Industry

AI-generated text

Holiday Extras rebuilt its app with Codex, boosting conversion and halving errors

Holiday Extras used OpenAI’s Codex alongside company-wide ChatGPT Enterprise access to accelerate app development, enabling a full rebuild in two weeks instead of six months.

Holiday Extras rebuilt its app with Codex, boosting conversion and halving errors

Holiday Extras used OpenAI’s Codex together with company-wide access to ChatGPT Enterprise to change how it builds customer-facing and internal software, enabling staff with operational and customer expertise to act directly on product improvements.

Leadership, tools and approach

  • CEO Matthew Pack says becoming an AI-driven travel technology company required organisational change: employees needed permission and the tools to directly improve products.
  • Holiday Extras provided enterprise access to ChatGPT and introduced Codex to support code generation and product build. OpenAI models power personalised app features, while a component called "dots" provides coordination and insight.

Rebuilding the main app and development speed

The company had spent years developing its mobile apps and previously timed major changes to an annual release. Using Codex, the team rethought both the user experience and the way they build software:

  • The team produced a fully working rebuild in two weeks — a pace they say would previously have taken more than six months.
  • Liam Thompson, VP of AI Innovation and former Head of Design, worked alongside engineers to translate his customer knowledge into software solutions.
  • Development was organised into small, documented packages that created reusable foundations, while engineers concentrated on architecture, APIs, security and development workflows.

Product and technical outcomes

The rebuilt app delivered a more personalised experience by leveraging users’ bookings, trips and history. Codex allowed iOS and Android features to be developed in parallel so Android did not lag behind iOS.

Key metrics reported:

  • Open-to-book conversion on the main app increased by 32.5% after the rebuild.
  • The main app’s error rate dropped from 19% to 9%.
  • Release cadence moved from an annual major release to weekly releases in the first month after the rebuild, with each release including multiple new features.
  • Countdown — a product within the business — shifted from adding roughly one feature every six months to multiple additions per week.

Customer feedback feeding development

Faster development made it possible for customer feedback to be actioned quickly. For example, when a Countdown customer complained about a missing list view, an agent prepared a fix that Liam Thompson approved; the update was published and the customer changed a one-star review to five stars.

Operational changes

David Lee, Director of Innovation and Growth, described an automated system that turns airport-transfer customer chats into actionable tickets, enabling the team to identify supplier-network issues more quickly. In this way, customer insight becomes work the team can act on.

Culture, governance and practical safeguards

Pack says leadership gave employees permission to try AI and actively encouraged experimentation. That freedom is paired with guardrails: engineers build packages, security controls and workflows to make experimentation practical and safe.

Lee points to an experienced accountant who rebuilt finance systems based on business understanding and colleague feedback, illustrating that the opportunity extends beyond learning to code — operational experience can be applied directly to the tools people use.

Partnership and future plans

Holiday Extras values Codex’s ability to explore ideas, reason through complex problems, propose plans and deliver independently. Thompson notes the tool can suggest alternative approaches during work.

ChatGPT Enterprise supports employees across the business, OpenAI models power personalised app features, and dots adds coordination and reporting — Thompson recalls dots proactively preparing a performance report after an app launch. Lee emphasises model quality and efficiency as well as the relationship behind the products.

Pack’s ambition is to have more Holiday Extras systems administered through AI with employees directing and shaping outcomes. The app rebuild demonstrates the potential: improved booking performance, fewer errors and ongoing capacity for enhancement.

Implications

With more employees able to act on their expertise, Holiday Extras has a new basis for choosing which product and market opportunities to pursue. Pack is asking the leadership team to reconsider the scale of the company’s ambitions as more people can influence product and operations.