NHS Resolution's Annual Report and Accounts for 2025/26 makes the direction of travel hard to miss. Under its "Resolution Through Collaboration" strategy, a record 84% of clinical claims were kept out of formal court proceedings. Its mediation service, now in its tenth year, handled 143 claims, with 71% settling on the day or within 28 days. The message is consistent across the report: resolve fairly, resolve early, and use the evidence to learn.
Earlier resolution changes what is asked of the medical evidence. When parties are trying to reach a fair view before proceedings begin, they need a clear, well-evidenced opinion sooner. That opinion still depends on the same starting point it always has, which is a careful reading of the records.
Where the expert's time actually goes
A modern clinical negligence or personal injury bundle can run to thousands of pages, often scanned, sometimes barely legible. Much of an expert's time goes into finding, ordering and cross-checking what is in those pages before any real analysis begins. That is time spent on retrieval rather than on the clinical reasoning only the expert can provide.
What Verify does
ALLDOQ Verify takes a first pass through the medical records so the expert can start from an ordered, referenced foundation instead of a raw bundle. It reads each page, extracts the clinical facts into typed tables, and builds a date-ordered chronology from what the records actually say.
Reads the whole bundle. Scanned and text-layer records are read page by page, with a confidence score on every page and a vision model for the poorest scans.
Extracts typed facts. Medications, diagnoses, investigations, procedures, allergies, observations, encounters and immunisations are pulled into structured tables, with UK drug naming kept as written.
Builds a chronology. Every fact is dated on a single timeline, derived from the records, so each entry can be traced rather than assumed.
Answers with citations. An expert can ask a plain-language question, such as whether a drug was ever prescribed or what the admission history looks like, and every answer returns with its source page.
Referenced and auditable by design
The point that matters most for early, collaborative resolution is verifiability. In Verify, every extracted fact and every answer links back to the document, the page and the exact quoted text. The draft an expert works from is a referenced record where each statement can be checked against its source in a single step, so nothing has to be taken on trust. Typed facts are queried first, with keyword and vector search as a safety net, and a verification step is used to filter out negations and mentions that belong to a different patient.
Efficiency, consistency and transparency
The value shows up in three ways. Efficiency, because the expert spends less time locating and ordering information and more time applying their expertise. Consistency, because the same disciplined first pass is applied to every bundle rather than depending on how much time was available that week. Transparency, because the reasoning behind an opinion can be followed back to the underlying pages by the expert, the instructing party and anyone reviewing the work.
The expert stays in control
Verify does not form the opinion. It prepares the ground so that the clinician can. The chronology, the extracted facts and the cited answers are there to sharpen clinical judgement, and the expert remains responsible for what the evidence means. Patient identifiers are encrypted, access is invite-only with two-factor sign-in, and no patient detail is sent by email.
Aligned with where the system is heading
NHS Resolution is asking the system to resolve more claims earlier, to work more collaboratively, and to treat the evidence as something to learn from. Tools that make the record faster to read, easier to verify and more consistent to work with support that direction. ALLDOQ Verify is built to do exactly that, by helping experts produce robust, well-referenced work more efficiently while keeping their judgement at the centre.
