How to Blend Your Experience With AI-Suggested Improvements

Ask ten job seekers how they wrote their CV and you will get ten different stories, but the ones who ended up with interviews rarely say “AI wrote it” or “I wrote every word myself.” They describe something closer to a conversation: a draft, a suggestion, an edit, another look. The blend is the point, and it is worth understanding how to do it properly rather than by accident.

Why neither approach works alone

A CV written entirely by hand tends to repeat the same three phrases you have used since your first job application. You know your experience too well to see what is missing, and you are too close to the wording to notice when a sentence has gone flat.

A CV written entirely by AI has the opposite problem. It is structurally sound and grammatically clean, but it smooths over the specific detail that makes your experience yours. Numbers get vague, achievements get generic, and a recruiter who has read five hundred CVs this year can usually tell.

The strongest CVs sit in between. They use AI for structure, phrasing and the blank-page problem, and they use human judgement for accuracy, nuance and the details only you would know.

Human first, AI second

This workflow suits people who find it easier to write badly than to write nothing. You draft each section in your own words, however rough, then hand it to AI and ask it to tighten the structure, fix the flow and suggest phrasing you would not have reached for. If you are new to working with AI this way, a structured ChatGPT workflow makes this far more effective than pasting the whole document and asking for improvements.

The advantage is that your voice and your facts are locked in before AI touches anything. There is less risk of the tool inventing detail, because there is nothing to invent; it is only refining what you gave it.

The drawback is time. Getting from a blank page to a rough draft is the hardest part of CV writing for most people, and this approach does not remove that step.

AI first, human second

This workflow suits people who freeze at a blank page but edit well once something exists. You give AI your job history, the target role and a rough steer, let it produce a full first draft, then go through line by line correcting anything that is wrong, vague or not quite you.

The advantage is speed. A usable first draft appears in minutes rather than hours, and editing is a much easier task than generating from nothing.

The drawback is that it is tempting to under-edit. If the draft reads well, it is easy to accept a sentence that sounds convincing but overstates what you actually did. This is the workflow most likely to produce a CV that sounds good and says nothing specific.

In practice, most people end up mixing both within a single CV rather than picking one workflow for the whole document. A candidate who has changed roles several times might draft their most recent job in their own words, because the detail is fresh, then let AI take the first pass at older roles where the detail has faded and structure matters more than specifics. There is nothing wrong with switching approach section by section; the CV does not know how it was written, only how it reads.

The same logic applies to the personal statement, which is usually the hardest section to write and the easiest for AI to overreach on. Draft this one yourself first, even in note form, since it is where your own tone matters most and where a generic AI paragraph is most obvious to a reader.

Knowing when to override

Whichever order you use, there are moments where your judgement should win over the tool’s suggestion every time. If AI proposes a number you cannot verify, cut it or replace it with one you can stand behind in an interview; there are specific details AI should never be left to write on your CV. If it uses a term that does not match how your industry actually talks, change it back. If a sentence describes a version of your role that is technically true but misleading, rewrite it honestly.

AI optimises for a CV that reads well. You are the only one who can check that it is also true, specific and defensible under interview questioning.

The final read-aloud pass

Before you submit, read the whole CV aloud, start to finish. This single habit catches more problems than another round of AI editing ever will. Sentences that look fine on screen often reveal themselves as clunky, repetitive or oddly formal when spoken.

If you stumble over a phrase, a recruiter’s eye will stumble over it too. Rewrite it in the words you would actually use if someone asked you about that achievement across a table. This is really the same test covered in making an AI-written CV sound like you: if you would not say it in an interview, it should not be on the page.

A Faster Way

Manually shuttling a CV back and forth between a blank document and a chat window is a valid way to blend your experience with AI suggestions, and plenty of people do it well. It does, however, ask you to manage the structure, the prompts and the editing discipline all at once.

Zappli is built for the back-and-forth described above rather than one-shot generation, so your own material stays the input at every step.

The free tier needs no card and no account: an ATS score, plus three CVs and three job matches. That allowance is a lifetime one rather than a trial, so it shows you what the tool does instead of running a countdown.

If you want more than that, passes are one off and priced by length. £9.99 for seven days and up to 25 applications, £24.99 for thirty days and up to 100, or £49.99 for ninety days and up to 250. One payment only, never auto-charged, and nothing to cancel. Zappli is paid by candidates, never by employers.

If a free diagnosis sounds more useful than managing prompts manually, pick whichever of these is closest to where you are.