Consistency
Every damage reduction, total-loss valuation, and arbitration contention is built the same way, against the same rules. Claims Agent prevents adjuster-to-adjuster variance on the same facts.
claimsolutions.ai
Claims Agent Suite
Recover more from every auto subrogation file. Claims Agent audits every disputed dollar, drafts evidence-backed negotiation responses, and prepares statute-grounded arbitration filings, inbound and outbound, while the adjuster keeps final control.
The same specialists also handle damage, total-loss, rental, statement, and scene work on first-party files, all logged to one auditable record.
Already a customer? Sign in$18BE-Subro Hub demand value, 2025
92%E-Subro Hub market share
50US states in statute RAG
The solution
Claims expertise embedded directly into the workflow.
A set of specialized AI agents, each an expert in one part of the claims lifecycle, reviews every estimate, valuation, and demand against your carrier specific rules, the same statutes, and the same evidence every time. New adjusters get senior-level consistency on day one; experienced adjusters get their routine work done in a fraction of the time so they can focus on judgment calls and negotiation.
Recovery & leakage prevention
Line-by-line audits catch inflated line items, missing documentation, and comparative-negligence opportunities before they cost real dollars.
Capacity & speed
Agents handle the repetitive, document-heavy work, ultimately freeing adjusters for detail-oriented judgment calls, and letting a fixed team absorb more volume.
Onboarding & institutional knowledge
Rules and statutory guidance are embedded in the workflow, not locked in a mentor's head. Claims Agent allows junior adjusters to produce senior-quality work immediately, and knowledge survives turnover.
Visibility
Every audit, transcription, and negotiation is logged and reportable. Senior leaders see outcomes and trends, not just spreadsheets.
Why not generic AI
- Statute-grounded arbitration arguments across 50 states, not generic AI reasoning.
- Handles both sides of subrogation, outbound demand defense and inbound reduction defense, with the same evidence standard.
- Labor rates audited as line items against the estimate itself, with body, mechanical, and paint rates logged to every audit record.
- A rules-based 50-state total-loss threshold engine, not a one-size-fits-all formula.
- Carrier-specific tone and rules via prompt overrides, with no code deployment required to onboard a new client.
Subrogation operating model
The same specialists work both sides of subrogation.
Inbound and outbound follow the same four control points, onset or outlay, audit or assembly, ESH negotiation, and arbitration, run by the same evidence-grounded agents.
01
Adjuster starts from chat, evidence upload, or dashboard history.
02
Claims Agent delegates to the correct specialized claim workflow.
03
Tools apply statutes, thresholds, evidence, screenshots, or scene photos.
04
Saved audit trail, exports, and next-step outputs stay tied to the claim record.
Inbound
The adverse carrier is demanding against us.
Inbound demand received
The adverse carrier issues an inbound subrogation demand across operational files.
Confirms claimant, carrier, and claim number, then classifies the demand and routes it to the right specialist.Automated reduction evaluation
The demand is audited against state guidelines, statutory boundary metrics, and company rules.
Audits the estimate, valuation, or rental figures and produces cited reductions.Reduction rebuttal counter
The adverse carrier responds to the reductions. The representative downloads an evidence-grounded rebuttal to use in the negotiation or arbitration response.
Builds a citation-backed rebuttal defending the reductions against the adverse carrier's pushback.Inter-company arbitration
A flagged scene location becomes aerial and street-level imagery plus an adjuster-tunable diagram, combined with the audit findings to contest reductions and capture backend leakage vectors. The adverse carrier's own demand and arguments can be packaged into the final output.
Compiles scene evidence and audit findings into a statute-grounded respondent filing.Outbound
We are demanding against the adverse carrier.
Final payment ledger
A final property damage ledger is generated, triggering file identification for recovery.
Confirms claimant, carrier, and claim number for the final payment ledger.Evidence compilation packet
The audited estimate, valuation, or rental findings are compiled alongside statement evidence into a submission-ready packet.
Assembles the audit findings and evidence into one demand packet.Portal settlement counters
The representative negotiates liability and damage recovery percentages inside the ESH portal, grounding the argument in industry standards and the file's own evidence.
Drafts the portal counter-argument, grounded in industry evidence and the file's own facts.Binding file submission
A flagged scene location becomes aerial and street-level imagery plus an adjuster-tunable diagram, paired with the evidence-cited negotiation position to secure final recovery metrics.
