By Angelo Martinez, founder of ApplyBuddy
Changing “helped” to “led” takes one word. On a CV, it can change the entire claim.
For a candidate who knows their work but struggles to describe it, AI can be a useful editor. It can untangle a long sentence, suggest a clearer structure or help identify experience relevant to a vacancy. Someone should not have to be an accomplished copywriter to explain why they could do a job.
The difficulty starts when a rewrite quietly changes what happened. A shared task becomes individual ownership. An informal responsibility becomes a management role. An outcome appears that the candidate never measured.
The boundary is whether the revised wording still describes the same experience, at the same level of responsibility, with evidence the candidate can explain.
One experience, three versions
Consider this fictional example. An operations assistant updated a customer onboarding spreadsheet each week, chased missing details with sales colleagues and prepared a status summary for their manager. They used Excel, but did not manage the process or measure its effect on onboarding time.
Their original CV bullet reads:
Helped with customer onboarding and updated spreadsheets.
This undersells the work because it leaves the reader guessing what “helped” involved. A useful revision would be:
Updated the weekly customer onboarding tracker in Excel, followed up with sales colleagues on missing information and prepared status summaries for the operations manager.
The second version gives the employer more information without changing the facts. It names the tool, the frequency, the actions and the reporting relationship. Every detail comes from the candidate’s actual experience.
Now consider an inflated rewrite:
Led a cross-functional onboarding programme, using advanced Excel automation to reduce onboarding time by 30%.
That sentence introduces ownership, technical proficiency and a measured result that the example does not support. It may sound more impressive, but the candidate now has to defend a different job.
Even if the team really did reduce onboarding time, that would not automatically justify claiming the result personally. The wording would need to reflect the candidate’s contribution and how the outcome was established.
Five checks before accepting a rewrite
1. Trace each new detail to its source.
Compare the original and revised versions side by side. Underline anything added: a number, a tool, a client type, a responsibility or an outcome. For each addition, identify where it came from. A project note, performance review or reliable recollection may help. If the only source is the AI’s suggestion, treat it as a question to investigate, not a fact to publish.
2. Check the verbs for changes in responsibility.
“Supported”, “coordinated”, “managed” and “led” make different promises. Ask who made decisions, who was accountable and what you personally delivered. You can describe a substantial contribution without claiming ownership of the whole project. Replace vague modesty with specific actions, rather than automatically choosing the strongest verb.
3. Verify numbers and their meaning.
A percentage needs a basis: what changed, over which period and compared with what? Check whether the figure describes your work, the team’s work or a wider business result. Do not accept a plausible number simply because the sentence feels incomplete without one. Where no reliable metric exists, describe the scope, frequency or purpose of the work using details you can support.
4. Match skills to what you can actually do.
A job description can help you choose relevant examples, but it cannot supply missing experience. Using an Excel tracker does not establish expertise in macros. Attending a project meeting does not establish project management experience. Use the employer’s terminology where it accurately describes your work, and keep distinctions between exposure, practical use and expertise.
5. Practise explaining the bullet aloud.
Imagine an interviewer asking: “What did you do, how did you do it, and what happened next?” Answer without reading the CV. If you cannot explain a term, separate your contribution from a colleague’s or account for an outcome, revise the wording. This is also a chance to catch language that sounds polished on paper but feels unnatural in your own voice.
Keep control of the final application
Give AI a clear brief: improve clarity using only the facts supplied, preserve the level of responsibility and ask questions where evidence is missing. That instruction helps define the task; it does not replace checking the result.
Review the final document and any populated application fields before submitting. Tailoring a CV to a vacancy should remain a deliberate choice about relevant evidence, even when software speeds up the work.
A useful edit makes it easier for an employer to understand what you have done. The candidate should finish with language they recognise, claims they can support and examples they are ready to discuss.
Read more: This contributor article follows our Onrec Explores feature, When every application looks perfect, how do you find the real talent?
About the author
Angelo Martinez is a developer and founder of ApplyBuddy, which helps jobseekers review and tailor CVs and cover letters. He built the tool after using AI to improve his own job applications, with an emphasis on helping candidates inspect suggested changes before using them.
