§ METHODOLOGY · HOW IT WORKS

Our methodology

How ClaimGavel produces an assessment — the data behind it, how the model uses that data, and the limits you should keep in mind.

Last updated · 19 August 2026

The data behind an assessment

ClaimGavel is built on published, public-interest data rather than marketing material. The main sources are:

  • Published judgments from Find Case Law — the National Archives service that publishes the decisions of the courts of England and Wales. Every case in our library has been read, recorded with a verbatim quotation from the judgment, and linked to its source. A script re-checks each quotation against the published judgment, and any record that fails the check is removed rather than published.
  • NHS Resolution settlements — the annual breakdown of NHS clinical negligence claims, payouts and outstanding provisions.
  • Judicial College Guidelines — the standard compensation bands used by courts for general damages (pain, suffering and loss of amenity).
  • Coroners’ Prevention of Future Death reports — published on judiciary.uk, used to understand which failures recur in NHS care. These concern deaths and inquests, not compensation claims, and we use them only as background.

Judgment material is used under the Open Justice – Licence v2.0 and is attributed wherever it appears. Coroners’ reports are used under the Open Government Licence v3.0. We do not reproduce judgments in full, and we remove any record whose judgment is later withdrawn or replaced.

How an assessment is produced

When you describe your situation, ClaimGavel:

  1. uses an AI model to identify the most likely claim type, the relevant UK jurisdiction, and the applicable time limit;
  2. compares the facts you give to our library of verified, published judgments of the courts of England and Wales (each linked to its source on Find Case Law) — the assessment can only cite judgments from that library, never a case of its own making;
  3. produces an indicative compensation range, a rough indication of how strong the claim looks, and the kinds of evidence that tend to matter.

The assessment is generated with the help of a third-party AI model (currently the Google Gemini API). The description you enter is processed by that provider to produce your result. Please avoid including more personal or sensitive detail than you need to.

What an estimate does and does not mean

An estimate is not a valuation, an offer, or a guarantee. Compensation ranges are drawn from how broadly similar matters have been resolved in the past. Your actual outcome depends on the full facts, the available evidence, the law as it applies to you, and decisions only a qualified solicitor (and ultimately a court) can make.

Limitations you should know about

  • The assessment is general information, not legal advice, and is AI-generated — it can be incomplete, out of date, or simply wrong.
  • Time limits for bringing a claim are strict and vary by claim type and jurisdiction. Do not rely on our indication alone — confirm your deadline with a solicitor as early as possible.
  • Our data covers many but not all claim types, and historic awards are not a promise of future results. Most claims settle without a judgment, so published cases under-represent some kinds of injury.
  • We cannot see documents, medical records or correspondence, so we work only from what you tell us.

Keeping it current

We refresh our underlying datasets periodically as new judgments and official reports are published. NHS figures reflect the most recent published NHS Resolution dataset at the time of writing.

Want the human version?

If your assessment suggests you may have a claim, the right next step is to speak to a qualified solicitor. You can ask to be contacted by Tomlin & Partners, the specialist medical negligence firm this site is run for, or take your assessment to any solicitor you like. Either way, the decision — and the legal advice — stays with you and the solicitor you choose.