The claim decisions you can't seeare the ones that leak.

Mysa reads the whole claim, scores each decision against how your team actually decides, and hands the adjuster a case that's ready to call.

How it works
Runs on your infrastructure · your data stays yours · you make every call
5–10%
Claim leakage
1,000s
Decisions scored on every claim
Instructions
1
Receive the FNOL
Read the inbound notification and extract the basics into Claim Intake.
2
Call the policyholder to confirm
Place a short confirmation call, then move on.
Voice Agent
Confirm identity and plate, capture the circumstances into Claim Intake.circumstances.
3
Decide liability, then hand it over
Apply the convention, write the apportionment and reserve into Settlement Model, and pass the adjuster a decision-ready case.
Suggested by your decision graph
Refer to SIU before this one settles.
Source this claimant and this repair shop have paired on 4 claims in 18 months — across 3 different adjusters.
Why every one settled just under your referral threshold. No single file shows the pattern.
Adjuster deskLIVE
ReadChaseAssembleReasoning
Automated · agentsYour call

Intake, coverage, fraud and liability ran on their own, most of it as deterministic code rather than model calls. What's left is your call.

Decision readiness
FNOL captured in full
14 of 14 fields · driver licence chased and received
Coverage validated
In force · rear-end peril covered · within limit
Fraud screen
6 indicators run · 1 needs your eyes
Liability decided
TP 100 / insured 0 · convention art. 4
Next actions queued
Notify TP insurer · book inspection · request estimate
Notified 9 days late, with a similar claim 4 months ago. Worth a look before payout.
This pattern turned out fraudulent in 12% of matched claims.
Decision-ready

The worst claim decisions never look wrong. That's the problem.

Your sharpest claims judgment lives in your senior people's heads, with no backup. Ask which adjuster is best on liability, or what last quarter's worst call cost you — nobody can answer. That blind spot leaks 5 to 10 cents on every dollar you pay.

01

No two adjusters agree

The same claim gets a different answer depending on who picks it up. The standard lives in people's heads, not a system you can see or hold onto.

02

The bar drifts, silently

Standards move case by case, and nothing tells you it happened.

03

You find out too late

The wrong call surfaces a quarter later, a blown reserve or a missed recovery, slow and costly to undo.

A claim is not a form.
It is a graph of decisions.

You've heard of a context graph. This is the one built for claims decisions — every node is a call your team makes, scored against how you decide, and it deepens with every claim.

What the decision graph does

Reads the whole claim

Every document, field and note becomes a node, wired to the evidence behind it — not a form, a graph.

Scores every decision

Each call is scored against how your book actually decides, not a generic model.

Flags what leaks

Cost anomalies, missed recoveries and patterns no rule caught — surfaced before payout.

Deepens every claim

Every validated call is written back. By claim twenty it finds its own precedents, unprompted.

Carries its trace

Each decision cites the rule, the confidence, the precedent, and who confirmed it.

Yours to own

Model-agnostic and sovereign. The graph you build is your asset, on your balance sheet.

We handle the case. Your adjuster handles the judgment.

Everything up to the decision, done and scored. The call itself stays yours.

Reading · CLM-2026-08470/4
FNOLreading…
Statementhandwritten
Repair estimate
Medical invoice

Read

The pile a claim arrives as, an FNOL, phone photos, a note scrawled at the scene, read and turned into one clean, structured record. Your team starts from a case, not a folder of PDFs.

See the reading
Voice Agentcalling
Central Auto Body
garage on the claim

Chase

Spots what's missing, a second estimate, an unsigned form, then picks up the phone and gets it, before the claim ever stalls on your desk.

See the chase
6 sources · 1 case0/4
Policyholdermerging…
Third party
Circumstances
Damages · $8,420

Assemble

Six documents across three channels, reconciled into a single clean case. No gaps, no duplicates, nothing left for your team to stitch together by hand.

Follow one claim

Reason

Scores every decision in the claim against how you actually decide, with a confidence on each.

See the decision desk

Catches the leak

A recovery your team would have written off. Mysa spots that the third party is at fault, flags what you can claw back, and catches it before it leaks away.

See the drift

Hands it to your team

Your adjuster gets a decision-ready case, evidence and confidence attached, with a heads-up wherever a call drifts from how your team has decided before. You make the call.

See who decides
Combined ratio & retention

The numbers you report on, scored on every claim.

Speed, accuracy, cost, experience. These four set your combined ratio and decide whether the customer renews. Mysa scores all four on every decision while the claim is still open, rather than in a quarterly audit three months after the money has left.

5–10%Accuracy
Loss ratio

of every payout leaks to a wrong call: a split that should have been 100/0, a reserve left too high, a recovery no one pursued. It stays invisible until it is paid.

HoursSpeed
Cycle time

The decision in hours, not weeks. Every day a claim sits accrues handling cost and lets recoveries go stale.

5–8 ptsCost
Expense ratio

Handler hours, rework, arbitration. The expense ratio holds five to eight points of combined ratio, and it is the work we do.

#1Experience
Retention

For most customers the claim is the only time they test the policy. It is the single biggest driver of whether they renew.

Decision qualityfirst 12 weeks
Scored against your closed claims before it touches a live one. You watch it work first.

We'll map where your claims leak, before you give us anything.

Drawn from your public filings and customers' complaints, with no data from you. 30 minutes, and the map is yours to keep.

Get your decision map
Trust

Your data never leaves. You make every call.

Sovereign by design. Mysa runs where you already do, logs every decision to its source, and leaves the final call with your team.

Sovereign deployment

Runs in your cloud, on-prem or your own VPC, across AWS, Azure or GCP, in your region. Your data never leaves without a signed frame.

You decide, always

Human-in-the-loop by design. We recommend with evidence and a confidence; your adjuster makes every call.

Auditable by decision

Every decision carries its trace: which rules fired, at what confidence, the precedent, who confirmed. Questioned later, the answer isn’t “the AI said so” — it’s the reasoning already on record. Built for disputes and GDPR Article 22.

Enterprise security

Encryption in transit and at rest, SSO / SAML, role-based access and granular permissions.

Your models, your graph, your asset

Model-agnostic, no lock-in, no third-party training on your data. The decision graph you build scores and sharpens every call — yours to keep, and to own on your balance sheet.

Privacy by default

Anonymized and masked before processing. Never any health data. Data minimization built in.

The full security postureSubprocessors, DPA, encryption details, residency and certifications — all in one place.
See our security posture

Every decision you run makes the next one sharper. Over time, the claims decision desk becomes the layer your whole claims operation decides on.