About Cortext

The bottleneck in AI is no longer compute.
It is expert judgment.

Cortext Labs, Inc. runs the marketplace that connects credentialed professionals to the labs training frontier models. 42,000+ experts in 130 countries, 40+ lab and research partners, paid every Friday in US dollars.

$18.4M
Paid to experts in 2025
42,000+
Active experts
$68
Average hourly rate
130
Countries
Why this exists

Models ran out of internet before they ran out of appetite

Pretraining consumed the readily available text. What decides how good a model is now happens after that, and it needs people who actually know the answer.

A model that has read everything still has no way of knowing which of two plausible answers a specialist would accept. That judgment has to be supplied from outside, by someone qualified to make it, and it has to be supplied in a form the training process can use — a preference between two responses, a score against a written rubric, a corrected reference answer, an adversarial prompt that breaks a safeguard.

This is post-training, and it covers a family of methods: supervised fine-tuning on expert-written demonstrations, reward models trained on ranked comparisons and then optimised against with reinforcement learning, and direct preference optimisation, which skips the separate reward model and tunes on the preference pairs themselves. All of them share one input. Someone has to decide which output is better, and be right.

For general-knowledge tasks a careful generalist is enough. For a differential diagnosis, a securities filing, a proof, or a production incident, they are not. The gap between a crowdworker's label and a specialist's label on the same item is where model quality now moves, and it is why a board-certified cardiologist grading clinical reasoning commands a rate an order of magnitude above generalist annotation.

What experts actually produce
Preference pairs
Two model responses, ranked, with the reasoning recorded.
Rubric scores
Multi-criterion grading against a written standard.
Gold responses
The reference answer a specialist would have given.
Adversarial probes
Prompts built to break a safeguard, plus the failure.
Trajectory reviews
Step-by-step judgment of an agent's tool use.
The business

A two-sided market with a quality problem in the middle

Labs do not want a labour pool. They want a delivered dataset that holds up under audit. Everything between those two things is what Cortext is.

01

Scoping with the lab

A research team arrives with a capability they want to improve and rarely with a rubric. We turn that into a written specification: what an item is, what counts as a correct one, which credential is required to judge it, and what agreement rate the delivered set has to clear.
02

Matching from the network

Roles are routed to experts whose verified credentials, assessment scores and prior quality history fit the specification. Contributors see only what they are eligible for, which is why the board looks different depending on who is signed in.
03

Delivery under review

Every submission is scored. Gold items seeded into the queue measure whether a contributor is still calibrated, agreement between reviewers is tracked per project, and nothing ships to a lab until the set clears its target. Contributors are paid for reviewed work in the Friday, 12:00 UTC run.
What we are not
Cortext is not a staffing agency and does not place people into jobs at the labs. It is not a crowdsourcing platform — there is no open task pool anyone can start working from. And it is not an employer: contributors are independent contractors who set their own hours, use their own equipment, and are free to work for our competitors.
Admissions

6.8% of applicants join the network

The funnel is deliberately narrow, and every stage past the first is paid. Here is what each one filters for and what it costs to get through it.

Applications received
214,300 in 2025, across every domain and level.
Credential verification
61% clear this stage. We check licences against issuing bodies, degrees against registries, and employment against public record. Unverifiable claims are not automatically rejected, but they cannot be used to qualify for a credentialed role.
Domain assessment
60–90 minutes, open-book, scored against the same rubric contributors will later apply. Pass mark is 80%; 34% of those who sit it pass. Up to 3 attempts, with 30 days between them.
Structured interview
A 25-minute recorded interview covering reasoning, not trivia. It exists mainly to catch credential fraud and to check that the person who sat the assessment is the person who applied.
Paid work trial
3–5 hours of real project work at the listing's rate. 71% pass. One trial per person — a failed trial does not bar you from applying to a different domain, but it cannot be retaken for the same one.
Admitted to the network
6.8% of applicants overall. Median time from application to first paid task is 17 days.
Rejection is domain-specific, not permanent
Failing a clinical assessment says nothing about whether you can do agentic evaluation. Applications are scored per domain, and a rejection in one leaves every other one open. The only hard stop is credential fraud, which ends the account.
How we operate

Six positions, and what each one costs us

Operating principles are only worth publishing if they constrain something. Each of these does.

Rates are published before you apply

Every listing shows its band on the card, before an account exists and before an application is submitted. It costs us negotiating room on the low end and it means contributors self-select out of work that is not worth their time — which is the point. We do not run reverse auctions and we do not ask what you currently earn.

Unpaid screening is capped at 30 minutes

Anything longer is a work trial and work trials are paid at the listing's rate, including the ones that end in a rejection. Our domain assessments run 60–90 minutes, which is over that line, so they are paid too. The rule costs real money on a funnel that saw 214,300 applications last year.

Rejected work gets a written reason

If a submission is scored down or an hour is removed, the reviewer records why, against the specific rubric clause, and the contributor sees it. Vague quality scores are how marketplaces quietly stop paying people. A reason you can read is a reason you can appeal.

A human decides anything that ends work

Automated checks flag drift, duplicate accounts and machine-generated submissions, and they are wrong often enough that they are not allowed to be the last step. Suspensions, offboardings and removed hours are reviewed by a person before they take effect, and the appeal goes to someone who was not part of the original decision.

No exclusivity, no non-competes

Contributors are independent contractors and we contract like it. Work for our competitors on the same afternoon if you want to. The trade-off is that we cannot guarantee volume — projects start and end on the labs' timelines, and there will be weeks where nothing matching your profile is open.

Rates are set by scarcity, not by your postcode

A role pays the same band whether the person doing it is in Lagos, Lisbon or Los Angeles. Marketplaces in this sector routinely price the same task three or four times lower based on where the contributor lives; we price on the credential and the difficulty of finding someone who holds it. Low-resource languages carry a premium for the same reason.
Leadership

Who runs Cortext

The board seats five: two founders, one seat each for Halyard Partners and Meridian Growth Partners, and one independent director.

AO
Amara Okonjo-Reyes
Co-founder & Chief Executive Officer

Ran the human-data organisation for a speech and dialogue research group before starting Cortext, where the recurring problem was that annotation vendors could supply volume but not judgement. Spent four years before that in clinical operations at a hospital network, which is where the conviction that credentialed reviewers produce categorically different labels came from. Holds a PhD in computational linguistics and still personally reviews the calibration report for every clinical project Cortext takes on.

DW
Daniel Wei
Co-founder & Chief Technology Officer

Built and maintained the reward-model training pipeline for a mid-sized lab, including the tooling that flagged annotator drift mid-run. Owns Cortext's matching system, the task-routing service and the review infrastructure that scores every submitted item. Writes the quarterly note explaining, in public, which internal quality metrics moved and which did not.

PR
Priya Raghunathan
Chief Operating Officer

Scaled a 9,000-person distributed operations function across eleven countries before joining Cortext in 2025. Responsible for project delivery, the regional hubs and the escalation path when a lab's timeline and an expert's availability disagree. Introduced the rule that no project ships to a lab until its rubric has cleared a human calibration set, which added roughly nine days to first delivery and cut post-delivery rework by a third.

TB
Tomás Berger
Chief Financial Officer

Twelve years in finance operations at payments and marketplace companies, most recently running treasury for a cross-border payouts business covering 90 countries. At Cortext, owns the weekly payout run, the contractor tax reporting stack and the decision to absorb foreign-exchange spread rather than pass it to contributors. Reports payout failure rates to the board every month alongside revenue.

NA
Ngozi Fola Adeyemi
Vice President, Expert Network

Joined from a professional licensing body, where the day job was verifying credentials at scale and adjudicating disputed ones. Runs recruiting, credential verification, the assessment bank and expert support. Wrote the thirty-minute ceiling on unpaid assessments into policy after auditing how long applicants were actually spending before anyone paid them anything.

