Case Study: loveholidays achieves 80% agent-assisted commits and higher code quality with CodeScene

loveholidays scales 80% agent-assisted commits with CodeScene

loveholidays, a UK-based travel agent, aimed to become an AI-first engineering organization but faced a challenge. As they adopted AI coding tools, their code quality began to decline because the AI agents generated complex, hard-to-maintain code that slipped past human reviews. They needed a way to scale AI adoption without sacrificing code quality or accumulating technical debt, and they had to prove the business impact. CodeScene provided the necessary guardrails to address this.

The solution from CodeScene integrated its CodeHealth metric directly into the AI development loop. This included setting a strict quality bar, using a CLI as a commit hook, and employing a CodeHealth MCP server to provide real-time feedback to the AI agents. As a result, loveholidays scaled to 80% AI-assisted commits, saw code health rise to elite levels with 70% of teams achieving a score above 9.75, and maintained AI build success rates above 90%. CodeScene enabled them to increase deployment frequency by 20-30% year-over-year while keeping the change failure rate stable.

Loveholidays