Holiday Extras hands software building to designers, accountants and its CEO

Anyone who has filed a ticket with an engineering team knows how the request queue works. Someone close to the customer spots a problem, writes it up and waits while it competes with every other ticket for a slice of engineering time. The person who understands the problem best is rarely the person allowed to fix it, and the fix arrives months later, if at all. At UK travel company Holiday Extras, which helps customers search, book and manage trips, that gap is starting to look like a choice rather than a law of nature.
Chief executive Matthew Pack wants the business to become an AI-based travel technology company. He describes that as "giving the people who understand customers and operations the ability to improve them directly." The order of steps matters. Company-wide ChatGPT Enterprise access came first, and Codex then extended that access into building software. Pack builds with Codex himself to understand what his colleagues need. "This isn't about one or two people anymore," he said. A chief executive who has hit the same walls as his staff is far better placed to remove them than one reading a dashboard about adoption.
What changes for the people involved
The main app used to get one major release a year. With Codex, the team had a fully working rebuild within two weeks, work that previously would have taken over six months. Liam Thompson, VP of AI Innovation and formerly Head of Design, worked alongside engineers to turn his understanding of customers into software, while the engineers concentrated on architecture, APIs, security and workflows. "It's not a question of can I build it anymore," Thompson said. "It's how I would build it." The results are hard to dismiss. Open-to-book conversion rose 32.5%, the app error rate fell from 19% to 9% and the team shipped weekly releases in the first month after the rebuild.
The small stories say as much as the headline numbers. When a customer complained about a missing list view, an agent prepared a fix, Thompson approved it and the customer changed a one-star review to five stars. David Lee, Director of Innovation and Growth, describes an automated system that turns airport-transfer customer chats into tickets so supplier problems surface. An experienced accountant rebuilt finance systems around his own understanding of the business. None of these people were hired as developers.
Permission and guardrails
It would be easy to read this as a story about taking engineers out of the loop. It is closer to the opposite. Pack gives everyone permission to try AI and then actively encourages them, but engineers build the packages, security controls and workflows that make that experimentation safe. The engineering role moves down a layer, from writing every feature to building the platform that lets a designer or an accountant ship changes without breaking things. The tension does not disappear. A human still approved the list view fix, and someone has to decide where approvals sit as the volume of changes grows.
Pack's own reading of the shift is blunt. "Our dreams are not hard enough, clearly," he said. "We need to go back and rethink what we're capable of doing." When a six-month rebuild takes two weeks, capacity stops being the excuse and ambition becomes the constraint.
What this means for L&D
For learning and development leaders, the lesson is that tool access is the easy part. Holiday Extras paired permission with active encouragement, a CEO who builds and an engineering team reorganised around guardrails. Training that teaches prompting without changing who is allowed to ship will stall. The more useful work is helping domain experts frame problems, test fixes with real customers and work with engineers on the safety layer, the kind of capability we build in AI Design Sprints. It also means rethinking how engineers develop, since their value increasingly sits in architecture and controls. And it means asking leaders a harder question than whether staff are using AI: which ideas were shelved because they once looked too slow to build, and are they still?
Sources:
News and insights for innovation, digital transformation, future of work and L&D leaders.
Stay ahead of learning and development, corporate innovation and digital transformation news. Plus the future of work. For leaders in AU, NZ, HK, SG, the US, the UK and Canada.





