A P&C carrier takes new-product launch from nine months to ten weeks.

# Insurance Customer Story — 74% faster product launch

**Industry:** Insurance
**Cloud:** Azure
**Outcome:** 74% faster product launch

## Summary
Thirty years of policy administration on WebLogic, claims running on an out-of-support Windows stack, and a rate-filing process that gated every new product. Decomposed, containerized, and shipped with state-by-state audit evidence intact.

## The Challenge
A US property & casualty carrier writing across 28 states ran its business on a policy administration system first written in the early 1990s — a Java monolith on WebLogic and Oracle, extended by three decades of state-specific rules, endorsements, and exceptions.The commercial problem was launch speed. Rating logic, forms, and product configuration were compiled into the monolith, so a new product — or a rate change in a single state — meant a code change, a full regression cycle, and a release train that ran quarterly at best. Filing to live averaged nine months. Competitors were quoting new coverage in a quarter.Underneath sat a harder deadline. Claims processing ran on .NET Framework 4.5 on Windows Server 2012 R2 — out of support, and named in the last state DOI examination as a finding. Any modernization had to preserve NAIC Model Audit Rule and SOX evidence, keep every state's filed rating behaviour bit-for-bit identical, and not pause claims for a day. A rewrite had been costed twice and shelved twice.

## The Solution
Discovery, assessment, and transformation ran as one workflow, so the dependency graph pulled out of the estate decided which services were extracted — and the rating behaviour that came out the other side was the behaviour that was already filed.CHAI Universe mapped all 640 applications across policy admin, claims, underwriting, and distribution, surfacing the dependencies that made the estate hard to touch — including 40+ undocumented batch jobs reaching directly into the policy schema. CHAI DART then graded each application on blockers, technical debt, and package health, which put the out-of-support claims stack and the rating engine at the top of the queue on evidence rather than opinion.CHAI Flow extracted rating, forms, and product configuration out of the monolith as separately deployable services, with rating rules lifted into configuration rather than code — so a state rate change becomes a config change, not a release. The claims application was containerized onto a supported .NET base image through a transform-step Dockerfile change, and CHAI Universe deployed everything to AKS as standardized Application Blueprints. Filed-rate parity was enforced by replaying 2.4M historical quotes through the extracted rating services and diffing premium against the monolith to the cent before any state cut over.

## The Outcome
Filing-to-live for a new product fell from nine months to ten weeks — 74% less calendar time — and a single-state rate change now ships in days as configuration, off the quarterly release train entirely.The claims stack is on a supported base image and the DOI examination finding is closed. Rating parity held: 2.4M replayed quotes matched the legacy engine to the cent, so no state needed to be re-filed to modernize the system that serves it.The carrier kept its audit position throughout. Every deployed workload traces back to a commit, which is what the Model Audit Rule walkthrough actually asks for, and the 40+ batch jobs nobody had documented are now in the map rather than in someone's memory. What was a twice-shelved rewrite became a sequence of state-by-state cutovers — 28 states, zero claims downtime.


## Stack
- Azure
- AKS
- Docker
- CHAI Universe
- CHAI DART
- CHAI Flow
- Application Blueprints
- Java
- WebLogic
- Oracle
- .NET Framework 4.5



## Contact
Book a demo: /contact/
Email: hello@cloudhedge.io