Holiday Extras rebuilt its main travel app in two weeks using OpenAI's Codex, a feat that previously took six months. The rebuilt app cut error rates from 19% to 9% and lifted open-to-book conversion by 32.5%.
TL;DR
- Holiday Extras rebuilt its main travel app in 2 weeks with OpenAI's Codex, replacing a six-month process.
- The rebuilt app cut error rates from 19% to 9% and increased conversions by 32.5%.
- Leadership now plans to let more teams use AI to directly shape products and operations.
What happened
Holiday Extras, a UK travel technology company, rebuilt its main mobile app in two weeks using OpenAI's Codex, a model fine-tuned for software development. The project replaced an annual release cycle that had previously required six months of work.
The rebuilt app delivered a more personal experience by integrating customer booking history and preferences. Engineers and designers collaborated to create reusable code packages while maintaining security, API design, and development workflows.
Open-to-book conversion on the main app rose 32.5% after the rebuild, and the app's error rate dropped from 19% to 9%. The team also shifted from annual releases to weekly updates in the first month, shipping multiple new features each week.
A separate Countdown feature moved from a six-month release cadence to multiple updates per week. Customer feedback loops accelerated: an agent submitted a fix for a missing list view, which went live the same day and changed a one-star review to five stars.
Why it matters
For AI startups and developers, the Holiday Extras case shows how AI coding tools can compress multi-month engineering cycles into weeks, enabling faster experimentation and iteration. The results demonstrate that AI-assisted development can improve both user experience and business metrics like conversion and error rates.
For startup founders, the story highlights the strategic shift from relying on a small engineering elite to empowering cross-functional teams to build directly. Holiday Extras' approach—training staff, setting guardrails, and scaling reusable components—offers a template for AI-driven product development.
Investors should note that Holiday Extras is reallocating engineering capacity from maintenance to innovation, suggesting a potential lift in product velocity and market responsiveness. The company's ambition to let more employees shape systems through AI could redefine its competitive edge.
The case also underscores the importance of partner choice: Holiday Extras credits OpenAI's Codex for its ability to reason through complex problems, propose plans, and deliver independently, while ChatGPT Enterprise supports broader internal adoption.
Key facts
- Holiday Extras rebuilt its main travel app in 2 weeks using OpenAI's Codex, replacing a six-month process.
- Open-to-book conversion on the main app increased 32.5% after the rebuild.
- The main app's error rate fell from 19% to 9% following the rebuild.
- The team shifted from annual releases to weekly updates in the first month after the rebuild.
- Countdown moved from adding a feature roughly every six months to multiple additions a week.
- The rebuilt app integrated customer booking history and preferences for a more personal experience.
- Holiday Extras provided company-wide access to ChatGPT Enterprise alongside Codex for the rebuild.
- Leadership plans to expand AI-driven development across more systems and teams.
Context
Holiday Extras' experience reflects a broader trend among AI-native companies that are moving from annual or quarterly release cycles to continuous delivery, enabled by AI coding assistants like Codex. This shift is particularly visible in consumer-facing travel and booking platforms where user expectations for personalization and responsiveness are high.
The travel industry has been an early adopter of AI coding tools, with companies like Booking.com and Expedia also reporting faster feature development and improved conversion metrics after integrating AI-assisted workflows. These tools are reducing the gap between product vision and execution, allowing teams to test hypotheses in near real time.
For startups, the Holiday Extras case illustrates how AI coding models can democratize software development, letting non-engineers contribute directly to product improvements while engineers focus on architecture, security, and scalability.