How the register is built.

Anyone can search the open part of the register without an account. Solicitors, paralegals, and trained helpline volunteers can additionally run the deep research flow, which combines AI-assisted disambiguation with a referenced report drawn from seven public sources.

Open search covers the organisations index, HSE enforcement notices, court venues and inquest coverage where in scope, direct links out to the original source for every entry, and bulk CSV download under CC BY 4.0 where applicable. No account, no login, no rate limit.

Gated research is for solicitors and helpline staff. When a claimant remembers a company name from decades ago, the flow disambiguates which legal entity they mean through a short Haiku 4.5 interview (up to four questions), then assembles a referenced dossier across Companies House, The Gazette, reported case law, the FCA Register and FSCS, the British Newspaper Archive and Internet Archive, and run-off insurer acquisitions. Every finding is a candidate match — verify against its source URL before relying on it.

The research flow is in alpha for partner solicitors and Mesothelioma UK helpline staff. Public access for self-represented claimants is gated behind the data-protection work in Phase E.

Live: HSE enforcement notices — fetched from the official register, archived byte-for-byte to private storage with SHA-256 content addressing, and parsed into the public dataset. Alongside them sit the stored media archive, the FSCS failed-firms directory, and a growing corpus of Gazette notices; the section below sets out what each covers.

Planned: HSE prosecutions, ONS mortality data, Companies House successor records. Each new source ships with its own per-source page documenting cadence, archival policy, and parsing approach.

Verified legal professionals only: court-judgment analysis, licensed from the National Archives’ Find Case Law, is gated to verified solicitors, barristers and their firms. It is not part of the free public dataset.

The register stores some sources in full and queries others live. Coverage is described here as windows and cadences rather than row counts, which go stale the week after they are written.

HSE enforcement notices — the stored corpus covers notices on the HSE public register from 2021 to the present, England and Wales, where an asbestos-related breach is cited. HSE publishes with a lag of up to nine weeks, so a scheduled weekly ingest re-checks a rolling 16-week window rather than trusting the latest page.

Media archive — stored articles from three publishers (Construction News, IOSH Magazine, and Thompsons Solicitors), each re-checked against the upstream URL on a rolling sweep so dead or amended links are flagged rather than silently trusted.

FSCS failed-firms directory — the full directory, refreshed weekly from fscs.org.uk as new failures are declared.

The Gazette — recipient pages search the full Gazette archive (1665 onwards) live at page load. Alongside that, a stored corpus of corporate-insolvency and Part VII transfer notices (1998 onwards) is building week by week for editorially-anchored organisations; name-only matches wait for editorial review before publication.

Query-time sources — Companies House, Find Case Law (verified legal professionals only), the FCA Register, and the historic press archives are queried live when a page or research report needs them. Nothing from these sources is stored beyond the citation shown.

The original source is canonical for every fact. AsbestosIQ does not edit, paraphrase, or infer beyond what the source published. Generated narrative paragraphs on notice and recipient pages are template-driven against the structured fields — never freehand. If the source is silent on a fact, the register is silent too.

1. Fetch from the official source URL on a published cadence.
2. Archive the raw page byte-for-byte to private storage with a SHA-256 content address.
3. Parse and validate structured fields against an explicit schema.
4. Write to the public register database; emit an idempotent change log.
5. Surface on the public site within minutes of archival, with the source link visible on every page.

Until Companies House identifiers land in Phase B, recipient pages group notices by exact recipient-name match. Two unrelated companies sharing a name will currently appear together under one recipient page; a permanent caveat explains this on every affected page. Cross-check Companies House before relying on a recipient page for due diligence.

The takedown and right-of-reply process is documented in the legal notice. Takedowns against the underlying source must be raised with the source operator (e.g. HSE for HSE notices); AsbestosIQ mirrors any source withdrawal in the register within 7 days of detection.

The full source code is published under AGPL-3.0. The methodology, schema, and fuzzy-matching pipeline are all documented in the public repository. Independent re-derivation of the dataset from the underlying sources is supported and welcomed.

The full source-by-source roadmap, with target dates and per-source dependencies, is maintained in the public project planning. Each new source will publish its own methodology page following the same discipline as this one.