loveholidays expands software building beyond engineering with Codex
Online travel agent loveholidays says OpenAI Codex has helped move software development beyond its engineering organisation. Over the past year, the share of AI-assisted code changes rose from 7% to 79%, while deployment frequency increased 73% and engineering headcount remained broadly flat.
The company operates across eight European markets and processes 60 trillion package combinations each day. It is using Codex to enable product managers, designers and commercial teams to contribute to codebases, prototype customer experiences, and carry out guided data and infrastructure changes that once depended on specialist engineering support.
Search experiments without an engineering queue
One visible implementation is Search Playground, built by loveholidays engineers with the company’s design system, frontend technologies and Codex. The tool lets employees turn an idea into a working customer experience, collect feedback and test its value without first persuading an engineering team to prioritise a prototype.
More than 10 new search experiences have been developed through the Playground. Most were created by non-engineers, and at least three are now running on the loveholidays website. These include Inspire Me, which lets travellers explore trip types such as beach breaks and foodie escapes.
The company also used the approach for its Crisps from Abroad marketing activation. Rather than using an external agency to produce a competition-entry and holiday-inspiration microsite, the marketing team built it in hours with Codex and Search Playground while retaining the existing design system.
Making specialist practices available through workflows
loveholidays is applying the same model to its Data Platform and infrastructure. These environments had been designed for technical users familiar with specialist tools, repositories, source control and internal procedures. When a request failed or stalled, specialist engineers had to intervene.
Its engineering teams now encode best practices, instructions and validations into workflows that Codex can guide other employees through. Users can focus on the intended change while Codex helps propose it, run checks and move it through the release process. This enterprise rollout reflects the direction set by structured Codex enterprise deployment support for scaling Codex adoption through structured deployment support.
The reported results are substantial. Successful AI-assisted Data Platform changes increased from 58% to 93% during the year, while the number of Data Platform changes per support request rose fourfold. In broader self-service infrastructure workflows, success increased from 63% to 90%.
Capacity redirected to platform and optimisation work
The company says this changes the division of work. Teams can progress without waiting for specialist help, while engineers spend less time troubleshooting routine requests and more time improving the platform. CTO Mike Jones said loveholidays measures AI against business outcomes rather than adoption alone.
That added capacity has also been used for optimisation work. loveholidays says its Data Engineering team has cut cloud storage costs by about £36,000 a year and is saving approximately another £100,000 annually by reducing data-processing waste.
For businesses considering similar workflows, the practical implication is to make domain expertise reusable through instructions and validations, retain release checks, and assess the programme through change success, delivery capacity and operational outcomes rather than access to an AI tool alone.