A Java estate modernized with AI in the loop
Three business-critical applications moved from Java 8-era stacks to current Java and Spring Boot, built on the verification harness that made AI-assisted change safe to merge.
Seats deployed. Timeline unchanged.
The client's core claims and policy operations ran on a portfolio of Java applications built in the Java 8 era. Three business-critical systems anchored the modernization backlog, including the platform handling claims appraisal.
The engineering organization had already rolled out an enterprise AI coding assistant. Seats alone, however, were not moving the modernization timeline.
The constraint
Not whether the tools could help, but how to make them reliably useful on decade-old code where a bad change lands in claims operations.
Ship, codify, transfer.
Our practitioners paired directly with the client's modernization team: real work alongside their engineers, not staff augmentation.
Audit the harness first
Before pointing AI tooling at production codebases, we audited the source-control organization for harnessability: the build health, test coverage, and verification scaffolding that decide whether AI-generated changes can be trusted at scale.
Real migrations, not demos
We paired with the client's modernization lead on live upgrade work, moving Java 8-era codebases to current Java and Spring Boot with AI-assisted modernization tooling inside the team's own IDE.
Turn what works into assets
Working patterns became upgrade playbooks, custom instructions tuned to the client's codebase conventions, and a technical deep-dive on AI-assisted modernization for Java teams.
Built to be taught onward
Enablement ran through the client's internal AI practitioner community: customization deep-dives, working sessions with engineering teams, and materials designed to keep the play running without us.
Upgrades, and the means to keep upgrading.
The migrations mattered, but the assets around them are what let the next application move without us on the bench.
Three business-critical applications on current Java and Spring Boot releases
A verification harness carrying unit test coverage from under 10% to over 90%
Upgrade playbooks and custom instructions tuned to the client's codebase conventions
Business-facing documentation for the claims appraisal system, generated from AI-extracted code analysis
Unit test coverage at close, up from under 10% before the engagement.
Business-critical applications moved to current Java and Spring Boot.
The upgrade playbook executing independently at engagement close.
Trust, not code generation, is the constraint. The tooling produced migration changes fast; confidence in them came from the harness.
Three things we would carry into the next one.
Trust is the constraint
Confidence came from tests, builds, and verification gates. That is what moving coverage from under 10% to over 90% bought: AI speed converted into merged, shipped upgrades rather than a pile of unreviewable diffs.
A judgment problem in a tooling costume
The mechanical parts of a Java upgrade are increasingly automated. Knowing which changes need a human's eyes, and what verified means for a claims system, is the durable skill. It transfers, but only if you codify it deliberately.
Enablement beats seat counts
The delta was not the license. It was embedded pairing on real work, playbooks written from that work, and structured transfer to the internal teams who own the codebase after we leave.
More engagements
View all →AI-assisted delivery inside an engineering org
An enablement program that sped up delivery without loosening discipline.
A platform foundation for agentic systems
A reference architecture that turned one-off prototypes into a repeatable way to ship.
Underwriting intake as a governed AI workflow
A high-friction intake process rebuilt as an agentic, auditable workflow.
Legacy setting the pace for everything else?
Bring us the application that has been next year's project for three years running. We will help define the first move and what verified has to mean before it ships.
Start a conversation →
AI-Native Delivery that turns ambition into execution across how you deliver, decide, and operate.