The talent market at a glance.

Key takeaways

• In a pre-registered field experiment with 67,056 randomized applications, applicants interviewed by an AI voice agent were 12% more likely to receive a job offer and 18% more likely to still be employed a month later than those interviewed by recruiters, with people making every hiring decision.arXiv (Jabarian and Henkel)
• Given the choice, 78% of applicants picked the AI interviewer, yet 7% of AI-led calls failed technically and 5% of applicants refused to continue with an AI, so every deployment needs a human fallback.arXiv (Jabarian and Henkel)
• Candidates punish badly designed AI interviews: 38% of US job seekers have walked away from a process because it included one, and 70% were never told upfront that AI would evaluate them.Greenhouse
• One-way recorded interviews deter applicants far more than live ones, cutting application continuation by over 50% in a randomized US experiment, even though the AI scoring in the same study predicted later employment better than recruiters did.CESifo (Avery, Ip, Leibbrandt and Vecci)
• Workday valued its acquisition of the conversational screening agent Paradox at $1.1 billion, as applicant tracking suites absorb the AI interview the way they absorbed interview notetakers.Workday
• The EU's AI Omnibus, in force since 27 July 2026, moved the high-risk obligations that cover AI used to evaluate job candidates to 2 December 2027.European Commission

Key takeaways

• LinkedIn's agentic-AI hiring products, led by Hiring Assistant, are on track for roughly $450 million in annual sales, the demand-side signal that reframed autonomous recruiting from a demo into a real, paid market.MicrosoftReuters via Yahoo Finance
• The word agentic is badly overused: Gartner estimates only about 130 vendors offer genuinely agentic capabilities out of thousands making the claim, so the three-question test (holds a goal, sequences its own actions, uses tools in a continuous loop) is the buyer's defense against agent washing.MarTech
• The honest ROI is bounded, not the vendor ceiling: organizations deploying AI in recruiting report roughly 30% lower cost-per-hire and about 31% faster hiring, far from the up-to-70% time-to-hire cut vendors cite as an upper bound.InCruiterPin
• Automating the funnel without human warmth is corrupting the pipeline it builds: job seekers ghosted by employers hit a three-year-high 53% in 2026 (up from 48% in 2025 and 38% in 2024), and roughly one in three candidates have walked away rather than sit through a one-way AI interview.FortuneReadySetExec
• The market has converged on human-in-the-loop autonomy: 52% of talent leaders plan to add autonomous AI agents in 2026 and 84% plan to use AI in some form, yet recruiter judgment overrides the AI in 58% of organizations when the two conflict, and 71% of Americans oppose AI making the final hiring call.Korn FerryiCIMS and Aptitude ResearchPew Research Center
• Supervision intent outruns supervision infrastructure: about 45% of companies report no formal AI governance framework in place, meaning nearly half of the organizations deploying these tools have no defined human checkpoint, no audit trail, and no bias-testing cadence.iCIMS and Aptitude Research

Key takeaways

• As of the 7 June 2026 transposition deadline, only 4 of the EU's 27 member states (Slovakia, Italy, Lithuania, and Malta) had complete national pay-transparency law in force, so the 'single European rule' is in practice a staggered, multi-year rollout stretching to 2028 for laggards like Germany and France.TrusaicPinsent Masons
• The directive does more than mandate salary ranges in job ads: it bans asking candidates about pay history, gives every worker an Article 7 right to a written pay-comparison response within two months, and forces a joint pay assessment whenever an unexplained gender pay gap of 5% or more appears in any worker category, which is why compliance requires a defensible job architecture and continuous pay-equity analytics rather than a job-ad tweak.European CommissionEurope HR Solutions
• Behavior lags the law and is driven by culture and enforcement more than the letter of the rule: as of March 2026 the UK (not even in the EU) led Europe with salary information in about 56% of job postings while Germany, the EU's largest economy, trailed at just 12%.Indeed Hiring Lab
• The US is the working preview of what the directive does, and its readiness data is sobering: Mercer found only about 54% of US employers felt prepared for pay-transparency compliance in 2025 (up from 39% in 2024) and only about 17% had fully implemented their approach, showing the binding constraint is executing the multi-quarter build, not awareness.Mercer
• Colorado shows enforcement, not the disclosure mandate alone, drives behavior: its Equal Pay for Equal Work Act carries per-posting fines of $500 to $10,000, and DaVita Inc. settled a Colorado pay-transparency penalty for $298,000 in 2025, turning a missing range into a material, per-req liability at scale.LegalClarityGoverning
• Transposition demonstrably moves behavior but with a lag: after Italy transposed the directive, the share of its job postings including salary information rose from about 22% to 36% within twelve months, evidence the directive works while a working directive still leaves roughly two-thirds of postings non-disclosing in year one.Indeed Hiring Lab

