Creating a Better Digital Experience for a Complex Insurance Process
Clinical trials conducted across multiple countries bring different insurance and regulatory requirements. Each trial can involve multiple sponsors, protocols, countries, quotes, policies, and changes that need to be managed throughout its lifecycle.
Lockton Life Sciences saw an opportunity to bring more of that experience into a secure, multi-tenant portal where clients could submit and track trials while Lockton teams could manage them in one place. The initiative also gave Lockton an opportunity to explore how Claude Code could support a new development approach while maintaining the standards and controls expected of a regulated, client-facing product.
A New Development Approach Built Around Claude Code
Perficient and Lockton formed a focused 3.5-FTE delivery team that used Claude Code, with Claude Opus and Claude Sonnet models supporting development. The team established clear project rules, connected requirements and designs directly to the development process, and incorporated structured planning and automated quality checks.
Engineers remained responsible for direction, review, and key decisions, while Claude supported coding, testing, documentation, and other development work. Different models supported different needs, from complex planning and debugging to implementation, refactoring, and summaries.
- Architecture, security, and design standards provided consistent guidance throughout development.
- Claude Code worked directly from requirements in Azure DevOps and designs in Figma.
- Engineers reviewed plans and test scenarios before development began.
- Automated checks helped reinforce established standards with each change.
From Concept to Production-Ready Portal in Eight Weeks
In eight weeks, Lockton delivered a production-ready Foreign Clinical Trials Insurance portal with a focused 3.5-FTE team. The delivery model resulted in an estimated 60–70% cost reduction against the project’s traditional delivery benchmark.
Over the eight-week engagement, the team closed or advanced 104 work items into QA and merged 160 pull requests. User acceptance testing took just a few days and surfaced only a handful of bugs.
The work also gave Lockton a repeatable foundation for future development. Project rules, workflows, and operating practices were captured in a runbook that can help additional teams apply the approach as Lockton continues exploring AI-assisted software development.
