Updated
Updated · OpenAI · Aug 26
loveholidays Lifts AI-Assisted Code Changes to 79%, Boosting Deployments 73%
Updated
Updated · OpenAI · Aug 26

loveholidays Lifts AI-Assisted Code Changes to 79%, Boosting Deployments 73%

1 articles · Updated · OpenAI · Aug 26

Summary

  • 79% of loveholidays’ code changes are now AI-assisted, up from about 7% a year ago, while engineering headcount has stayed broadly flat.
  • Codex is letting product managers, designers and commercial staff build prototypes, make data and infrastructure changes, and deploy software without waiting in engineering queues.
  • More than 10 new search experiences have been built through Search Playground—mostly by non-engineers—and at least three are already live on the loveholidays website.
  • 93% of AI-assisted Data Platform changes now succeed, up from 58%, while broader self-service infrastructure workflow success rose to 90% from 63%.
  • About £136,000 in annual savings from lower cloud storage and reduced data-processing waste supports loveholidays’ push to make Codex a shared interface across the business.

Insights

As AI shifts tech bottlenecks from coding to governance, what hidden costs emerge when marketing teams bypass traditional engineering queues?
Unhealthy code doubles AI token usage; how much technical debt was secretly incurred before guardrails fixed early code quality drops?
If AI lets anyone build software, how do companies prevent non-engineers from accidentally breaking live systems despite automated guardrails?