Buyer's Guide

AI CV Formatting for Recruitment and Staffing Agencies: What It Does, the 2026 Numbers, and What to Check

What AI CV formatting does for a recruitment or staffing agency, the 2026 adoption and workload statistics with sources, our own measured speed and accuracy, and seven checks before you buy.

AI CV formatting is software that reads the CV a candidate sent you, extracts the content into fields, and rebuilds it in your agency's template so it is ready to send to a client. It replaces the copy, paste and retype step, not the recruiter's judgement. The tool does not decide who is a good candidate. It decides where the dates go, which font the headings use and whether the phone number is on the page or not.

That distinction matters more in 2026 than it did a year ago. Candidates now write most CVs with AI, applications per role have doubled, and the EU has fixed a date for its rules on AI that evaluates people. This post puts the numbers in one place, with the source for each, shows what we measured on our own pipeline, and ends with the seven things to check before you buy any tool in this category. If you want to run the checks as you read, RefineCV gives you 10 free CVs with no card.

Key takeaways

  • AI CV formatting reads the CV a candidate sent, pulls the content into fields, and rebuilds it inside your agency's template. It does not screen, rank or score anyone. That is why it sits outside the EU AI Act's high-risk list for recruitment, which covers systems that filter applications or evaluate candidates.
  • The workload it attacks is growing. Applications per job on Greenhouse rose from 116 in 2022 to 244 in 2025, and 77% of US job seekers used AI in their search in the past year. More CVs arrive, in more shapes, and more of them were written by a machine.
  • Agencies that use AI anywhere in the process were 3.5 to 4.5 times more likely to have grown revenue in 2025, yet only 10% have AI embedded through the whole workflow. Admin and data entry is the task agency recruiters most often hand to AI.
  • On our own pipeline, a raw CV becomes a branded PDF in a median of about 3 seconds, the parser scored about 99 out of 100 on word-for-word field accuracy across five layouts, and one-click anonymisation left zero identifying fields behind. All three tests are published with their method.
  • Before you buy: test on your worst CV, insist on editable Word output, check anonymisation deletes rather than hides, read where the data goes and for how long, and price it per CV at your real monthly volume, not per seat.

What AI CV formatting does

The workflow has four steps in every product in this category.

  1. Parse. The AI reads the file (PDF, Word, a scan, a LinkedIn export) and extracts name, contact details, employers, titles, dates, education, skills and so on into structured fields.
  2. Map. Those fields are poured into your template: your logo, your section order, your consultant's details, the fields your client wants and none of the ones they do not.
  3. Review. A recruiter reads the result and fixes anything the parser missed or misread. This step is not optional and no honest vendor says it is.
  4. Export. PDF for the client, Word if the client edits, anonymised if the client should not see the name yet.

What it does not do is read the candidate's suitability. A formatter that also ranks or shortlists is a screening tool with a formatter attached, and it is regulated differently (see the EU AI Act section).

The 2026 numbers

Every figure below was read on the publisher's own page on 30 October 2026. Where a source is a vendor, it says so.

Applications per job (Greenhouse benchmark) North America Europe 116 99 2022 189 131 2023 223 159 2024 244 183 2025
Applications per job, 2022 to 2025. Greenhouse Hire Standard benchmark, March 2026: North America (over 6,000 companies) and Europe (over 800 companies). Both roughly doubled while recruiters per organisation fell by more than half.

