interface, integrations and checks it needs.