Claude as part of the development workflow
Once the direction had been established, Claude Code became an integrated part of the development workflow.
Claude was used to generate and refine code, tests and documentation, while the solution built on Cardioscan’s existing evaluation logic. The technology also helped establish a reproducible local development environment across multiple codebases, including legacy runtimes, local HTTPS and test data.
Rather than using Claude as a standalone chat tool, Consid configured it as part of the delivery system. Project-specific instructions defined commands, testing rules, deployment steps and documentation requirements. Recurring workflows could also be packaged for reuse, while quality assurance was automated through tests, linting, type checks and code coverage checks.
This allowed more of the consultant’s time to be spent on architecture, data-related questions and business understanding, while AI supported more routine and time-consuming parts of the development process.
A smaller team and approximately 30 percent faster delivery
The AI-assisted approach also had tangible effects on how the project could be staffed and delivered.
The delivery was completed by two people: one Consid consultant working closely with Cardioscan’s CTO. According to the project team’s assessment, a comparable feature developed using a more conventional approach would typically have required a team of at least 3–5 people.
Tasks that might otherwise have involved several specialist roles could be handled within the same delivery. Claude Design supported UX and design work, Claude served as a development partner during implementation, and AI also supported work related to the development environment, testing and documentation.
Overall, the solution is estimated to have been delivered approximately 30 percent faster than with conventional development.
AI did not, however, replace human responsibility for the solution. Claude worked from specifications created and owned by the consultant, while the implementation was verified through automated testing and functional validation together with Cardioscan’s CTO.
Consid remained responsible for architecture, implementation decisions, code quality and technical validation. Domain-specific and legal questions – particularly those concerning health data and anonymisation thresholds – were deliberately not delegated to AI.
From initial idea to a solution in production
The project progressed from an initial product idea to a new BGM reporting feature that has been live in production since September 2026.
The completed solution gives Cardioscan a new way to present aggregated employee health data in a structured, accessible and actionable format. At the same time, it provides a foundation for the continued development of Cardioscan’s corporate health offering.
The project also demonstrates how AI-assisted development can be applied throughout the entire product development lifecycle – rather than being limited to generating individual pieces of code. In Cardioscan’s case, AI was used from early design and requirements validation through to implementation, testing and documentation.
The result is a production-ready feature, a smaller delivery team and an estimated 30 percent faster delivery – while responsibility for architecture, quality and business-critical decisions remained firmly with the people involved in the project.
