Most people use ChatGPT for CVs the wrong way. They paste the whole document and ask “make it better.” The output is generic, overwrites things that were already good, and sounds nothing like them. Here is the workflow that actually works.
The Core Principle: Section by Section, Not All at Once
A CV is not a single piece of writing, it is a collection of discrete sections, each with a different purpose and a different audience expectation. When you ask an AI to improve the whole document at once, it treats everything as equally important and applies uniform changes. The result is a document that is consistently mediocre rather than selectively strong.
The better approach is to work one section at a time, with a fresh brief for each. This takes slightly longer but produces output you can actually use.
Step 1: Set Context Once at the Start
Before any section-specific prompts, give ChatGPT a brief that stays active for the whole session.
“I am a [job title] with [X years] of experience in [sector]. I am applying for [target role] at a [type of company/organisation]. Please use this as background for all prompts in this session. British English only. No em dashes. No buzzwords.”
This anchors every subsequent output to your actual situation and eliminates the generic framing that makes AI CVs easy to spot.
Step 2: Rewrite the Personal Statement
Paste your current personal statement and the target job description, then use a structured prompt.
“Here is my current personal statement: [paste]. Here is the job description I am targeting: [paste]. Rewrite the personal statement to be 3–4 sentences. Lead with my role and years of experience. Include one specific capability that maps to this role. End with what I am looking for. Do not use the words ‘passionate’, ‘driven’, or ‘results-oriented’.”
Review the output. If it sounds unlike you, add: “This reads too formally. Rewrite in a more direct, plain-English tone.” Iterate once or twice until the voice is right.
Step 3: Strengthen Individual Bullet Points
The most impactful single change you can make to a CV is shifting from duty statements to achievement statements. Most CVs read as a job description. The best ones read as a track record.
“Rewrite this bullet point to focus on outcome rather than duty. If I have not given you a number, suggest what metric I should try to verify and add: [paste your bullet point].”
The instruction to suggest metrics, rather than invent them, is important. Numbers on a CV must be accurate. AI can help you identify what to measure; it cannot supply the figure.
Step 4: Align the Skills Section
Paste the job description and your current skills section. Then:
“Compare this job description [paste] with my current skills section [paste]. List any keywords or capabilities in the job description that are absent or not closely matched in my skills section. Then suggest a revised skills section that includes these, grouped into two or three logical categories.”
Do not add skills you do not have. Use this output to surface things you genuinely possess but had not thought to include.
Step 5: Check the Whole Against the Job Description
Once individual sections are revised, paste the full updated CV and the job description together for a final check.
“Here is my updated CV [paste] and the job description [paste]. On a scale of 1–10, how well does my CV match this specific role? List the three strongest alignments and the three most significant gaps. Do not rewrite anything, just assess.”
This surfaces issues you may have missed and gives you a sense of where the application stands before you submit.
The Limitations of Free ChatGPT
The workflow above produces good results, but it has real constraints. ChatGPT does not know UK ATS conventions, UK CV formatting expectations, or the specific parsing behaviour of different systems. It has no memory between sessions, so the context you set at the start is lost the next time you log in. And on the free tier, output quality is noticeably lower.
For occasional applications, this is manageable. If you are applying to multiple roles simultaneously, the overhead compounds, each application requires resetting context, re-prompting each section, and manually checking ATS alignment.
A Purpose-Built Alternative
The prompting workflow above is valid and many candidates use it well. The trade-off is the time and skill required to get consistent results, writing good prompts, iterating on output, and manually checking ATS alignment for every application.
Zappli is built specifically for UK CV tailoring. Upload your CV and a job description, and it handles the keyword alignment, ATS formatting, and section rewriting automatically, no prompts needed. The free Diagnose tier gives you an ATS score and gap analysis; Pro-Pass costs £7.99 for seven days, or £11.99 a month on Pro-Monthly.
If reducing the prompt engineering overhead sounds useful, the link below is where to start.
Start free today
Your free ATS score
in under two minutes.
No card. No catch. Free forever, upgrade to Pro or Agent when you’re ready.
Leave a comment