Build resumes recruiters can read. See every gap before you send it.

An ATS resume is one whose text can be extracted and assigned to the right fields by the applicant tracking software an employer uses. In practice that means real selectable text rather than an image, a single column so nothing is read out of order, conventional section headings, and dates in a recognisable format. It is a document property rather than a design style: a plain resume can fail it and a well-designed one can pass.

You build the resume one section at a time with the page rendering beside you. As you type, three separate scores are computed in your browser: whether the document can be extracted, whether the writing is specific, and, once you paste a job description, how strongly it matches that posting. Every finding quotes the text it is about. When you export, preflight checks run first, and you get PDF, DOCX, plain text or a JSON backup.

No, and neither can anyone else. Workday, Greenhouse, Taleo, Lever and iCIMS parse and rank differently, and most employers configure their own rules on top of whichever they use. What this tool can do is show you the text an extractor receives from your document, tell you which conventional headings were recognised, and name every check it ran. That is a published checklist, not a prediction.

It is one of the three scores, and it answers a single question: can this document be read correctly by software? Ten categories are scored, covering text extraction, reading order, section recognition, contact detection, date recognition, page structure, links, unsupported characters, columns and tables, and font embedding. Each says whether it was measured from your document or is a rule of thumb. Things this workspace guarantees, such as having no images and no multi-column layout, are listed separately and score nothing, because a check that cannot fail is a claim rather than a check.

Resume Quality asks whether the content is clear, specific and convincing: section completeness, achievement evidence, action verbs, structure, readability, specificity, conciseness, skills coverage, repetition and formatting consistency. It is separate from Parseability because the two have different fixes. A well-written resume in an unreadable layout and an unreadable resume that is beautifully written would score identically on one blended number, and neither owner would learn anything.

Job Match is weighted coverage of the terms a specific posting emphasises. The posting is first tiered into essential requirements, desirable requirements, responsibilities and the role title; terms are then extracted with a quota per tier so a long essential list cannot crowd out the desirables. A matched term earns its full weight, a partial match earns half, and a missing term earns nothing. Separately, stated requirements for years of experience, degree level, certifications and seniority are checked against your own dates and education. Those are shown beside the percentage but deliberately do not change it.

No. Nothing is added unless you do it. Where a requirement is not evidenced, the matrix says so, quotes the line of the posting it came from, and offers to add the term, with the words add this only if it reflects work you have actually done beside the button. Keyword stuffing gets a resume through a screen and then fails in the interview, so this tool is built to show you gaps rather than paper over them.

It is not permitted to, and there is a check that does not rely on it complying. The rewrite prompt forbids introducing any employer, title, date, technology, certification, figure or claim of rank that is not already in your text. On top of that, every suggestion is diffed against the line you wrote and the rest of your resume: any number, year, named entity or superlative that appears in the suggestion but in neither source is flagged as a potential unsupported claim, and you are offered four ways out, which are to provide evidence, remove the claim, edit it, or keep it and accept that you are asserting it is accurate.

Yes, as PDF, DOCX or TXT up to 5 MB. The file is uploaded so the text can be extracted, because PDF and DOCX parsing needs libraries that do not run in a browser tab. Extraction is then reviewed field by field before anything is applied: you see which fields were found and which need checking, and you approve the result. If the file turns out to be image-only, nothing can be extracted, and that is worth knowing, because an employer's software would read exactly as little.

All three, plus a JSON backup. PDF is produced by your browser's own print engine from the same page you have been looking at, so the text stays selectable and the line breaks are the ones you approved. DOCX is a real Word document built from your content rather than a renamed PDF. TXT is exactly the text shown in ATS Read View. Nothing is paywalled and no export is watermarked.

Yes. Keep a master resume holding everything you have done, then branch a targeted copy per application. The copy is a deep copy, so trimming a variant can never reach back into the profile it came from, which is the whole reason to keep one. Each variant carries its own target job and its own match score.

A view of your resume as a text extractor receives it: the header lines, then each recognised section heading with the text beneath it, in the order it is read. It is not a mock-up. The text is produced by the same function that writes the plain-text export, and that function walks the same section order the printed page does, so what you see is genuinely the extraction order. Beneath it are the diagnostics that produced each verdict, including which headings were recognised and which were not.

They are the most common cause of scrambled extraction, because a parser reads cells in storage order rather than visual order, so a side rail of skills can end up interleaved with your job history. Every template here is single-column for that reason. That is also why the reading-order check in this workspace can be stated with more confidence than an uploaded-file checker manages: the order of the extracted text is the order of the rendered page, because both come from the same section order.

In your browser, and no. The document lives in this browser's local storage; there is no account, no sign-in and no sync, and no resume record is written on our side. Building, scoring, matching against a job description, previewing and every export happen in this tab. Only two optional actions reach a server, which are importing a file and the AI features, and neither is stored. Because nothing is synced, download the JSON backup before clearing site data or moving to another machine.