New records arrive already compared to the old ones.
Supplemental record review software that compares new medical records to the existing file automatically. Each supplemental production is deduplicated against the records you already have, summarized on its own, and flagged wherever it agrees with the existing file, conflicts with it, or adds something new.
Deduplicated against the whole production.
A supplemental batch usually re-sends half the file. New pages are matched against everything already produced — not just each other — so what lands in review is only what’s genuinely new.
What’s new, summarized on its own.
The batch gets its own summary: new visits, new diagnoses, new work-status entries. A reviewer who already knows the file reads the delta, not the record from page one.
Flagged where it agrees, conflicts, or adds.
Each new record is compared against the existing analysis. Where a new note corroborates the file, it’s flagged as agreement; where it contradicts a date, a finding, or a prior report’s conclusion, the conflict is stated with citations to both pages.
A conflict is only useful if you can prove both halves.
When a late-arriving record contradicts the file, the flag carries two citations: the original page and the supplemental page. What changed, and where, is source-linked on both ends — audit-grade, legally defensible, and ready to hand to whoever wrote the first report.
See Verifiable AI CitationsFrom late batch to updated file.
The comparison work happens on arrival, before a reviewer opens a single page.
The late production goes in next to the existing file — no manual matching or renaming.
Duplicates are removed against the whole production; new records are compared to the existing analysis.
A cited summary of what’s new, plus agree / conflict / add flags on every record — both sides cited.
Who absorbs late records with it.
The people for whom a supplemental production lands a week before the deadline.
Supplemental review, answered.
A supplemental batch is treated as a comparison problem, not a new pile. It is deduplicated against everything already produced, summarized on its own, and each genuinely-new record is flagged for whether it agrees with, conflicts with, or adds to the existing file.
Each new record's extracted content — dates, findings, work status, diagnoses — is checked against the file's existing analysis. Corroboration is flagged as agreement; contradictions are flagged as conflicts with the specific disagreeing content quoted and cited.
If a new record contradicts something a previously drafted report relied on, that report is flagged for update, with the conflicting passage cited on both sides. The reviewer decides what changes; nothing is rewritten silently.
The full production. Supplemental batches routinely re-send pages from earlier productions, so matching only within the batch would miss most duplicates. Anything already in the file is removed and logged before review begins.
Yes. Every conflict flag carries two citations — the original page and the supplemental page — so the disagreement can be verified at both sources in one click each.
Related capabilities
Adjacent features on the same platform — every output source-linked and cited to page.
Duplicates found, confidence-scored, and collapsed without losing a citation.
ExploreThe visits and date ranges that should exist but weren’t produced, flagged.
ExploreTwo versions of the same record get scrutiny, not a silent merge.
ExploreBring the batch that just arrived.
Upload a supplemental set next to its file and read the delta, cited on both sides. Handled under our BAA; never used to train a model.