# Claims Triage Automation Software > Claims triage automation that bulk-processes a claim portfolio against your criteria — closures cited, claims needing judgment routed to a human queue. Canonical page: https://medrecords.ai/product/automated-claims-triage/ --- NewWe now read the actual MRI and CT imaging — not just the radiologist's 1-page report. [See how →](https://medrecords.ai/product/imaging/) [Home](https://medrecords.ai/) ›[Product](https://medrecords.ai/product/) › — Automated Claims Triage CLAIMS TRIAGE AUTOMATION ## 128 claims in. 9 needed a human. Claims triage automation software from Medrecords AI bulk-processes a claims portfolio against criteria you configure. Only the claims that genuinely need judgment reach a person; the rest are documented, cited, and closed under your rules. Portfolio-scale intake still gets file-level reading, and the human queue starts at the decision point. Closure is rule-based against criteria you configure, with every closure documented and cited. Claims needing judgment route to a human queue — the AI never makes adjudication decisions. [Test a file ](https://medrecords.ai/test-a-file/?src=product-automated-claims-triage) [Book a demo](https://medrecords.ai/demo/) Portfolio run · 128 claims Your criteria applied 119 Met your criteria · documented 9 Routed to human queue Human queue · with cited reasons Treatment gap vs. plan criteria NEEDS JUDGMENT — p.38 Documentation incomplete — packet 2 NEEDS JUDGMENT — p.140 History conflicts with claim form NEEDS JUDGMENT — p.14 Rule-based against **your configured criteria** — humans make the decisions. 128 in One portfolio, processed in bulk 9 Reached the human queue, with reasons Every closure Documented and cited to the file ### Your criteria. Applied exactly, every time. Triage runs on rules you configure: coverage windows, documentation completeness, treatment-to-injury consistency, and whatever your book requires. The system applies those rules the same way on claim one and claim 100 — it never substitutes its own judgment for yours. Criteria configured per line, client, or program Anything outside the rules defaults to a person Your criteria · configured once Coverage window matches dates of service — RULE Required documentation present and legible — RULE Treatment consistent with reported injury — RULE Anything else — route to a person — DEFAULT The system applies **your rules** . It doesn't invent judgment. Portfolio intake 128 CLAIMS Every page of every claim read — 342 pp/file max Duplicates removed automatically — 11 dupes Illegible pages quarantined, not skipped — p.140 Bulk doesn't mean shallow — **each file is read in full** . ### Portfolio-scale intake, file-level reading. Drop a portfolio and each claim file is read in full — every page OCR'd, duplicates removed, illegible pages quarantined rather than skipped. Bulk processing here means the whole record gets read faster, not that less of it gets read. Every page of every claim read, at any volume Dupes removed and low-quality pages flagged automatically ### The human queue starts at the decision point. The claims that reach your team arrive with the reason they were routed — the conflicting passages, the missing documentation, the criterion that didn't resolve — each cited to its page. Adjusters spend their time deciding, not re-reading files that were never in question. Routing reasons attached, cited to the source pages Queue ordered by what actually needs judgment Adjuster's queue · one claim Why this claim reached you Claimed history conflicts with the intake note — both passages attached, cited to their pages. p.14 — p.38 — NEEDS JUDGMENT Adjusters start at **the decision point** , not page one. Every outcome · documented and cited Met criteria — closure memo written, cited DOCUMENTED Needs judgment — routed to a person, with reasons HUMAN QUEUE AI decides the claim NEVER Rule-based triage · adjudication stays human Drawn line ### Documented, cited, and rule-based — or it goes to a human. Straight-through processing is only defensible if you can show your work. Every auto-closed claim carries a closure memo citing the pages that satisfied your criteria; every routed claim carries the reason it needs a person. The AI applies rules and drafts documentation — adjudication stays human. [See Verifiable AI Citations ](https://medrecords.ai/product/citations/) ### From portfolio to worked queue. 3 steps between a claims dump and a queue that contains only real decisions. 01 Configure the criteria Define the rules per line or program — coverage, documentation, consistency checks. 02 Run the portfolio Every claim file is read in full and measured against the rules, at bulk scale. 03 Work only the queue Claims meeting criteria close with cited documentation; the rest reach your team with reasons. ### Who triages claims with it. Teams measured on cycle time and leakage — who still have to defend every closure. [ TPAs & case management Client-specific criteria per program, applied identically by every desk, every day. For TPAs ](https://medrecords.ai/solutions/tpas/) [ Insurance carriers Straight-through handling for the claims that never needed an adjuster — with the paper trail intact. For carriers ](https://medrecords.ai/solutions/insurance-carriers/) FAQ ### Claims triage automation, answered. For claims that fully satisfy your configured criteria, yes: they are documented, cited, and closed without touching a desk. But the term comes with a caveat we enforce — closure is rule-based against your criteria, and anything requiring judgment routes to a person. The AI never adjudicates. Criteria are set per line of business, client, or program: coverage-window checks, required documentation, treatment-to-injury consistency, and thresholds you define. Rules are explicit and reviewable, so compliance can read exactly what closes a claim and what routes it. Any claim that fails or falls outside a rule: conflicting histories, incomplete documentation, treatment inconsistent with the reported injury, or anything your rules do not cover. Each arrives with its routing reason cited to the source pages, so the adjuster starts at the decision point. Every auto-closed claim gets a closure memo stating which criteria were met and citing the pages that satisfied them. If a closure is ever questioned — by an auditor, a regulator, or a claimant — the supporting record is attached, line by line. Yes. Portfolios are processed in bulk, and each file inside the batch is still read in full: every page OCR'd, duplicates removed, illegible pages quarantined for review rather than silently skipped. ### Related capabilities. Same engine, same citation standard — claims-side capabilities on the same cited engine. [ Demand Package Audit Every diagnosis, treatment claim, and billed figure in an incoming demand, checked against the record. Explore ](https://medrecords.ai/product/demand-package-audit/) [ Fraud & Inconsistency Signals Intra-record contradictions and cross-claim provider patterns, flagged and cited. Explore ](https://medrecords.ai/product/claims-fraud-signal-detection/) [ Relatedness & Compensability Signals Each visit and charge tagged related, unrelated, or disputed — cited, for your reviewer. Explore ](https://medrecords.ai/product/treatment-relatedness-compensability/) [ Mass Tort Batch Processing The same bulk engine, pointed at dockets: chronology and criteria across every file at once. Explore ](https://medrecords.ai/product/mass-tort-batch-processing/) ### Point your criteria at a real portfolio. Book a demo and watch a batch run against rules you define — closures documented, judgment calls routed. Handled under our BAA; never used to train a model. [Test a file ](https://medrecords.ai/test-a-file/?src=product-automated-claims-triage) [Book a demo](https://medrecords.ai/demo/) [All capabilities ](https://medrecords.ai/product/)