Juratum
Domain experts, measured and developed
The expert network for AI data work · London
We measure what an expert knows and how they work, and we invest in making both sharper.
Juratum organises lawyers, clinicians, engineers and other domain experts for the data work frontier AI depends on: grading answers, writing the tests a model must pass, catching what it gets wrong. Whether that work is any good comes down to two things, and we measure both: real domain competency, and the working habits that produce consistent, high-quality judgement. Then we develop them, so a member’s work gets better, and more valuable, over time.
Membership · open now
Anyone can sign up, with the details you would give for any job application. Then sit a friendly, structured interview with Juratum’s AI, and from there your profile is yours to build: test yourself, gain Juratum qualifications, and the work offered to you improves with it. What to expect is set out plainly on the joining page.
Reviewing, grading and correcting model-written code: the largest demand in AI data work.
Sign upWriting and marking the reasoning problems frontier models are trained and tested on.
Sign upLegal judgement for frontier and legal AI, across the practice areas.
Sign upGrading and guiding medical AI against real practice.
Sign upGrading financial analysis, modelling and judgement from AI against how the work is really done.
Sign upWhat we measure
What they know. Domain-specific, empirical tests with known answers measure a member’s judgement where it counts: real work, marked against ground truth, discipline by discipline. A plausible-but-wrong answer is exactly what AI needs an expert to catch, so that is exactly what we test for.
How they work. A psychometric and behavioural battery measures the habits that separate reliable expert data work from the rest: consistency, calibration, how someone reasons under ambiguity, whether they flag what they are unsure of. These are not pass-or-fail gates. They show each member precisely what to sharpen: the raw material of getting better.
We invest in our members
Most talent networks test you once and rank you. On Juratum your profile is yours to build: sit further tests whenever you choose, gain new qualifications, and watch the work you are offered improve with it. Members who keep testing move into higher-quality, higher-value work, and developing them is how the whole network’s judgement compounds. It is why the best experts stay.
How Juratum works
Create your profile with the same details you would give for a job application or contract work. We verify proactively: your identity, and where your profession has a register, your registration against it.
A friendly, structured interview with Juratum’s AI. We want to know what you practise, where your judgement is strongest, and what work suits you.
Sit domain tests and gain Juratum qualifications whenever you choose. Each one strengthens your profile, and a stronger profile is offered stronger work. Nobody ranks you once and files you away.
A client brings a project and we assemble a balanced group of experts around it, working together to produce great work. Everyone signs the Juratum Oath first: a commitment by all parties, client and professionals alike, to work with each other passionately and professionally.
Solicitors, barristers and paralegals across contract, litigation, employment, property, regulatory and public law. Every authority cited gets checked against the primary source, because a plausible fake citation is precisely the failure legal AI needs humans for.
Registered clinicians who grade medical AI the way they practise. NHS pathways, UK-licensed medicines, safety judgement first.
Evaluation benches assembled to the customer’s security requirements. We are building this vertical with defence buyers rather than guessing at what they need.
The instruments
The competency tests and habit measures that admit and develop the network were designed by HCPC-registered, chartered occupational psychologists: careers spent selecting for defence, for national health and for boardrooms, where a wrong judgement about a person is the most expensive mistake there is.
Juratum’s engineers build what AI newly makes possible in reading expert performance at scale.
The Standard
How members are verified, measured, developed and quality-controlled runs under the Juratum Standard: a published, versioned document covering identity, competency testing, habit measurement, agreement and independent audit. Every engagement cites the version it ran under.
No talent marketplace publishes one. We consider that the point.
The Juratum Standard, v0.1. Publication forthcoming on this page.
Where it matters · on the record
For government, defence and jurisdiction-bound work, who supplies the judgement, and where the data sits, is part of the specification. So every engagement can carry a declared sovereignty class, provable person by person from member profiles. It is a guarantee we can point to, not the whole story.
| UK‑sovereign | UK-based, UK-national professionals; UK data residency throughout. |
|---|---|
| EU‑sovereign | EU-based, EU-national professionals, for buyers under the EU AI Act. |
Measured against the Juratum Standard, version 0.1, in the City of London, this year two thousand and twenty-six.
Sign-up and enquiries: portal.juratum.com/join