More CVs, and more of them written by AI

  • Applications per job rose from 116 in 2022 to 244 in 2025, up 111%, while recruiters per organisation fell 56%. That is Greenhouse's North American benchmark across more than 6,000 companies and 640 million applications. Its Europe edition shows 99 to 183 per job, up 85%, with recruiters per organisation down 57% [1]. It is corporate hiring data, not agency data, but the same CVs reach agencies.
  • 67% of US HR leaders say reviewing AI-generated applications has slowed hiring, and 67% are using staffing firms for support. Robert Half, March 2026, more than 2,000 US hiring managers [2]. The overflow lands on agencies.
  • 77% of US job seekers used AI in their job search in the past year, and 60% have changed or considered changing how they write so their application is not mistaken for AI. Greenhouse, October 2026, 2,800 job seekers across five countries [3].
  • 78% of candidates across the UK, Ireland and Germany use AI to tailor their CVs or applications. Greenhouse, February 2026 [4].
  • Bullhorn's 2026 Talent Trends survey of nearly 2,300 people who had worked with a staffing firm found "three-quarters of talent are using AI" for tasks such as writing resumes and applying [5].
  • 34% of US employer respondents said they spend up to half their week filtering spam and junk applications. Greenhouse, November 2025, 1,236 recruiters and hiring managers [6].
  • 83% of UK recruiters encounter signs of AI-written applications frequently or occasionally, and 36% are less confident than a year ago that a CV reflects a candidate's true skills. CV-Library, September 2026, 233 recruiters and employers, so a small base [7].
  • Hiring managers think they can spot it: 80% say they can often tell when a CV was written by AI (Resume Genius, 1,000 US hiring managers, January 2026) [8]. In the one blind test found, 600 US hiring managers shown four CVs picked out the AI-written ones 33.5% of the time (TopResume, June 2025) [9]. Confidence is high, detection is not, which is one more reason a formatter must never rewrite the candidate's words.

Put together: the CVs landing in an agency inbox are more numerous, more uniform in tone and less uniform in layout than they were two years ago, because each one was shaped by whatever tool the candidate used.

What agencies are doing about it

  • Only 10% of staffing and recruitment firms have AI embedded throughout their workflow. A year earlier 52% were "only experimenting with basic generative AI"; now 29% remain there and 30% use some agentic AI. Bullhorn GRID 2026, nearly 2,300 recruitment professionals, surveyed November to December 2025 [10].
  • Firms using AI at any stage were "3.5 to 4.5 times more likely to have grown revenue in 2025", up from 25 to 40% more likely the year before. Same report [10]. In the UK and Ireland cut, 4 to 8 times [11].
  • 78% of recruitment agencies use AI, but only 21% can see measurable return from it, and only 6% run it as a core operating layer. The Access Group, June 2026, 159 agency leaders. Vendor survey [12].
  • Admin and data entry is the task agency recruiters most often give to AI: 80.8% of respondents in ATLAS's 2026 benchmark of over 1,000 agency recruiters, ahead of outreach (61.6%) and sourcing (49.3%). Vendor survey with thin methodology, so treat the ranking as the finding, not the decimal [13].
  • Recruiters spend an average of 17.7 hours per vacancy on administrative work, and 61% named manual data entry as a cause of delayed decisions. Totaljobs, 748 UK HR leaders, July 2025. In-house recruiters, not agencies [14].

Formatting is the data entry nobody counts. Every CV an agency sends to a client has been retyped, restyled or at least re-saved by someone on a salary.

Buyers now ask AI which tool to use

  • 51% of B2B software buyers start research with an AI chatbot more often than with Google, up from 29% a year earlier. G2, surveys of 1,076 (March 2026) and 1,169 (April 2025) B2B decision makers [15].
  • 94% of technology buyers who used AI in a purchase fact-check its answers at least some of the time. TrustRadius, 1,862 buyers, January 2026 [16].
  • On our own site, 29% of the Google queries that showed refine-cv.com in October 2026 were full questions of seven words or more ("our staffing agency processes hundreds of CVs a month, which tool..."), up from 0% in May. Google now counts AI Mode in Search Console and logs each follow-up as a new query [17]. Those questions are the brief for this post.

What we measured on our own pipeline

Industry surveys tell you what people say. These are what we timed and checked, each published with its method: the speed test, the parsing accuracy test and the anonymisation test [18, 19, 20].

WhatResultMethod
Time from raw CV to branded PDFMedian about 3 seconds, 2 to 4 across five layoutsThree timed runs per CV on the live pipeline; excludes human review
Parsing accuracyAbout 99 out of 100 on word-for-word field accuracy; lowest CV 96.2; two-column and sidebar CVs 99.8 and 99.2Five hand-checked CVs, scored field by field
AnonymisationEvery identifying field (name, email, phone, LinkedIn) gone from the output; zero leaksText extracted from each anonymised PDF and searched
Three first-party tests on the same five CVs (single column, dense, two-column, sidebar, five employers). Each is published in full with its method.

Five CVs is a sample, not a benchmark, and we say so on each page. We publish them because nobody else in this category publishes per-field numbers at all.

Why the review step stays

On the hardest CV the parser inferred an age the document never stated and listed a few skills that appeared in job descriptions but not the skills section. You see and edit every field before export, so none of that reaches a client. Budget a minute of review per CV, every CV.

Where formatting sits under the EU AI Act

Annex III of the EU AI Act lists, as high risk, "AI systems intended to be used for the recruitment or selection of natural persons, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates" [21]. A tool that only re-lays out a CV does not filter applications or evaluate anyone, so on the Annex's own wording it is not in that list. A tool that also screens, scores or shortlists is.

The timing changed this summer. The AI Omnibus, published in the Official Journal on 24 July 2026, moved the Annex III high-risk obligations from 2 August 2026 to 2 December 2027 [22]. The classification itself did not change. If your shortlist includes a "formatting plus matching" product, ask the vendor in writing which side of that line it sits on. We are not lawyers; your DPO should read the Annex, which is two paragraphs.

GDPR applies to all of it regardless. The relevant questions are where the file goes, how long it is kept and whether the original is deleted. Our answers are on the security page (EU hosting, original file deleted after processing, formatted documents in private storage your agency controls) and in the GDPR guide for candidate CVs.

Seven checks before you buy

  1. Test on your worst CV, not the vendor's sample. A two-column CV with a sidebar and a photo. If the parser scrambles it, nothing else matters. Ask for the vendor's accuracy numbers and how they were measured; if there are none, that is the answer.
  2. Insist on editable output. Clients edit. A PDF-only tool sends your consultant back to Word. RefineCV exports PDF and Word, and you edit the document on screen before export.
  3. Use your own templates. Many agencies hold client-owned Word templates. Check the tool rebuilds those, not just its own designs. RefineCV's template builder is included on every plan, and the team rebuilds a client's Word template for you free.
  4. Check anonymisation deletes rather than hides. A black box over live text is a GDPR incident waiting for a copy-paste. Extract the text from an anonymised output and search it for the name [20].
  5. Read the data path. Where is the file processed, where is the output stored, when is the original deleted, is your agency's data isolated from other agencies'. These should be written on a page, not in a sales call.
  6. Price it per CV at your real volume. Seat pricing hides the number that matters. At 100 CVs a month RefineCV's pay-as-you-go is $40; at 500 a month Pro is $50 for 200 plus $0.30 for each of the other 300, $140. Unlimited team seats on both [23]. Work the same sum for every vendor on the list. Many sell a monthly allowance or a per-user licence, so convert each to a cost per CV at your volume. Our 2026 pricing benchmark does that sum for 18 tools.
  7. Ask about your ATS honestly. If your Bullhorn, Loxo or JobAdder workflow needs a native integration, ask for it by name. RefineCV has no Bullhorn integration today; agencies on Bullhorn download the CV, format it, and attach the output, which is the two-minute workflow described here. A vendor that claims every integration should show you each one.

Frequently asked questions

Is AI CV formatting software high-risk under the EU AI Act?

Not on the Annex's wording. Annex III lists systems used to place targeted job adverts, analyse and filter applications, and evaluate candidates. A tool that only rebuilds a CV in a template does none of those. A tool that also screens or ranks candidates does, and its high-risk obligations now apply from 2 December 2027.

Does AI formatting change what the CV says?

