Skip to content

01Research

Ask the law of the world.

An autonomous research system that reads judgments, legislation and court records across every jurisdiction we cover, compares how similar legal systems decide, and cites the source of every sentence.

It works the way a researcher works. Only across every court at once.

It reads the law
Judgments at every level, statutes as they stood on the day, and the court record around them — in the language each was written in.
It cross-references
Where one system is silent, it reads the systems that share its tradition, and says which is which.
It checks
Every citation is matched to the record before it is used. A passage that cannot be found is not quoted.
It writes
A memo, a comparative table or a short answer — each sentence linked to the paragraph it rests on.
ResearchSample
QuestionCan a court order a government to cut its emissions faster?
  1. Reading case law across jurisdictionsWaiting
  2. Cross-referencing similar legal systemsWaiting
  3. Checking every citation against its sourceWaiting
  4. Writing the answerWaiting

In several legal systems, yes. The Supreme Court of the Netherlands upheld an order requiring the State to cut emissions by at least 25% by the end of 2020, against 1990 levels1. Germany’s Federal Constitutional Court held the Federal Climate Change Act incompatible with fundamental rights for failing to set out reductions after 20302. The European Court of Human Rights found that Switzerland had violated Article 8 of the Convention by failing to put an adequate framework in place3.

1The State of the Netherlands v. Urgenda FoundationHoge Raad · 20 Dec 2019 · ECLI:NL:HR:2019:2007
2KlimaschutzBundesverfassungsgericht · 24 Mar 2021 · 1 BvR 2656/18
3Verein KlimaSeniorinnen Schweiz and Others v. SwitzerlandECtHR Grand Chamber · 9 Apr 2024 · No. 53600/20

02Comparative law

Similar systems, read side by side. A question decided in one informs how the next will decide it.

A question from Israel is read against the United Kingdom, Canada and Australia — and against the civil-law courts Israeli law has borrowed from. A German question is read against Austria and Switzerland; a Brazilian one against Portugal and the rest of the Latin tradition.

Common law

United Kingdom, United States, Canada, Australia

New Zealand · Ireland · India · Kenya

Civil law · French tradition

France, Belgium, Italy, Spain, Portugal

Luxembourg · Netherlands · Brazil · Colombia · Costa Rica · Ecuador · Peru · Uruguay

Civil law · German tradition

Germany, Austria, Switzerland

Czechia · Slovakia · Slovenia · Hungary · Poland · Estonia · Japan · South Korea · Taiwan

Nordic law

Denmark, Finland, Iceland, Norway, Sweden

Read together, and with the European courts above them.

Mixed systems

Israel, Scotland, Quebec, the Philippines

Common law and civil law in one system, read against both.

International

The European courts

European Court of Human Rights · Court of Justice of the European Union · EPO Boards of Appeal

03Provenance

Answers from the record, not from memory.

Each sentence of an answer links to the paragraph it rests on — the court, the date, the citation and the body that publishes it. If the record does not say it, the answer does not either.

Decision20 December 2019
Hoge Raad der Nederlanden · Supreme CourtThe State of the Netherlands v. Urgenda Foundation
Citation
ECLI:NL:HR:2019:2007
Case number
19/00135
Chamber
Civil
Language
Dutch, as published

Published by the Dutch judiciary · rechtspraak.nl

04What it delivers

From a single question to a full memo.

Research memo

A structured answer to a hard question: the rule, how courts have applied it, where they disagree, and every authority cited.

Comparative table

One question, answered jurisdiction by jurisdiction, side by side — with the leading decision and statute for each.

Citation check

Every authority in a brief or an opinion matched against the record, with the passage it is cited for.

05Under the hood

Fast where it should be, thorough where it must be — and yours throughout.

Fast search
The whole record searched by words and by meaning at once — lexical and semantic retrieval together, across languages — as fast as you can type.
Vectors through the API
The corpus embedded and served as a vector index: retrieve the passages that answer a question, by meaning, each with its citation — for your own models and agents.
Task harnesses
A purpose-built harness for each task — research, comparison, citation checking, drafting, screening review — with its own tools, sources, steps and checks.
Zero data retention
Zero-data-retention agreements with the model providers we use: what you send is not stored by them, and not used to train their models.
Bring your own key
Run on your own model-provider keys, under your own agreements — and hold the encryption keys to your matters.
Private deployment
The whole system inside your network, with our engineer on site to install, adapt and train.
RequestSample
POST /v1/retrieve
{
  "query": "duty of care of the state to reduce emissions",
  "jurisdictions": ["NL", "DE", "ECHR"],
  "mode": "hybrid",
  "top_k": 8
}
Response · 200
{ "passages": [ {
    "text": "…",
    "citation": "ECLI:NL:HR:2019:2007",
    "score": 0.91
  }, … ],
  "retention": "none" }

Bring us your hardest question. We will show you how the record answers it.