Holiday Extras reports stronger app bookings after Codex rebuild

Holiday Extras has reported a 32.5% increase in open-to-book conversion on its rebuilt main app after using OpenAI Codex, while the app’s error rate fell from 19% to 9%. The UK and European travel business says it produced a fully working rebuild in two weeks, a level of development progress that had previously taken more than six months.
The company, which helps customers search, book and manage trips, had historically centred major changes to its mobile app on an annual release. Its rebuild also changed the delivery cadence: in the first month after launch, the main app moved to weekly releases, each containing multiple new features.
A rebuild designed around customer knowledge
Holiday Extras says the new app delivers a more personal experience based on customers’ bookings, trips and history. Liam Thompson, VP of AI Innovation, worked with engineers to translate his understanding of customer needs into software. The teams created small, documented packages that could be reused, while engineers concentrated on architecture, APIs, security and development workflows.
Codex supported concurrent work on iOS and Android features, allowing improvements to reach both platforms rather than leaving Android to follow behind. The company also cites its Countdown product, where feature delivery moved from roughly one addition every six months to multiple additions each week.
In one Countdown case, a customer complained that a list view was missing. An agent prepared a fix, Thompson approved it and the update was submitted; the customer then changed a one-star review to five stars. Holiday Extras presents this as an example of a faster route from customer feedback to product change.
Broader access, with engineering controls
Company-wide ChatGPT Enterprise access formed the foundation for the rollout, and Codex extended that access into customer-facing applications and internal systems. CEO Matthew Pack says the aim is not limited to a small group of specialists, but is to bring the wider team into the work of using AI.
That experimentation is paired with guardrails. Employees are encouraged to try AI, while engineers provide the software packages, security controls and workflows required to make experimentation practical. David Lee, Director of Innovation and Growth, points to an experienced accountant who rebuilt finance systems using his operational knowledge and feedback from colleagues.
The operating model also reflects the expanding enterprise tooling around Codex: OpenAI Codex workplace plugins and tools highlights plugins and workplace tools that can extend AI-assisted work beyond code, while Holiday Extras is applying its own customer and operational expertise to applications and systems.
Customer signals become work items
Holiday Extras is also using automation in operations. Lee describes a system that converts customer chats about airport transfers into actionable tickets, helping teams identify issues across the supplier network. The stated goal is to turn customer insight into work that teams can address.
Alongside Codex, ChatGPT Enterprise supports staff across the business, while OpenAI models power the app’s personalised experience. The company says its next ambition is for more systems to be administered through AI, with employees directing work and shaping outcomes. For businesses, the practical implication is to combine domain expertise, reusable engineering foundations and security controls so that faster experimentation can produce changes that teams are able to operate responsibly.