It should not. The content is extracted and re-laid out; the words stay the candidate's. The exception is anonymisation, where you choose which identifying fields to remove. Review every output before it goes to a client.

Can it handle CVs that candidates wrote with AI?

Yes. An AI-written CV is text in a layout like any other; the parser does not care who wrote it. What changes is volume and the spread of layouts, which is the reason to automate the step.

How long does AI CV formatting take per CV?

On our pipeline a raw CV became a branded PDF in a median of about 3 seconds across five layouts, measured end to end and excluding human review. Add a minute of review per CV.

What does AI CV formatting cost at agency volumes?

RefineCV is $0.40 per CV with no minimum, or $50 a month for 200 CVs ($0.25 each) and $0.30 per CV above that, with unlimited team seats. 10 CVs are free with no card. Competitors price by monthly allowance, credits or seats, so convert each to a cost per CV at your volume before comparing.

The bottom line

The 2026 numbers point one way. Candidates write CVs with AI, volume per role has doubled, and the agencies growing fastest are the ones that automated the admin first. CV formatting is the admin that touches every placement and is the easiest to automate without touching the decision, which is also why it sits outside the AI Act's high-risk list.

Pick a tool the way you would pick a consultant: test it on real work, read what it publishes about accuracy and data, and price it at your volume. RefineCV publishes all three. Related reading: the best CV formatting software for recruitment agencies, and for US readers, resume formatting software for staffing agencies.

Test it on your worst CV

Upload a two-column CV with a photo and see every parsed field before anything is exported. Start free with 10 CVs, no credit card. Then $0.40 per CV, or $50 per month for 200.

Start Free, 10 CVs

Sources

Bracketed numbers in the text refer to this list. Every external figure was read on the publisher's own page on 30 October 2026.

