RESEARCHYou Can't Buy Taste. We Tested It.

Invisible Technologies vs Mercor: evidence or scale

Invisible Technologies and Mercor are both described as Scale AI challengers, which obscures how differently they are built. Invisible is a managed services company that publishes dated audit documents, named case studies with metrics, and co-authored research. Mercor is a marketplace that has raised far more money, grown far faster, and publishes almost none of that. One of them gives you more to verify before signing; the other gives you more people, sooner. This comparison works through what each operates, what each will show you, and which trade fits which program.

Key takeaways

  • Invisible delivers managed outcomes and is accountable for them. Mercor sources and matches experts, bills their hours at a markup, and you generally manage the work.
  • Invisible publishes the most substantial trust documentation in this category: a dated SOC 2 Type 2, HIPAA policies, Cyber Essentials Plus, and six named penetration-test reports.
  • Their revenue figures are not comparable. Mercor's 2 billion dollars is gross marketplace volume before contractor payouts. Invisible's 134 million is estimated company revenue.
  • Invisible covers substantially more modalities, including video, audio, and 3D. Mercor is text and code, plus wearable-camera video through partner businesses.
  • Invisible publishes named case studies with metrics. Mercor publishes none, only anonymized outcomes.
  • Mercor's expert network, capital, and growth rate are all larger. If speed to staff is the constraint, that matters more than documentation.

About Invisible Technologies

Founded in 2015 by Francis Pedraza, remote-first with New York listed by most trackers. Matt Fitzpatrick, previously a McKinsey AI leader, became chief executive in January 2025, with Pedraza moving to chairman. In September 2025 the company raised 100 million dollars led by Vanara Capital at a 2 billion dollar valuation, and it describes itself as having run profitably for half a decade before that round.

The platform is organized into five modules: Neuron for data, Atomic for process building, Meridial for the expert network, Synapse for evaluations, and Axon for agents. Its AI-training business covers six lines: expert domain training, continuous model evaluation across step-based agentic workflows, internationalization across more than 80 languages, reinforcement learning environments, multimodal data generation and labeling spanning 3D modeling, video annotation, and audio engineering, and red-teaming.

Workforce figures vary by source and definition. Invisible states 24,000 or more vetted experts and 250-plus full-time staff; outside analysis puts the contractor network at 3,000 or more across 35-plus countries. In March 2026 it agreed to acquire WeCP, a technical assessment platform bringing more than 18,000 scope-specific assessments and over two million interview records, explicitly to raise the precision of expert evaluation.

About Mercor

Founded in January 2023 by Brendan Foody, Adarsh Hiremath, and Surya Midha, based in San Francisco. It raised a 100 million dollar Series B at 2 billion in February 2025 and a 350 million dollar Series C led by Felicis at 10 billion in October 2025. A further round at roughly 20 billion has been reported in talks since July 2026 and has not closed.

Candidates complete an AI-run interview of about twenty minutes, with up to three retakes, and qualified experts are matched to client projects with Mercor billing a marked-up hourly rate. Mercor states 30,000 or more active experts and an average contracted rate of 107 dollars per hour, with published listings running from around 60 dollars per hour to 450 for clinical specialists. Its documentation lists 107 supported payout countries.

Beyond staffing it sells data monetization, licensing enterprises' own operational data to AI labs after extraction and de-identification, off-the-shelf data, and a robotics line licensing wearable-camera footage captured at partner businesses. It publishes the APEX benchmark family with partner benchmarks built alongside Harvey, Ramp, Cognition, and Sierra, and acquired Sepal AI in February 2026 and Deeptune in July 2026.

What buyers report

Neither has a usable review base, and both are thin for different reasons.

Invisible holds 5.0 out of 5 on G2 from two reviews, which is a sample too small to report as a rating. It has no Gartner Peer Insights or Capterra listing. Its Glassdoor rating sits near 3.1 from roughly 290 reviews, but those are employees rather than customers.

Mercor holds 4.4 out of 5 on G2 across 114 reviews, but the profile sits under applicant tracking and AI recruiting categories, so the reviewers are candidates and contractors. It has no Gartner Peer Insights listing.

