Spot the AI tells in your CV before a recruiter does
Upload your CV and get the exact lines that read as machine-written, plus a specific rewrite for each one.
What actually makes a CV read as AI-written
Recruiters rarely think "this was written by a chatbot." They think "this tells me nothing." That's the real tell. A drafted-by-AI CV describes the shape of a job instead of what you personally did in it: you spearheaded initiatives, you collaborated cross-functionally, you drove impact. Every claim is true of a hundred other candidates, so none of it helps a hiring manager choose you.
The second tell is rhythm. AI drafts tend to build every bullet the same way, with the same clause length and the same tidy verb-then-outcome pattern, so eight bullets blur into one. Human CVs are lumpier. One bullet names a tool and a number, the next is six words long, and that unevenness is exactly what makes it read as lived experience rather than a template.
None of this means you should stop using AI. It means you should use it as a first draft and then put your own facts back in: the system you migrated, the queue you cut from three days to four hours, the team of five you inherited mid-quarter. That's the work HRLens automates. It reads your CV the way a screener does, flags the lines that carry no information, and tells you what to replace them with.
Four things we check that recruiters notice first
Specificity
We flag bullets that could belong to anyone and ask for the tool, the team size or the system you actually touched.
Metrics
Claims without numbers get marked. You'll see which bullets need a figure, a timeframe or a before-and-after.
Phrasing
Repeated sentence patterns, stacked buzzwords and filler verbs show up highlighted, line by line.
Formatting
Tables, columns, text boxes and odd characters break ATS parsing. We show what the parser sees instead.
How it works
From upload to flagged lines in three steps
1. Upload the file you're actually sending
PDF or Word, in any language you apply in. No account setup before you see results.
2. Let it read the CV like a screener
It parses your file the way an ATS does, then scores each section on clarity and evidence.
3. Fix the flagged lines one by one
You get the weak bullets highlighted with a concrete rewrite prompt for each. Edit, re-upload, compare.
The phrases that keep showing up in AI drafts
| Line in the draft | Why it reads as AI | Write this instead |
|---|---|---|
| Spearheaded cross-functional initiatives | No team, no project, no result. Pure job-description language. | Led a 4-person squad across design and backend to ship the new checkout |
| Leveraged data-driven insights to optimise processes | Three abstractions in a row and not one concrete noun. | Rebuilt the weekly sales report in SQL, cutting prep from 6 hours to 40 minutes |
| Results-oriented professional with a proven track record | A self-description anyone can copy. Recruiters skip it. | Delete the line. Open with your role, years and the thing you're best at |
| Collaborated with stakeholders to drive impact | Says you worked with people. Everyone works with people. | Ran monthly reviews with finance and ops to agree the hiring plan for 3 regions |
| Passionate about delivering exceptional customer experiences | Adjectives instead of evidence, and the same opener as every other draft. | Handled 60+ support tickets a day and lifted CSAT from 4.1 to 4.6 in two quarters |
| Utilised industry-leading tools and technologies | Hides the tool names, which is exactly what the ATS is searching for. | Daily work in Jira, Figma and Looker; owned the reporting dashboard |
Find the empty lines before a recruiter skims past them
Upload the CV you're about to send and see exactly which bullets need your facts instead of adjectives.
Questions job seekers ask us about this
Can recruiters really tell if a CV was written by AI?
Not with certainty, and most don't try. What they notice is the symptom: bullets full of generic verbs and no evidence. A screener reading twenty CVs for the same role spots the pattern within seconds and moves on. So the fix isn't hiding the AI, it's putting your own numbers, tools and outcomes back into every line.
Do AI detectors work on CVs?
Poorly. Detectors were built for essays and long-form text, and a CV is short, fragmented and full of standard formatting, which pushes false positives high in both directions. A clean human-written CV can be flagged, and a polished AI draft can pass. Judge your CV on whether each bullet carries a specific fact, not on a detector score.
Is it a problem to use AI to write my CV at all?
No. Employers care about what you did, not which editor you opened. AI is genuinely good at structure, tightening long sentences and keeping tense consistent. It only becomes a liability when you ship the first draft untouched, because that draft invents nothing and knows none of your numbers. Use it to shape, then supply the substance yourself.
What does HRLens actually flag?
Bullets with no measurable outcome, repeated sentence patterns, stacked buzzwords, missing tool names, sections in an order that confuses a parser, and formatting that an ATS reads incorrectly, such as tables, columns and header-box contact details. Each flag comes with the rewrite direction, so you know what to add rather than just that something is wrong.
What if I genuinely have no numbers to add?
Most people have more than they think. Count what you handled: tickets per week, accounts managed, people trained, systems migrated, hours saved. If a number truly doesn't exist, name the thing instead. The specific client, product, tool or process you worked on carries almost as much weight as a percentage and it's still uniquely yours.
How long does a check take?
Upload a PDF or Word file and the analysis comes back in under a minute, with an ATS compatibility read, section-by-section feedback and a prioritised fix list. It's free to run, so most people check once before editing and once after, to see which flagged lines actually cleared.