Key takeaways

• Roughly 70 to 85 percent of employers now claim skills-based hiring, yet a Harvard and Burning Glass Institute study found fewer than 1 in 700 hires (about 97,000 of 77 million annual hires) were actually affected by dropped degree requirements. The gap between announcement and practice is the central fact of the debate.The Burning Glass Institute and Harvard Business School
• Among companies that announced skills-based hiring, about 45 percent made no real change in practice, about 37 percent followed through as genuine leaders and raised non-degree hiring by nearly 20 percent, and about 18 percent were backsliders who reverted. The 37 percent is the true operating rate, not the survey headline.The Burning Glass Institute and Harvard Business School
• AI has broken the unproctored take-home: CodeSignal detection data found 35 percent of proctored assessments were flagged for cheating in 2025, more than double the 16 percent of 2024, rising to 40 percent on entry-level tests and 48 percent in Asia-Pacific. HackerRank reports 83 percent of candidates would use AI if they believed it would go undetected.CodeSignal (PR Newswire)HackerRank
• The measurable return is real but conditional: an IBM study reported AI recruiting cut time-to-hire about 24 percent, and PwC's skills-first approach across its EMEA and Asia Pacific Deals practices with Workday cut mean hiring time 45 percent, a gain that came from wiring the skill signal into the system of record rather than bolting on a standalone test.PwC (with Workday)WeCreateProblems (citing IBM)
• Skills scarcity is the backdrop forcing the shift: the World Economic Forum found 63 percent of employers cite skills gaps as the biggest barrier to transformation over 2025 to 2030, with 39 percent of workers' core skills expected to be transformed or obsolete by 2030. SHRM found 43 percent of organizations now use AI for HR tasks, up from 26 percent in 2024.World Economic ForumSHRM
• The proxy is measurably dying at the screening stage: NACE found only 42 percent of employers now screen candidates by GPA, down sharply from 73 percent in 2019, a 31-point collapse consistent with cheap skill measurement displacing cheap proxy measurement at the top of the funnel.National Association of Colleges and Employers

Key takeaways

• A single AI vendor, Workday, now sits as the named defendant in a nationwide collective action tied to roughly 1.1 billion rejected job applications, an order of magnitude no single-employer discrimination case has ever approached.Proskauer (Law and the Workplace)
• The Northern District of California held that an AI hiring vendor can be sued directly as an agent of the employer, and as an employer itself, under Title VII, the ADEA, and the ADA, shifting bias liability from buyer to builder.Seyfarth Shaw
• Vendor-side audit data show audited AI systems averaging a 0.94 four-fifths impact ratio versus a 0.67 human baseline, but 15% of audited systems still fail at least one demographic threshold, and a single failing class in a single deployment is all a plaintiff needs.Warden AI
• The litigated risk concentrates in a class the assurance market barely measures: age was tested in only 5% of audits and disability in 5%, while Mobley's certified collective is an ADEA age-discrimination case.Warden AI
• A 2024 University of Washington study of over 3 million resume-job comparisons found leading text-embedding models favored white-associated names 85% of the time and female-associated names only 11% of the time, showing name-based bias can be baked into the embedding layer.University of Washington
• A January 2026 class action reframes AI candidate scoring under the Fair Credit Reporting Act rather than disparate impact, alleging Eightfold AI scraped data on over one billion workers and scored applicants zero to five, a theory immune to the DOJ's constitutional attack on disparate impact.Built In
• The federal disparate-impact machinery is being dismantled just as enforcement migrates to private litigation: in June 2026 the DOJ's Office of Legal Counsel concluded the EEOC's disparate-impact guidelines are unconstitutional.U.S. Department of Justice
• The bias-assurance layer is monetizing the exact liability the litigation created: Warden AI raised in July 2025, reaching roughly $1.6M in total funding at a $10M valuation, on surging demand for independent HR-AI audits driven directly by Mobley v. Workday.Tech Funding News

