Who writes the guides, and how
The guides are written by Naomi Shiels, CVelo's founder, not by an anonymous team or an unsupervised model. Facts about a formal process (selection criteria, APS classification, professional registration, graduate programs) come from the body that actually runs it, such as the Australian Public Service Commission, AHPRA, or the relevant state health service, not from other sites' blog posts. Where a guide turns on a rule that could change, it links to the primary source so you can check it yourself. The full detail on research approach is on the About page.
Review, updates and dates
Registration requirements, classification levels and application processes change, and a guide that was correct in one year can be wrong the next. Guides carry a publication date, and pages that have been substantively revised show a "last reviewed" date, as this page does above. A visible date is a claim you can check; an undated page is not.
Sourcing standard
A number of statistics circulate in the CV and recruitment space that trace back to vendor marketing rather than published research. The most common is a claim that a fixed percentage of CVs are automatically rejected by recruitment software before a human sees one. CVelo does not repeat figures it cannot trace to a real source. The standard is plain: a number goes on this site only if it is cited to something you can open and read, or it does not go in at all. A weaker, true claim beats a stronger one that is not.
What the AI is not allowed to do
Every prompt that writes about your experience, whether that is a CV rewrite, a cover letter, a selection criteria response or an interview coaching answer, carries one rule that overrides everything else: use only facts already in your CV or in what you told us. Never add, upgrade or invent an employer, job title, date, qualification, licence, clearance, number or outcome. Where the source is silent, leave it out rather than fill the gap with something plausible.
That rule exists because it was tested and found wanting once. Testing a real CV against two roles surfaced a rewrite that had quietly invented a stated responsibility the original CV never claimed, and moved a role's dates and location. The prompt was partly to blame: it had been handed missing keywords as though they were confirmed skills. The fix was two defences rather than one instruction: the rule above, written into every prompt, and a second, independent pass that checks the finished text against your source material before you ever see it.
You see that second pass as the fact-check note under your rewritten CV and cover letter, showing exactly what, if anything, it corrected. If it could not check something, it says so plainly rather than staying silent about the gap.
Your responsibility, stated plainly
A truthfulness rule and an automated check reduce the rate of a model inventing something about you; neither makes the output correct by definition. Language models make mistakes that have nothing to do with truthfulness, wrong dates transcribed correctly from a source, an odd sentence, a tone that does not sound like you. You are the last check before anything is sent to an employer, and every generated document is offered for you to review and edit, never sent on your behalf.
Corrections
If a guide, a tool's output, or anything on this site is wrong, tell us and it will be fixed, with the correction noted where relevant. Get in touch with what you found and where.
See it on your own CV
One free check a day, no account and no card. Nothing you paste is stored.
Score my CV free