Pairs scene evidence with a statute-grounded applicant filing to secure recovery. Product system
Platform services that support every specialist agent.
A common platform layer sits under every claim agent: a state-specific statute vault, carrier-level prompt overrides, two-pass PII redaction, and a live streaming chat and dashboard experience.
Statute evidence vault
Retrieval-augmented search over 582 statute documents across 50 US states.
Every arbitration argument is grounded in the actual applicable law, not a generic model guess.Automatically queried when generating contentions.50-state total-loss threshold engine
Applies the correct statutory total-loss formula for all 50 states automatically, instead of one generic calculation.
Consistent, statute-correct total-loss decisions nationwide.Backed by a dedicated evidence vault for towing, ACV, and salvage rules.Multi-tenant prompt overrides
Carrier- or company-specific instructions can be stored and swapped without a code deployment.
Each carrier gets its own tone and rules without custom engineering.Falls back to a sensible default automatically.2-pass PII redaction
Cloud DLP automatically detects and redacts personally identifiable information before and after every model call.
Reduces compliance risk when handling sensitive claim data.Applied across every agent.Live streaming chat + dashboards
Real-time, conversational interface with live status updates and dashboard views of past audits.
Adjusters see progress in real time; managers see historical results.Includes exportable Word-document reports. Security & trust
Adoption doesn't mean new compliance risk.
Everything runs inside a dedicated Google Cloud project on Vertex AI: no third-party AI, no model training on your data, PII redacted by Cloud DLP, and every upload tied to an authenticated user and expiring on a schedule. The answers your security team will ask for, up front.
Your data never trains a model
Prompts, uploaded documents, recorded statements, and agent outputs are customer data under Google Cloud's Vertex AI enterprise terms. Google doesn't use them to train its models or share them with other customers.
No third-party AI in the pipeline
Transcription, document reading, and drafting all run on Gemini through Vertex AI inside a dedicated Google Cloud project. No OpenAI, no aggregators, no proxies between your files and the model.
PII redaction
A two-pass Cloud DLP process scans adjuster-entered text and every agent output, replacing Social Security, driver's license, and phone numbers before anything is stored or shown.
Google Cloud BAA in place
A HIPAA Business Associate Agreement with Google Cloud covers the services the product runs on: Vertex AI, Cloud Run, Firestore, Cloud Storage, and Sensitive Data Protection. Bodily-injury and casualty files can be reviewed under it.
User-scoped access
Every upload, session, and dashboard query is tied to a verified user identity, so one adjuster's files and sessions are never served to another user.
Least-privilege backend
The backend runs on a dedicated service account holding only the IAM roles it needs, and every API key and credential lives in Secret Manager, never in code or config files.
Uploads expire automatically
Raw uploads are deleted 7 days after upload and generated artifacts at 30, by bucket lifecycle rules with public access prevention enforced. Nothing lingers waiting for a cleanup job.
Audit trail
Every audit, transcription, filing, and negotiation is written as a timestamped record tied to the user who ran it, with platform activity captured in Cloud Logging.
ROI & proof points
Model your own subrogation file capacity.
Every figure here carries its underlying assumption. Recompute against a carrier's own volume before quoting it to that carrier.
Subrogation file capacity calculator
Drag to your own subrogation file volume to see hours and FTE capacity reclaimed, scaled from the suite's modeled cadence.
38,880 files / year
19,440Hours reclaimed / year
13.05FTE equivalent capacity
Reference cadence: one steady-state capacity, three views
| Volume | Hours reclaimed | FTE equivalent |
|---|---|---|
| 180 files/day | 90 hours/day | 13.05 FTE |
| 3,240 files/month | 1,620 hours/month | 13.05 FTE |
| 38,880 files/year | 19,440 hours/year | 13.05 FTE |
Run a pilot
Price a pilot pack, or see it first.
An output is one finished agent deliverable: a negotiation response, an arbitration filing packet, or a damage audit. Pilot packs are one-time and prepaid, with outputs used within 90 days of purchase.
Select a pack above to continue.
Bring this discipline into onboarding, too.
Claims School is the same evidence-grounded standard, built for training: sequenced curriculum, hands-on simulations, and grading tied to statute-cited remediation.