Key takeaways

• Entry-level hiring did not collapse uniformly: highly AI-exposed entry-level postings fell about 40% from January 2023 to August 2025 versus a 33% decline for low-exposed roles, and the senior pattern inverts (high-exposed senior postings fell 27% versus 16% for low-exposed), showing AI fragility runs by seniority.Revelio Labs
• Two independent instruments converge on the same seam: Stanford's ADP payroll data puts employment among workers aged 22-25 in highly AI-exposed occupations about 19% below where it would be had it kept pace with less-exposed peers of the same age (data through June 2026), while experienced workers show no comparable gap, and the adjustment ran through reduced hiring rather than wage cuts.Stanford Digital Economy Lab
• The core driver of graduate labor-market weakness is a depressed hires rate, not a layoff wave: employers are not firing junior staff in unusual numbers, they are simply not opening the door, which means a freeze that could thaw quickly rather than a structural deletion.Economic Policy Institute
• The AI-exposed decline began before ChatGPT: unemployment-insurance records show risk rising in AI-exposed occupations starting in early 2022, months before ChatGPT's late-2022 release, suggesting AI acted as an accelerant on a pre-existing skill-shift slope rather than the origin of the trend.arXiv
• Aggregate studies find no clear AI signal because exposure is concentrated: the NY Fed found fewer than 10% of workers and vacancies sit in occupations with an AI exposure score of 0.4 or higher, while about 40% of workers are in positions with zero measurable AI exposure, so a real seam-level effect can stay invisible in national averages.Liberty Street Economics
• The bifurcation plays out inside single org charts: Salesforce is hiring 1,000 new grads and interns to 'ride the AI exponential' after reducing customer-support headcount from about 9,000 to roughly 5,000 as AI agents took over support tasks.Fortune
• Physical presence and regulated sectors are the strongest moats: healthcare entry-level postings rose about 13 percentage points against the broad downtrend, because a model cannot insert an IV or bear the legal liability of signing a regulated filing.Rezi
• AI fluency is the highest-leverage graduate investment: the arXiv study found graduates who took more AI-exposed coursework had higher first-job pay and shorter job searches after ChatGPT, pointing to fluency becoming a hiring differentiator once model capabilities became broadly usable.arXiv

Key takeaways

• The barriers to faking a candidate have all collapsed: a Unit 42 researcher with no image-manipulation experience built a real-time deepfake capable of passing a live video interview in 70 minutes using a five-year-old consumer GPU and free tools.Unit 42
• Practitioners now see fraud as routine: in Checkr's survey of 3,000 managers, 31% interviewed a candidate later revealed to have a fake identity, 35% confirmed a proxy participated in a virtual interview, and 23% reported hiring-fraud losses over $50,000 in the past year.Checkr
• There is a wide detection-confidence gap: 62% of managers believe job seekers are now better at faking identity than HR is at detecting it, and only 19% are extremely confident they could catch a fraudulent applicant.Checkr
• The DPRK IT-worker pipeline is a prosecuted, industrial-scale operation: an April 2026 case brought a nine-year federal sentence for placing fraudulent workers at more than 100 US companies.DOJ
• North Korea-linked actors accounted for 47% of all state-backed activity targeting the US tech sector between April 2025 and May 2026, per CrowdStrike's 2026 Global Threat Report.TechCrunch
• The hiring-fraud kill chain ends in catastrophic breaches: the record $1.46 billion Bybit crypto theft was executed by compromising a third-party developer's laptop and credentials, making the hiring decision and the breach the same event separated by time.Fortune
• Detection structurally trails generation: Gartner projects 30% of enterprises will consider identity verification unreliable in isolation by 2026 due to AI deepfakes, and HYPR found 95% of organizations experienced a deepfake incident in the prior year with nearly 40% suffering a GenAI-related breach.Gartner
• A verification stack materialized in a single quarter, and its fast-growing edge is hiring: the employee-onboarding identity verification market is projected to reach $6.01 billion by 2031 at a 16.67% CAGR, outpacing the overall IDV market's 11.18% growth.Mordor Intelligence