The difference is that Invisible compensates with published case studies, covered below. Mercor does not.

The comparison

Invisible TechnologiesMercor
ModelManaged delivery, accountable for outcomesMarketplace, hours billed at a markup
Founded2015January 2023
FundingRoughly 144 million raised; 2 billion valuation, September 2025Roughly 484 million raised; 10 billion priced, October 2025
Revenue134 million dollars estimated company revenue, 20242 billion dollars gross marketplace volume, June 2026
ModalitiesText, code, image, video, audio, 3D, multimodalText and code, plus wearable-camera video
Languages80-plusNot published as a count
Network24,000-plus vetted experts stated; 3,000-plus contractors across 35-plus countries30,000-plus active experts; 107 payout countries
ComplianceSOC 2 Type 2, HIPAA policies, GDPR processor, Cyber Essentials Plus, CMMC badgeSOC 2 Type 2 only
Security evidenceSix dated penetration-test reports, 16 named policiesPenetration testing noted, reports gated
Public sectorTradewinds awardable vendor, NASA SEWP V contracts, Carahsoft partnerNone published
Case studiesNamed, with metricsAnonymized only
IP indemnificationNot published; no public terms of serviceNot published; public terms indemnify Mercor only

The revenue comparison that everyone gets wrong

Put 2 billion next to 134 million and Mercor looks roughly fifteen times the company. That is not what the numbers say.

Mercor's 2 billion is gross annualized revenue, meaning total customer spend before contractor payouts, with contractor share estimated at 60 to 70 percent. Mercor does not publish net, and outside estimates of it span 180 million to 800 million. Invisible's 134 million for 2024 is estimated company revenue on a managed-services profit and loss, with reported EBITDA of 15 million, and it is not split gross and net because the model does not work that way.

The honest comparison is Mercor's net against Invisible's revenue, and on that basis the gap narrows sharply and may not exist at all at the low end of the estimate range. Both figures are third-party estimates rather than disclosure. Mercor is unambiguously growing faster and has raised more than three times the capital. But if vendor scale is a procurement criterion, use the funding record and the customer evidence, both documented, rather than revenue headlines that are measuring different things.

Compliance and security evidence, where Invisible is clearly ahead

This is the sharpest difference between them and the easiest to verify yourself.

Invisible operates a trust portal with document-level detail: a SOC 2 Type 2 report last updated February 2026, HIPAA compliance and workstation security policies, a GDPR data processor badge, a Cyber Essentials Plus certificate updated June 2026, and a DoD CMMC badge. Unusually, it also publishes six named penetration-test reports covering annotations, the marketplace, Neuron, Atomic, and virtual desktop infrastructure, dated between October and December 2025, alongside sixteen named security policies with last-updated dates. On the public-sector side it is a Tradewinds Solutions Marketplace awardable vendor, a Carahsoft authorized partner, and holds two NASA SEWP V contracts.

Mercor operates a trust center listing a SOC 2 Type 2 report, penetration testing, business continuity plans, cyber insurance, and its subprocessors. ISO 27001 appears in its job postings as a framework being stood up rather than achieved. HIPAA and GDPR attestations are absent.

Two caveats keep this honest. Invisible's PCI DSS document on that portal was last updated in 2018 and is stale. And ISO 27001 does not appear on Invisible's own portal despite a third-party case study claiming it, so treat that claim as unverified. Neither company publishes IP indemnification or data-ownership terms, and Invisible's privacy center explicitly directs those questions to customer agreements that are not public.

Evidence of delivery

Invisible publishes named case studies with numbers attached. Cohere evaluated Command A using PhD-level experts and expanded into ten languages, with reported figures including a 51.7 percent average win rate, 91.5 on IFEval, and 92 percent on RepoQA, alongside a named quote from Cohere's director of data and evaluations. Nasdaq is credited with saving 10,000 developer hours. Boosted.ai is credited with 90 percent lower cost. Named logos include Microsoft, AWS, Thomson Reuters, ElevenLabs, and SAIC.

All of these are vendor-published and unaudited, and the company's headline claim that it has trained more than 80 percent of the world's leading AI models is self-reported and not verifiable. But named customers with specific metrics and an attributed quote are a materially stronger artifact than what most of this category offers.