HL
Hannah Lindqvist
Chief Quality Officer

Spent nine years as a research methodologist designing inter-rater reliability studies for medical and legal evidence review, and published on adjudication protocols for low-agreement domains. At Cortext, owns rubric standards, gold-set construction, reviewer certification and the appeals process for removed hours. The only executive with authority to halt a project mid-delivery on quality grounds.

MO
Marcus Oyelaran
General Counsel

Practised employment and commercial law for eleven years, including four advising platform companies on contractor classification across US, UK, EU and Indian regimes. Owns the contractor agreement, the confidentiality and IP terms labs require, data protection and the sanctions screening that determines which countries Cortext can pay. Maintains the position that Cortext does not ask contributors for non-compete or exclusivity terms.

Backing

$147.2M raised across three rounds

Cortext is venture-funded. Contributors are paid from revenue, not from the balance sheet — payouts do not depend on the next round closing.

SeedMarch 2024
$6.2M

Raised three months after incorporation, against a network of 400 experts and two pilot contracts.

Led by Fernwood Capital. Rho Street Ventures participating.
Series AFebruary 2025
$31M

Funded the credential-verification stack, the Nairobi and Bengaluru hubs, and the move from flat annotation work to rubric authoring.

Led by Halyard Partners. Fernwood Capital, Anselm Ventures participating.
Series BApril 2026
$110M

Post-money valuation of $1.4B. Earmarked for clinical and legal expert recruitment, the agentic-trajectory tooling and a second payout rail.

Led by Meridian Growth Partners. Halyard Partners, Fernwood Capital, Kanto Fund participating.
Where we are

One headquarters, four regional hubs, and a network everywhere else

The hubs exist to recruit and support experts in their regions and to keep working hours overlapping with the labs we deliver to. Contributors are never required to be near one.

Headquarters
San Francisco
1 Sansome Street, Suite 3500
San Francisco, CA 94104
United States
Research operations
New York
United States
EMEA expert network
London
United Kingdom
Africa expert network
Nairobi
Kenya
APAC expert network
Bengaluru
India
Press

Media enquiries and company boilerplate

Journalists on deadline should write to press@cortext-ai.uk — we answer press mail ahead of everything else.

Recent coverage
The Economist12 May 2026
The people teaching machines to be experts
Feature on the shift from crowdsourced annotation to credentialed review.
TechCrunch14 April 2026
Cortext raises $110M to put licensed professionals in the training loop
Series B announcement.
Financial Times3 March 2026
AI labs are bidding up the price of specialist judgement
Quotes Cortext on clinical and legal rate inflation through 2025.
Business Daily9 February 2026
Nairobi's doctors and lawyers pick up a second shift grading AI
Reporting on the East Africa expert network and local rate setting.
Wired21 January 2026
Inside the market for expert disagreement
On inter-annotator agreement and why labs pay for adjudication.
Boilerplate

Cortext Labs, Inc. operates an expert marketplace for AI training data. Domain professionals — physicians, attorneys, engineers, quantitative analysts, linguists and research scientists — evaluate model output, author reference answers, build and apply grading rubrics, red-team frontier systems and annotate agent trajectories for the labs building large language models. Founded in 2024 and headquartered in San Francisco, Cortext works with more than 40 AI labs and research groups and pays its contributor network weekly in US dollars.

Quote this verbatim. For logos, product screenshots and executive headshots, write to press@cortext-ai.uk and ask for the brand kit.
Other enquiries
Working at Cortextcareers@cortext-ai.uk
Legal and policylegal@cortext-ai.uk
Security disclosuressecurity@cortext-ai.uk
Cortext has no phone line. Every channel above is email, which keeps a written record for both sides across 130 countries and every timezone.
A note on this site
Cortext Labs, Inc. is a demonstration build. The company, its leadership, funding history, press coverage and figures on this page are invented, and the openings on the careers page are illustrative. The product flows are real and the policies are written as they would be for a live marketplace.

Your expertise is worth more.

Apply once, get matched to projects that fit your credentials, and cash out every Friday.

Find your role →How it works