Key takeaways

• The global staffing market is large and stagnant (Staffing Industry Analysts put it at roughly $620 billion in 2024 and expects 2025 essentially flat, still below the $648 billion 2022 peak), so individual firm growth now depends almost entirely on taking share rather than riding a rising tide.Staffing Industry AnalystsStaffing Industry Analysts
• AI adoption is now the clearest dividing line in staffing: top-performing firms are four times more likely to use AI, 78% of high-growth firms use AI in their ATS, and 56% of the fastest-growing firms place candidates in under 10 days.Bullhorn
• The gap between AI adopters and laggards is set to widen because agentic AI is still early: only about 10% of agencies run agentic AI across their full workflow, and 45% cite data concerns (not technology) as the binding constraint, meaning data-readiness work must start now.HR Tech FeedBullhorn
• Independent data confirms the AI productivity shift: recruiter call time hit a record 286 minutes per week in Q1 2026 (double early 2024), interactions rose 60% year over year, and recruiters now average 1.36 AI tools, showing humans being redeployed from sourcing busywork to relationship work.American Staffing Association
• Contingent demand is not disappearing but relocating: roughly 76% of companies now use freelancers or contractors as a permanent strategy, while that demand migrates from traditional agencies toward marketplaces, EOR infrastructure, vetted networks, and AI-native platforms.StaffingHub
• Bullhorn is rolling up an end-to-end agentic staffing stack (Textkernel for sourcing AI, TargetRecruit for Salesforce-native healthcare operations, and the Amplify agentic layer), and adoption by giants like The Adecco Group signals the agentic model is becoming the industry's operating assumption.Staffing Industry AnalystsBullhornBullhornBullhorn
• AI acts as a filter that pushes value up the stack: commodity freelance work is being absorbed by models directly while high-value work grows, evident in Upwork's record $5,100+ GSV per client and Fiverr's high-value projects (over $1,000) growing 23% to about 15% of GMV.The Motley FoolSEC
• Capital is repricing staffing value into software economics: AI-native and infrastructure platforms (Deel at $17.3B with $1B+ profitable ARR, Mercor at $10B, Scale AI at $29B) command valuations that dwarf larger-revenue traditional staffing firms, which trade at services multiples in a flat market.CalcalistHR GrapevineTechCrunchWikipedia

Key takeaways

• Only 28% of companies are satisfied with their current applicant-tracking system, 82% report significant gaps, and one in four plan to replace it this year.Eightfold AI
• SAP completed its acquisition of SmartRecruiters on September 11, 2025, and plans to retire the SuccessFactors recruiting module entirely and migrate its customers to SmartRecruiters.SmartRecruiters
• SuccessFactors recruiting customers will have three to five years to migrate to SmartRecruiters, with the SAP-enabled version available for implementation in March 2026.CIO
• By the App Runs the World measure iCIMS holds the number one ATS position at roughly 11% share, supporting 25% of the Fortune 500.iCIMS
• Workday, Oracle, and SmartRecruiters were all positioned as Leaders in the 2025 Gartner Magic Quadrant for Talent Acquisition (Recruiting) Suites.Oracle
• Workday is used by more than 11,000 organizations worldwide, including more than 65% of the Fortune 500.Workday
• Workday's Illuminate AI agents drove a 72% increase in screening efficiency at General Motors and enabled AdventHealth to reduce agency spend by around $68 million in twelve months.Workday (HiredScore customer story)Workday
• The worktech sector, which includes the ATS market, saw between 90 and 100 companies acquired every quarter through 2025, with 110 acquisitions in the third quarter alone.Venero Capital