Mercor publishes no named enterprise case studies. Its enterprise page carries outcome metrics with every customer anonymized, and its stories section profiles contractors rather than clients. Its named relationships come through benchmark partnerships with Harvey, Ramp, Cognition, and Sierra, and through third-party reporting naming OpenAI, Anthropic, Meta, and Google as customers, plus Nvidia's own disclosure of using Mercor data in recent Nemotron models.

Research

Invisible publishes co-authored work with named institutional partners: a text-to-image evaluation paper with Meta on arXiv, a paper on data quality thresholds and emergent misalignment, and a defense simulation whitepaper with SAIC. It has no Hugging Face or GitHub organization and publishes no open datasets.

Mercor publishes the APEX benchmark family and partner benchmarks. Neither company has an open data footprint comparable to the larger vendors in this category.

Which to choose

Choose Invisible when you need multimodal work beyond text, particularly video, audio, or 3D; when your security review wants documents rather than assurances; when public-sector contract vehicles matter; or when you want a vendor contractually accountable for an outcome rather than for supplying hours. Ask about the stale PCI document and get IP terms drafted explicitly.

Choose Mercor when speed to staff is the binding constraint, when you need coverage across many countries, when the work is text and reasoning in professional domains, or when you want the adjacent lines Invisible does not offer, including data monetization and wearable capture. Ask what certifications exist beyond SOC 2 Type 2.

Why HumanSignal is the better choice when the data has to be created

Both companies here work on material that already exists or on judgment applied to it. Invisible will annotate footage you bring and says so plainly; its computer vision offering begins with the phrase bring us your footage. Mercor's wearable line is the closest either comes to producing something new, and it is opportunistic capture at partner businesses rather than a protocol designed to close a gap in your model.

HumanSignal Services starts from the gap instead. We write the collection protocol, recruit for the physical or professional characteristics it requires, run the capture in our studios, on site at your facility, or in the field with synchronized motion, vision, audio, and sensor streams, and deliver a dataset verified against the spec. The difference from opportunistic capture is control over the distribution: which tasks, which conditions, which failure modes, how many repetitions, and which edge cases are deliberately included rather than hoped for.

The quality mechanism is inspectable rather than asserted. Review workflows run in Label Studio Enterprise, the commercial edition of the open source labeling tool we build, so inter-annotator agreement, reviewer decisions, and rework rates are reportable to you while the work is running. We publish contributor pay ranges, which makes our sourcing claims checkable. Our customer agreement gives you exclusive ownership of customer data and of insights produced specifically for you, which neither company here publishes. We hold SOC 2 Type 2 and HIPAA compliance, documented on our security page.

The capture operation came in through the acquisition of Erud AI in November 2025. If your constraint is designed capture rather than annotation or expert hours, physical AI data services is where to start.

Which company is bigger

The public figures do not settle it, because they measure different things. Mercor reports 2 billion dollars in gross marketplace volume before contractor payouts; Invisible is estimated at 134 million in company revenue for 2024 with 15 million EBITDA. Mercor has raised roughly three times the capital and is growing faster. On revenue actually retained, the gap is far smaller than the headlines suggest.

Does Invisible do data collection or only annotation

Primarily annotation and generation on material you supply. Its computer vision offering is framed around customer-supplied footage, and no first-party physical collection capability is published. Its one physical-domain research contribution is a simulation whitepaper with SAIC rather than sensor collection.

Which has better security documentation

Invisible, by a clear margin. It publishes a dated SOC 2 Type 2, HIPAA policies, Cyber Essentials Plus, a CMMC badge, six named penetration-test reports, and sixteen security policies with dates. Mercor publishes a SOC 2 Type 2 behind an access request and little else. Note that Invisible's PCI DSS document dates to 2018 and that ISO 27001 does not appear on its own portal.

Can either work with government agencies

Invisible has published contract vehicles: Tradewinds awardable vendor status, NASA SEWP V contracts, and a Carahsoft partnership. It publishes no FedRAMP authorization. Mercor publishes no public-sector vehicles at all. For work requiring FedRAMP or an IL authorization, neither qualifies on published evidence.

Related Content