Industries → Insurance
Insurance
Modernising claims, underwriting, and risk data so decisions move at the speed of the customer.
60%
Faster fraud triage
35%
Fewer false positives
100%
Claims scored in real time
Featured use case
A problem we solved
The problem, solved
A mid-market insurer couldn't score fraud until after a claim was already paid. We moved scoring to claim intake, cutting investigation time by 60% within two quarters of go-live.
The challenge
A mid-market insurer's fraud signals were scattered across claims, policy, and third-party data sources. Scoring ran in an overnight batch, so fraud was often flagged only after a claim had already been paid out.
How we transformed it
Karsient consolidated claims, policy, and external risk-signal data onto a governed Databricks Lakehouse, then built a real-time scoring pipeline that evaluates every claim at intake instead of after the fact — with investigator feedback written back to retrain the model on a schedule.
Architecture
How data moves through a Insurance platform
Sources through to outcomes — the layers, and the live movement of data between them.
Live data flow · sources to outcomes
Benefits
What the business gained
Fraud scoring at claim intake, not after payout
One governed view of claims, policy, and risk data
A reusable feature store now powering additional risk models
Investigator time redirected from data assembly to actual investigation
What's next
Where this is heading
The same feature store built for fraud scoring is now being extended into pricing and reserving models — turning a single fraud use case into a reusable risk-data foundation for the wider business.
More industries
Where else we work
Let's talk