Key takeaways

• Structured interviews rank as the most predictive selection procedure in the 2022 re-analysis by Sackett and colleagues, with a validity of about 0.42, down from the 0.51 Schmidt and Hunter reported in 1998 and more than double the 0.19 for unstructured interviews.Journal of Applied Psychology (Sackett, Zhang, Berry and Lievens)
• Unstructured interviews predict job performance at a validity of only about 0.19 in the 2022 re-analysis, less than half the 0.42 of structured interviews, which makes structure the strongest single lever on interview quality.Journal of Applied Psychology (Sackett, Zhang, Berry and Lievens)
• Automatic speech recognition costs fell roughly 92% from 2021 to 2025, from about $0.25 per minute to $0.02 per minute, as free transcription became bundled into every major meeting platform.Tool Directory
• HireVue has hosted over 70 million video interviews and run more than a thousand validation studies, illustrating how interview data compounds into a defensible science advantage.HireVue
• Zoom invested in BrightHire in 2021 and acquired the company outright in November 2025, bringing its interview intelligence directly into Zoom Workplace.CB Insights
• Employ acquired Pillar in March 2025 and cites it enabling recruiters to save 40 hours per month, reduce time to fill by 26%, and cut year-one attrition by 32%.Employ
• HireVue acquired the technology behind Hireguide in March 2026 to accelerate agentic, skills-based AI hiring.GlobeNewswire
• Metaview raised a $35 million Series B led by Google Ventures in June 2025, bringing total funding to $50 million as it expanded beyond notetaking into a multi-product hiring platform.FortuneMetaview

Key takeaways

• The 2025 funding wave (Juicebox $30M from Sequoia, Loxo $115M from Tritium, Findem $51M, ConverzAI $16M, Tezi $9M, Jack and Jill $20M, Dex $5.3M) shows capital flowing to many independent sourcing vendors at once rather than one winner, which keeps the category competitive and gives buyers genuine negotiating leverage.TechCrunchDHRMapFindem
• Sourcing is the task recruiters and hiring managers most often say AI helps with (58% in LinkedIn's 2018 global survey, narrowly ahead of screening at 56%), but the uplift figures vendors repeat for intent-based search over Boolean strings trace to unsourced blog posts rather than a published benchmark, so the switch should be proven on a buyer's own live requisitions.LinkedIn Talent Solutions
• LinkedIn sets the baseline every standalone vendor must beat: its Hiring Assistant agent went GA in September 2025 and cut profiles reviewed by 62% to 81%, lifted InMail acceptance 69%, and its agentic talent products are on track for about $450M in annual sales, so buyers should assume the native capability keeps advancing within any contract term.PinStaffing Industry AnalystsLinkedIn
• Sourcing, CRM, and ATS are converging into single platforms (hireEZ, Gem, Loxo), and the data moat of seeing the full candidate record (past applicants, silver medalists, interview history) gives converged AI a context advantage that standalone public-index tools cannot match.GemhireEZLoxoGem
• Autonomous sourcing agents (ConverzAI, Jack and Jill, Dex, and Tezi's Max until Headway acquired Tezi in early 2026) are a distinct product category that performs hiring rather than equipping a recruiter, but verifiable revenue is modest (around $1.8M each at Tezi and Dex), so the right use today is a contained pilot measuring candidate experience and quality, not wholesale adoption.The AI InsiderPR NewswireFortuneTezi
• Passive candidates do not optimize profiles for keywords, so literal matching misses the people sourcing exists to find, and rigid criteria fail on the other side of the market too: Harvard Business School estimates more than 27 million US hidden workers whom automated hiring tools and job criteria screen out, and 88% of the employers it surveyed agreed qualified high-skill candidates are vetted out for not matching a job description's exact criteria.Harvard Business School and Accenture
• The market has split into two pricing tiers with little overlap: an accessible, published tier (Juicebox $99 to $179 per seat, Fetcher from $379/mo, SeekOut self-serve Recruit Core at $149/mo) and a negotiated enterprise tier running roughly $15,000 to $90,000+ per year, where SeekOut contracts vary several-fold and competitive pressure plus data-portability terms drive the real outcome.JuiceboxTestGorillaVendr
• The decisive procurement principle is matching the tool to team structure and hiring type, not demo polish: LinkedIn plus a specialist for in-house generalist teams, SeekOut for deep technical/diverse/cleared talent, Findem for skills- and trajectory-based hiring, Juicebox or AmazingHiring for hidden technical talent, and Loxo as the converged system of record for multi-client agencies.SeekOutFindemJuiceboxRecruiter DailyLoxo

