Do companies use AI to screen resumes? What happens to yours

Do companies use AI to screen resumes?

Some do, and the tools are now built into the systems many employers already use to collect applications. Greenhouse sells a Talent Matching feature, part of its Real Talent add-on, that sorts applicants into match categories, and Workday's HiredScore assigns candidates a letter grade based on how well their resumes match the job description. Neither vendor's page lists which employers use these features, so applying through one of these systems does not by itself tell you your resume was scored.

We could not find a current, primary-source survey that reliably measures what share of employers use AI to screen resumes, so this guide gives no percentage. What the vendor documentation does show is the shape of the process: software reads your resume, compares it with criteria the hiring team set, and puts you in a ranked or labelled pile that a recruiter then reviews.

That means two things can go wrong for you. The software can misread your resume, or it can read it correctly and still conclude you match poorly because the evidence the posting asks for is missing or buried. You can control both.

What ATS parsing does, and what AI matching adds

An applicant tracking system, or ATS, is the software behind most online job applications. Its oldest job is parsing: pulling your name, contact details, job titles, employers, dates, education and skills out of the file you upload and storing them as fields a recruiter can search and filter. Parsing is not judgment. It is extraction, and it fails in boring ways: text inside images, unusual section names, tables that scramble reading order, or dates the system cannot recognize.

AI matching sits on top of parsing. Instead of only letting a recruiter search for a word, it compares what was extracted with the requirements of the role and produces a score, grade or category. Greenhouse describes this in its own help center: the hiring team first sets up a calibration of required skills, relevant experience, job titles and optionally industry, with a weight for each, and Talent Matching then sorts the pool into categories such as Strong, Good, Partial, Limited or Needs manual review.

Greenhouse also says that for skills it compares related terms on each resume with the calibrated skills, and that several terms can map to the same skill, so a longer list of matched terms does not always mean a higher match. In plain terms, repeating a keyword is not the same as showing the skill.

• Parsing: reads your file and fills in structured fields. Formatting problems hurt here.
• Keyword search and filters: recruiters search those fields for skills, titles or locations. Missing or oddly worded terms hurt here.
• AI matching or grading: software compares your fields with weighted job criteria and sorts you. Weak evidence for the main requirements hurts here.
• Human review: a recruiter or hiring manager reads the sorted list and decides. Everything on the page matters here.

What Greenhouse and Workday say their AI does

Vendor documentation is the most concrete public evidence of how AI screening works, because it describes the feature for the employers who buy it. Here is what two large systems say, in their own terms.

Workday's HiredScore reference page says it assigns each candidate a grade from A to D in its Spotlight view and on a Pending Actions dashboard, based on how well the resume matches the skills and qualifications in the job description. Grade A means the candidate meets all basic, or required, qualifications, most preferred ones, and has an average match above a set threshold. Grade B means all basic qualifications are met. Grade C means most but not all are met, and Grade D means most are not met. The page is dated 2023-06-23, so check it for changes.

Greenhouse's Talent Matching FAQ says the feature is assistive, does not automatically disposition candidates, and does not auto-reject or auto-advance anyone, including people in lower match categories. It says recruiters and hiring managers make all advancement and rejection decisions. Greenhouse also says employers can turn on an opt-out that lets candidates ask for review without AI, and that applications the system cannot analyze, for example because of rare formatting or technical issues, are labelled Needs manual review for a person to read.

• Both systems grade against the job's stated requirements, so the requirements section of the posting is your best guide to what is being measured.
• Workday's grades turn on required qualifications first, so make sure each one you meet is plainly visible.
• Greenhouse's categories depend on criteria each hiring team sets, so two employers on the same system can weigh your resume differently.
• Neither page says the employer must tell you which grade or category you received.