  1. Greenhouse, "The Hire Standard" recruiting benchmarks, North America (March 2026): Over 6,000 companies and 640M+ applications, 2022 to 2025: applications per job 116 to 244 (+111%), recruiters per organisation 10.43 to 4.62 (-56%). Europe edition: 99 to 183 per job (+85%), recruiters per organisation down 57%.
  2. Robert Half, "67% of HR leaders report AI-generated applications are slowing hiring" (10 March 2026): More than 2,000 US hiring managers, fieldwork November 2025. 67% say reviewing AI-generated applications has slowed hiring; 67% are using staffing firms for support.
  3. Greenhouse, "6 in 10 job seekers are rethinking how they write to avoid sounding like AI" (20 October 2026): 2,800+ job seekers in the US, UK, Ireland, Germany and Australia; US figures n=1,200. 77% used AI in their search in the past year; 60% have changed or considered changing how they write.
  4. Greenhouse UK, "Why transparency now defines great hiring", quoting The 2026 AI in Hiring Report: UK, Ireland and Germany (February 2026): 1,700 jobseekers, 198 recruiters and 373 hiring managers. 78% of candidates use AI to tailor their CVs or applications. Full report gated.
  5. Bullhorn, GRID 2026 Talent Trends Report, press release (23 September 2026): Nearly 2,300 people who worked with a staffing firm in the past three years. "Three-quarters of talent are using AI" for tasks such as writing resumes and applying.
  6. Greenhouse, "An AI trust crisis", 2025 AI in Hiring Report (19 November 2025): 4,136 respondents including 1,236 recruiters and hiring managers. 34% of US employer respondents spend up to half their week filtering spam and junk applications.
  7. CV-Library, Candidate Behaviour Barometer 2026, reported by Onrec (10 September 2026): 1,043 UK jobseekers and 233 recruiters and employers, fieldwork August 2026. 83% of recruiters encounter signs of AI-written applications frequently or occasionally; 36% are less confident a CV reflects true skills.
  8. Resume Genius, "2026 Hiring Insights Report" (29 January 2026, updated 24 March 2026): 1,000 US hiring managers, Pollfish. 80% say they can often tell when a resume was written by AI. Self-reported.
  9. TopResume, "Survey: Where employers draw the line on the use of AI in hiring" (3 June 2025): 600 US hiring managers, Pollfish, 15 to 16 May 2025. Shown four resumes, 33.5% spotted the AI-generated ones. The only study found that tests detection rather than asking about it.
  10. Bullhorn, GRID 2026 Industry Trends Report (25 February 2026): Nearly 2,300 recruitment professionals in North America, UK and Ireland, Benelux, DACH and APAC, surveyed November to December 2025. Only 10% have AI embedded throughout the workflow; 52% were only experimenting a year earlier, 29% now; firms using AI at any stage 3.5 to 4.5 times more likely to have grown revenue in 2025. Press release.
  11. Bullhorn UK, GRID 2026 Industry Trends Report, UK and Ireland press release (25 February 2026): Nearly 300 UK and Ireland recruitment professionals. Agencies using AI at any stage 4 to 8 times more likely to have grown revenue in 2025.
  12. The Access Group, "AI in Recruitment: 2026 Outlook" (June 2026): 159 recruitment agency leaders (sample size from the APSCo listing). 78% use AI; 21% see measurable ROI; 6% run AI as a core operating layer. Vendor survey. APSCo listing.
  13. ATLAS, "The State of Agency Recruitment: 2026 Benchmark Report" (16 March 2026): 1,000+ agency recruiters; countries and fieldwork dates not stated. 80.82% use AI for admin and data entry, 61.64% for outreach, 49.32% for sourcing. Vendor survey.
  14. Totaljobs, "What's slowing down your hiring?", reported by People Management (19 August 2025): 748 UK HR leaders and 2,025 jobseekers, 25 June to 3 July 2025. 17.7 hours of administrative work per vacancy; 61% cite manual data entry as a cause of delayed decisions.
  15. G2, "The Answer Economy: G2 2026 AI Search Insight Report" (March 2026): 1,076 B2B decision makers. 51% start software research with an AI chatbot more often than with Google. 2025 Buyer Behavior Report (1,169 decision makers, April 2025): 29%.
  16. TrustRadius, 2026 B2B Buying Disconnect Report (15 July 2026): 1,862 technology buyers and 444 vendors, January 2026. 94% of buyers who used AI fact-check its responses at least some of the time.
  17. Google Search Console Help, "What are impressions, position, and clicks?" (read 30 October 2026): "If a user asks a follow-up question within AI Mode, they are essentially performing a new query." Our own figure is from Search Console for refine-cv.com: query rows of seven or more words, May against October 2026.
  18. RefineCV, "How long does automated CV formatting take?" (October 2026): First-party test: five CVs, median of three timed runs on the live pipeline.
  19. RefineCV, "How accurate is CV parsing?" (October 2026): First-party test: five hand-checked CVs scored field by field.
  20. RefineCV, "Does CV anonymisation actually remove the data?" (October 2026): First-party test: text extracted from each anonymised PDF and searched for the candidate's identifying fields.
  21. EU AI Act, Annex III, point 4(a) (Regulation (EU) 2024/1689): High-risk: "AI systems intended to be used for the recruitment or selection of natural persons, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates".
  22. Future of Privacy Forum, "The AI Act implementation timeline: what changes under the AI Omnibus" (28 July 2026, updated 5 August 2026): The AI Omnibus was published in the Official Journal on 24 July 2026; the Annex III high-risk requirements apply from 2 December 2027 rather than 2 August 2026.
  23. RefineCV pricing (October 2026): $0.40 per CV pay-as-you-go with no minimum; Pro $50 a month for 200 CVs, then $0.30 per CV; unlimited team seats; 10 free CVs with no card.

The RefineCV Team

Written by the team building RefineCV, CV formatting software for recruitment agencies.

Format your next CV in 10 seconds

Try RefineCV with 10 free CVs. No credit card. Then $0.40 per CV, or $50/month for 200 on Pro.

Start Free, 10 CVs

No credit card required