Key takeaways

• AI adoption in recruiting is broad but shallow: organizational AI use in HR jumped from 26% to 43% in a year (a roughly 65% rise) and 69% of HR professionals now use it for recruiting, yet only 18% of companies use AI broadly across hiring and just 37% of TA pros use generative AI daily. The competitive frontier has shifted from whether you use AI to how deeply it is integrated into actual decision points.SHRMSelectSoftware Reviews
• The marquee paradox of 2026 is that the headline hiring metrics did not fall the way an efficiency story predicts: executive cost-per-hire rose sharply while nonexecutive cost stayed flat and nonexecutive time-to-fill actually fell. The cause is system-level (a cautious low-hire/low-fire market that slows senior hiring deliberately, plus AI-inflated application volume), not tool failure, and it is masked by the fact that 56% of organizations never formally measure their AI investments and only 16% use a custom ROI metric.SHRMSHRMSHRM
• The strongest evidence supports a combined model, not full automation: organizations that use AI to screen and humans to make the final call cut time-to-hire by 40% while improving first-year retention by 25%, and recruiter judgment overrides AI in 58% of organizations. The healthy design treats AI screening as a filter that ranks and surfaces, never an auto-rejecting decider.SelectSoftware Reviews
• Appetite for autonomous agents is now mainstream (52% of talent leaders plan to add them in 2026), but readiness lags badly: only 22% believe their leaders can effectively manage mixed human-AI teams. In 2026 the delivered reality is augmentation, and agents will work first only in high-volume, well-bounded contexts like frontline and hourly hiring.Korn Ferry
• When cheap intelligence commoditizes the mechanical layer of recruiting, value migrates to judgment, relationships, and accountability rather than disappearing. SHRM finds organizations are 5.7x more likely to shift job responsibilities than displace jobs (only 7% report displacement, 39% report shifts), and the top-ranked recruiting skill for 2026 is critical thinking (73%), with AI skills only fifth. Leaders should redeploy recruiters into higher-value work, not cut them.SHRMKorn Ferry
• AI has restructured the vendor landscape: total HR tech funding cooled to roughly $3.7B in 2025 (down 40% YoY) yet talent acquisition led all categories at about $1B across 197 deals, while suite vendors absorbed AI-native point solutions (Workday bought Paradox for ~$1B, plus SAP/SmartRecruiters and Zoom/BrightHire) and AI-native challengers like Mercor (~$10B valuation, a gross annualized run rate that reached about $2B by mid-2026) emerged to compete with the recruiting function itself.Crunchbase NewsVenero CapitalGlobal Legal ChronicleTechCrunchTechCrunch
• Hiring has become an adversarial-trust problem, not just a matching problem. Gartner projects one in four candidate profiles will be fake by 2028, 59% of hiring managers suspect AI misrepresentation, 91% of recruiters have spotted candidate deception of some kind, and 41% of security and fraud leaders at large organizations say their company has hired and onboarded a fraudulent candidate. Identity verification and liveness checks must be designed into the process, especially for remote roles with sensitive access.GartnerCheckrGreenhouseGetReal Security
• A governance gap could turn AI advantage into liability: only 49% of organizations have an AI use policy and just 25% call it clear and future-proof, 57% of HR pros in regulated jurisdictions are unaware regulations exist (only 12% of the aware are compliant), and 52% of organizations exclude HR from AI strategy. Leaders should give HR ownership of a durable evaluation framework and shift the scorecard toward quality of hire (the 25% retention-lift metric AI demonstrably improves), which 61% of TA pros believe AI can help measure.SHRMSelectSoftware ReviewsRecruit AI Suite

Key takeaways

• The talent technology sector recorded between 90 and 100 acquisitions every quarter through 2025, with 110 in the third quarter alone.Venero Capital
• By one widely cited measure iCIMS holds the number one ATS market-share position at roughly 11%, followed by Oracle, Workday, and Greenhouse in the top tier.iCIMS
• Workday acquired the conversational hiring platform Paradox for approximately one billion dollars in 2025.Global Legal Chronicle
• Paradox's assistant Olivia serves more than 500 companies and reports cutting time-to-hire in half for high-volume employers.Paradox
• Mercor raised a $350 million Series C in October 2025 at a $10 billion valuation, a fivefold increase in eight months, on the back of a gross annualized run rate above $450 million that then quadrupled to about $2 billion by mid-2026 (lower net of contractor payouts).TechCrunchTechCrunchTechCrunch
• Total HR technology investment reached roughly $3.7 billion in 2025, a 40% decrease year over year.HRO Today (HR Tech Report)
• Talent acquisition led all HR-tech categories in 2025 with about one billion dollars raised across 197 deals.HR Executive
• Gartner projects that one in four candidate profiles worldwide will be fake by 2028, and in a September 2025 GetReal Security survey 41% of IT, security, risk and fraud leaders at organizations with 1,000+ employees said their company had hired and onboarded a fraudulent candidate.GartnerGetReal Security

