Skip to main content

How we build & verify

The honest version. What caselaw is, who makes it, where every line of law comes from, how the AI-assisted study material is generated and checked, and — just as important — what we do not claim.

Written for the academics and students deciding whether this is trustworthy. If anything here reads as overclaiming, email [email protected] and we will fix it.

In one paragraph

caselaw is built by one founder in the UK. Every case page is anchored to a primary judgment from an open, official source. Briefs, the AI answer cards, and search answers are generated on demand from that judgment text and grounded in it — they are AI-assisted study aids, not a substitute for reading the case, and not legal advice. The UK corpus is broad; the Canadian corpus is currently federal-focused, with provincial appellate coverage being expanded. We say so plainly everywhere, including here.

Who builds it

caselaw is a solo project. One founder — a UK law student — does the data ingest, the engineering, the content design and the support inbox. There is no editorial board, no panel of barristers, and no firm behind the brand.

We mention this for a reason: it sets the right expectation. We do notclaim that briefs are “reviewed by barristers” or “checked by practising solicitors”, because they are not. What you get is a carefully-built study layer over primary law, with the original judgment always one click away so you can check our work. More on the person behind it on the about page.

Where the law comes from

Every case page begins with a primary judgment from an open, official source — never a third-party headnote, citator, or a competitor’s study notes. The two corpora most relevant to UK and Canadian academics:

United Kingdom

  • Find Case Law — The National Archives (caselaw.nationalarchives.gov.uk) — judgments under the Open Justice Licence, ingested from the structured LegalDocML XML. The cleanest UK source we use.
  • BAILII (bailii.org) — the long-standing free judgment surface for England & Wales, Scotland and Northern Ireland, accessed with attribution and linked back from the case footer.
  • legislation.gov.uk — the official statute text behind references to UK Acts and statutory instruments.

Canada

The Canadian corpus is currently federal-focused. It is built from official, open court sources:

  • Supreme Court of Canada (the official SCC reporter)
  • Federal Court of Appeal
  • Federal Court
  • Tax Court of Canada
  • federal tribunals (e.g. the Canadian Human Rights Tribunal and immigration/refugee tribunals)

Provincial appellate coverage (provincial Courts of Appeal and superior courts) is being expanded — we call this phase 1, and we will not pretend the provincial side is complete until it is. We do not scrape CanLII: its terms forbid bulk redistribution, so we use court-direct and open-licence datasets and link to CanLII as a reference. The full per-jurisdiction breakdown — licences, ingest method, and what we explicitly do not redistribute — lives on the sources & attribution page.

How briefs and answers are generated

Case briefs, the AI answer cards on a case page, and the answers to a search query are generated on demand from the primary judgment text and grounded in it. They are not pulled from a model’s loose recall of a case and they are not copied from anyone’s proprietary summary. The source text is the judgment; the output is a structured study aid built from that text.

A brief is laid out the way a student would structure their own note: facts, issues, what was held, the reasoning, the significance, obiter, and the cases cited. An answer card distils a specific question down to the parts of the judgment that decide it. In both cases the goal is the same — make the structure of the law visible faster, so you can then go and read the case properly.

That last point is load-bearing. This material is an AI-assisted study aid, not a substitute for reading the judgment. Use it to orient and to revise; cite the judgment itself.

Honest limits

The disclaimer on every answer card

AI-synthesised — verify every case citation against the linked source before citing in coursework or essays. Not legal advice.

  • Not legal advice. caselaw is a study aid and research tool. Nothing on it is legal advice or a substitute for a qualified lawyer. For an actual legal problem, instruct a solicitor or barrister (in Canada, a member of your provincial law society).
  • The SQE bank is continuously verified — not “fully verified”. Our SQE question bank goes through ongoing adversarial review against primary sources. When a question is found to have a wrong keyed answer or a genuinely ambiguous one, it is quarantined— pulled from every selection pool (practice, mock, diagnostic, question-of-the-day) until it is re-written and re-checked. That process is real and ongoing, which is exactly why we will not describe the bank as “fully verified”. It is verified continuously, and defects are removed as they are found.
  • Citations trace to primary sources. Where a citation is shown, it is traceable to the primary source — the case page links back to the official judgment, and citation strings are formatted to the locally-authoritative guide (OSCOLA for the UK, the McGill Guide for Canada). Always verify a citation against that linked source before relying on it in coursework or an essay.
  • Coverage is uneven, and labelled. The UK corpus is broad; the Canadian corpus is federal-focused while provincial coverage expands. AI-generated study aids are labelled as such, and material that fails citation validation is held back rather than shown.

Privacy

Briefly and factually: we collect what we need to run the service — your account details and how you use the study tools — and we do not sell your data. The full policy is written in plain English on the privacy page; security disclosures and our practices are at /security.

Dig deeper

This page is the summary. The detail lives in three places, and a correction always reaches a human: