Every claimant scored against the same matrix — with the page behind each element.
Settlement claim scoring support reads each claimant file for the elements a court-approved allocation matrix turns on: injury tier, exposure window, causation interval, and aggravating or mitigating modifiers. Every element comes back cited to its page, with a deficiency list for everything the file is still missing. The administrator scores it.
The matrix is court-approved. So is the referee.
We do not build for a case type until we can name the standard that defines a correct output and the person who grades the file against it. In matrix-scored settlement programs, both are on the docket in writing.
That is what makes the work checkable. The elements are enumerated in the allocation exhibit, the grader is appointed by the court, and every determination is written, appealable, and attached to money.
Six hundred pages on a good day. Eight thousand on a bad one.
On 3 August 2026 the JPML counted 205,540 pending actions across 162 MDLs. Scoring is the step between that docket and money reaching claimants, and it is done one file at a time, by hand, against a grid the court has already published.
Pages in one claimant file
A single matrix-scored claimant file typically runs between 600 and 8,000 pages, with the middle of the range sitting around 4,300. The scored elements are a few dozen words somewhere inside that.
Never one production
Records arrive in waves: the plaintiff fact sheet, the first medical production, the supplemental production after the deficiency notice, then the exhibits. In the demo case, 342 pages arrived across 2 packets and 11 pages were duplicates of pages already read.
Degraded, handwritten, and mis-filed
Faxed-then-rescanned charts, handwritten intake forms, and pages that belong to a different claimant entirely. In the demo case, page 140 was another patient's record; it was quarantined on its own line rather than folded silently into the read.
Four steps, starting with one real file.
Every allocation matrix is different, so we do not pretend to have yours pre-loaded. We build against your matrix, with your files, and you keep the scoring call the whole way through.
Send one real file
One claimant file as it actually arrives, plus the allocation exhibit or CMO that defines the elements. Any format, degraded scans included, under a signed BAA.
We return the read
Every matrix element we can find in the file, each one cited to the page it came from, and a deficiency list naming what is still missing against which element. No point total, no dollar figure.
We scope the build
You mark where the read is wrong or thin. That correction becomes the element definition, the output template, and the QA rules for your inventory. If it is not buildable to your standard, we say so at this step.
You run it
Your reviewers work from the cited read; the Claims Administrator scores and issues the determination. Related in this lane: asbestos trust claim review and abuse claim matrix review.
The rules the platform never breaks.
EVERY LINE CITED
HIPAA, under a signed BAA
Every file is handled under our Business Associate Agreement, from the first byte.
Never trains a model
Your records are never used to train any AI model — ours or anyone else's.
Every line cited
If we can't cite it, we don't say it. Every sentence links to its source page.
Deleted after delivery
Files are deleted 30 days after delivery, with a full audit log of every access.
Four capabilities behind every scored file.
Matrix element reads are one deliverable of the same platform that reads, sorts, and cites the whole record. The output format is yours, built in the custom report builder.
Mass Tort Batch Processing
Inventories are read in batches against the same element list in the same order, so the output does not drift from claimant to claimant.
Medical Chronology
The exposure-to-diagnosis timeline builds itself from the production, which is what the causation interval is measured across.
Missing Records Identification
The gaps are named, not guessed at: which matrix element has no supporting document, and where in the file that document should sit.
Verifiable AI Citations
If we can't cite it, we don't say it. Every element in the read links to the page it was pulled from, which is what an audit or an appeal asks for.
Settlement claim scoring, answered.
Send one file. We'll tell you what we can read.
No obligation. If the case type is buildable we'll scope it; if it isn't, we'll say so.