Which systems large employers' job listings run on

Applystead's Job Market Report 2026, with data captured September 30, 2026, counted 7,077,787 open job listings on employers' career sites across 9 applicant tracking systems. Workday hosted 3,301,159 of them (46.6%), followed by SmartRecruiters (30.5%) and Oracle Taleo (12.9%). Together those three hosted 90.1% of open listings. iCIMS hosted 5.3%, Greenhouse 2.1%, and Ashby and Lever 0.8% each. The report notes these are shares of listings, not of companies, and that employers on other systems are not included.

So if you apply to large employers, there is a good chance your resume goes through Workday, SmartRecruiters or Taleo first. Whether any AI ranking is switched on is a separate choice each employer makes. You can see which employers run each system on our pages for companies that use Workday, companies that use SmartRecruiters, companies that use Taleo, companies that use iCIMS and companies that use Greenhouse, or look up a single employer with the which application system tool.

What the rules say about AI in hiring, as of October 5, 2026

A few US jurisdictions regulate automated hiring tools. Each rule below is summarized from the official government page as it read on October 5, 2026. This is general information, not legal advice; if you think a rule was broken in your case, contact the agency named or a lawyer.

We did not include court cases or federal agency guidance, because we could not confirm them from an official source today within this guide's source limits. Laws in this area change quickly, so check the linked official pages before relying on any of this.

• New York City: as of October 5, 2026, the city's Department of Consumer and Worker Protection page says Local Law 144 of 2021 bars employers and employment agencies from using an automated employment decision tool unless it has had a bias audit within one year of use, information about the audit is publicly available, and required notices have been given to candidates or employees. The page says enforcement began July 5, 2023, that the notice must be given 10 business days before the tool is used, and that you can file a complaint with the department.
• Illinois: as of October 5, 2026, 820 ILCS 42/5 on the Illinois General Assembly site says an employer that asks applicants to record video interviews and uses artificial intelligence to analyze them, for positions based in Illinois, must tell each applicant beforehand that AI may be used, explain how it works and what general types of characteristics it evaluates, and get the applicant's consent. Employers may not use AI to evaluate applicants who have not consented. This covers video interviews, not resumes.
• Colorado: as of October 5, 2026, the Colorado General Assembly page for Senate Bill 26-189 shows it was signed May 14, 2026. Its summary says it repeals and reenacts the 2024 artificial intelligence law, Senate Bill 24-205, with requirements on automated decision-making technology used in consequential decisions, which include employment. Developer documentation duties start January 1, 2027, deployers must give consumers notice at the point of interaction, people can request human review after an adverse decision, and the attorney general enforces it.

How to format a resume so the system reads it correctly

Parsing problems are the one part of screening that is purely mechanical, so they are worth eliminating before you think about matching. Greenhouse's own FAQ lists rare formatting or technical issues as one reason it cannot analyze a resume, which sends it to a person instead, so a clean file is the safer default. Our guide to an ATS-friendly resume covers the layout in detail; these are the checks that catch most problems.

1. Use a single-column layout with standard section headings such as Experience, Education and Skills.
2. Keep all text as real text: no text inside images, logos or charts, and nothing important in headers or footers.
3. Write each job as title, employer, location and dates on their own lines, using one consistent date format.
4. Save as a .docx or text-based PDF unless the posting asks for something else, and check that you can select and copy the text.
5. Paste the resume into a plain-text box or run it through an ATS resume checker to see what a parser is likely to extract, then fix anything that comes out garbled or out of order.

How to tailor for AI matching without keyword stuffing

Both vendors describe grading against the job's requirements, so tailoring is about making true evidence for those requirements easy to find, not about gaming a score. Start by marking the required qualifications in the posting, then the preferred ones. For each requirement you genuinely meet, make sure the resume names it in the posting's own words at least once and shows where you used it.

Use a resume keyword checker to compare your resume with the posting and spot real gaps, then close them only with experience you actually have. Our guides on resume keywords and how to tailor your resume to a job description walk through the process. Avoid white text, pasted job descriptions and long keyword blocks: they add nothing a matching system values, and the person who reads the shortlist will see them.