Two strong drafters. Neither can score the file you send.
We judged both on draft quality, ATS-safe formatting, honesty about your numbers and job-ad matching. Here's what each one gets right.
The short verdict, before the details
Use ChatGPT when you're staring at an empty page and need momentum. Use Claude when you have a long, messy CV and want it tightened without losing your voice. Both write cleaner English than most CV templates you'll download. Neither of them has ever seen the file a recruiter opens, and that gap is where good drafts quietly fail.
Screening software doesn't read prose. It parses a file: headings, dates, job titles, columns, tables, and anything you tucked into a header or footer. A chat window judges your words; a parser judges your structure. You can get a beautifully written CV out of either assistant and still watch half of it land in the wrong field.
The second gap is the more dangerous one. Ask any assistant to make your achievements sound stronger and it will hand you a confident percentage you never measured. You then carry that number into an interview and have to defend it. Keep the drafting help, keep your own figures, then check the finished PDF against the way a screening system reads it.
Where each assistant genuinely earns its keep
ChatGPT: momentum
Fastest at turning a blank page into six usable bullet points, and quick to spin variations on a line you almost like.
Claude: the careful rewrite
Holds a long CV in view and trims it without flattening your voice. Better at cutting waffle than at adding flourish.
Both: translating your jargon
They turn internal job titles and project code names into plain language a recruiter outside your company will recognise.
Five things a chat window can't do for your CV
It never sees the actual file
You paste text. The recruiter opens a PDF. Your layout, tables and header text disappear before the conversation even starts.
It fills your gaps with invented numbers
Ask for stronger achievements and you'll get a plausible figure you never measured, and a question you can't answer in the interview.
It matches keywords by eye
Paste in a job ad and you get a general sense of overlap, not a checked list of the exact terms and titles you're missing.
It adds when you need it to cut
Assistants expand by instinct. A screening pass rewards a shorter, denser CV, so someone still has to decide what leaves the page.
It can't score the result
No pass or fail, no section-by-section rating, no ranked fix list based on how parsers actually read a document.
Side by side on the jobs that matter
| What you need | ChatGPT | Claude | HRLens |
|---|---|---|---|
| A first draft from scratch | Fast and fluent | Fast, more restrained | Builds from your details |
| Rewriting a long CV | Good, can drift off tone | Strongest of the three | Section-by-section notes |
| Sticking to your real numbers | Embellishes if asked | Embellishes if asked | Feedback on vague claims |
| Reading your finished PDF | Text only, layout lost | Text only, layout lost | Parses the file itself |
| ATS compatibility score | Not available | Not available | Yes, with a fix list |
| Tailoring to one job ad | Good with the ad pasted in | Good with the ad pasted in | Shows the gaps to close |
| Cost to try | Free tier available | Free tier available | Free CV analysis |
See how a screening system reads your CV
Upload the file you actually send. You'll get an ATS compatibility check, section-by-section feedback and a ranked list of fixes.
Questions job seekers ask us about this
Which is better for writing a CV, ChatGPT or Claude?
For a first draft from nothing, ChatGPT is quicker and looser, which helps when you're stuck. For editing a long CV down to two pages while keeping your own tone, Claude is the safer pair of hands. Most people get the best result by drafting in one, tightening in the other, then checking the finished file separately.
Can ChatGPT or Claude check if my CV is ATS friendly?
Not properly. Both can advise on general formatting rules, such as avoiding tables or putting contact details in the body rather than the header. What neither can do is open your PDF and confirm the parser read every section correctly, because they see pasted text, not the structured file a screening system processes.
Is it safe to paste my CV into an AI chat?
Treat it like any other cloud upload. Check the data settings in your account and switch off model training if the option is offered. Strip out anything you wouldn't post publicly: full home address, ID or passport numbers, referee phone numbers. Your job history, skills and education carry far less risk.
Will recruiters know my CV was written with AI?
They notice the symptoms, not the tool. Generic phrasing, identical bullet structures and achievements with no specifics all read as machine-made. The fix isn't hiding it, it's editing: keep your real numbers, name the actual systems you used, and cut any sentence that could sit on somebody else's CV unchanged.
What's the best way to combine an AI assistant with a CV checker?
Draft and rewrite in the chat, where the language work happens. Export the finished CV as a PDF, then run that file through an analysis tool that parses it and scores each section. Fix what the report flags, reread it once as a human, and only then send it to an employer.