Key takeaways

• The value in the talent-marketplace economy has inverted: an AI-data-labor company, Mercor, is worth 10 billion dollars while Upwork, the largest public freelance platform, books only 788 million dollars of annual revenue. The growth has moved from platforms that move commodity labor to those that move scarce, AI-relevant human cognition.TechCrunchU.S. Securities and Exchange Commission
• Classic freelance has split into two categories: a commoditizing base under direct AI substitution pressure (Fiverr's active buyers fell 13.6 percent even as revenue grew) and a defensible premium layer. Value Upwork and Fiverr as mature cash machines, not growth stories, and never apply the category CAGR to the incumbents.Fiverr International Ltd.Yahoo FinanceGrand View Research
• Take rates are diverging rather than converging, spanning from Braintrust's zero on the talent side to Toptal's 30-plus percent premium, with the EOR category settling on a flat fee around 599 dollars per worker per month. The take-rate spread is a map of moats: community alignment, curation, compliance, and AI-native matching of scarce cognition.BraintrustGetLatkaWhichPayroll
• Employer of record is the most defensible of the four categories thanks to its compliance-and-entity moat, with Deel surpassing 1 billion dollars in ARR at a 17.3 billion dollar valuation while processing 22 billion dollars in payroll across 150-plus countries. Buyers placing multi-year compliance bets must now price in governance risk, as the category leader is under a DOJ criminal probe over alleged corporate espionage against Rippling.TechCrunchDeelPYMNTS
• AI data labor is the fastest-growing and highest-valued marketplace category because when intelligence becomes cheap the bottleneck shifts to the human judgment that trains and aligns it. AI-native matching lets these platforms scale to hundreds of millions in payout with few employees, with Mercor paying over 1.5 million dollars a day to roughly 30,000 contractors for buyers like OpenAI and Anthropic.CNBCWikipedia
• On a near-29-percent CAGR, AI data labor crosses over the freelance and EOR categories in absolute size within the forecast window, moving from the smallest of the four in 2025 toward the largest by 2028. Any sizing that files data labor under AI infrastructure rather than talent misses where the marketplace economy is heading.Coherent Market InsightsBusiness Research Insights
• AI data labor is the most fragile category despite its growth, exposed to three concentrated risks: a shift toward synthetic data, demand concentration in a few frontier labs that may internalize the work (as Meta's Scale AI stake hints), and labor-and-regulatory risk, including a US Department of Labor FLSA investigation into Scale AI and lawsuits over underpayment and worker conditions.The RegisterWikipedia
• The actionable rule for every audience is to stop treating this as one gig economy: match marketplace to job, capital to moat, and take rate to defensibility. Buyers should use the specialized platform built for each job, investors should buy the durable EOR moat at a fair price and the fragile data-labor hypergrowth only at a price that respects its variance, and operators must pick a take-rate strategy that matches their achievable moat.TechCrunchMarketing AI Institute

Key takeaways

• Hiring a data professional now takes 19.5 interviews on average, more than double the 9.5 it takes to hire a customer support agent.Ashby
• Data roles consume 24.9 interviewer-hours per hire and engineering 24.7, against just 8.9 hours for customer support.Ashby
• Technical roles take roughly 15 days longer to fill than business roles, and senior roles take 37% longer to fill than junior positions.Ashby
• Nearly 40% of senior-level roles take more than 90 days to fill, while most entry-level roles close inside 30 to 60 days.Mitratech
• SHRM publishes no all-roles benchmark: its 2026 data splits into a 39-day median time to fill for nonexecutive roles (down from 44 in 2025) against 45 days for executives, and a $1,300 median cost per hire for nonexecutive roles against $15,000 for executives.SHRM
• Applications per hire have tripled since 2021 to roughly 291 in Q1 2026, up from about 100 in early 2021.Ashby
• LinkedIn now processes an average of 11,000 applications per minute, a 45% jump in a single year.eWeek
• B2B sales teams run average annual turnover around 35%, with SDRs the highest at roughly 45% and account executives near 30%.Optifai