As part of my New Year’s resolutions, I’ve decided to share more about the AI work I’ve been doing 🚀

As a design leader working on the world’s largest travel website, I’ve spent the past few years immersed in the fascinating and ever-evolving world of AI. Working at the intersection of UX and AI has offered countless opportunities to rethink how we approach design, shifting the focus from building flashy new features to creating meaningful, intuitive user experiences powered by this cutting-edge technology.

But here’s the challenge: AI often feels complex, even overwhelming, and it’s not always clear how it delivers real value. Too many products are designed around AI — emphasising the technology rather than integrating AI seamlessly into the features users already rely on. This has led to some impressive innovations but also plenty of missed opportunities to make AI feel simple, natural, and human.

Through my journey of leading AI-powered design projects, I’ve come away with some key insights that have shaped my approach to UX and AI. Here are five of the most important lessons I’ve learned along the way:

1. Keep it simple, stupid

The primary goal of AI in user experience should be to enhance, not complicate. Too often, users are expected to become “prompt engineers,” learning how to interact with AI systems through trial and error. This places an unnecessary burden on the user and risks alienating them from the experience altogether.

Great design integrates AI invisibly. Users shouldn’t need to know that AI is working behind the scenes; they should simply notice that the experience is smoother, faster, and more personalised. For example, in projects I’ve worked on like Smart Filters at Booking.com, the AI operates quietly in the background, translating user requests into actionable filters. The result is a seamless, intuitive search process where users feel empowered to express their needs naturally, without navigating a steep learning curve.

2. Ask why, why, why, why, why

In the race to innovate, it’s tempting to start with the technology and try to build something impressive around it. However, this approach often leads to solutions that feel disconnected from real user needs. AI features that exist just to “do something with AI” often lack the clarity and purpose that make them truly useful.

The key is to start with a deep understanding of the problem. What pain points do users face? What needs aren’t being met? Once these questions are answered, AI can become a tool to address those specific needs. In my experience, this approach ensures that AI serves a clear purpose, enhancing the user’s experience in ways that feel intuitive and valuable.

For example, when designing the AI Trip Planner, we focused on simplifying the often-overwhelming task of travel planning. By understanding how users search and explore options, we were able to design an AI-driven experience that felt more like a personal travel assistant than just another search tool.

3. Test it like you mean it

AI introduces a layer of unpredictability that makes user testing more important than ever. Unlike static designs, AI systems can produce different results for different users, meaning the potential for blind spots is much greater.

User testing is crucial to uncovering these blind spots and understanding how AI features perform in real-world scenarios. It’s not just about ensuring functionality; it’s about observing how users interact with the system and identifying areas for improvement. Iteration based on these insights is essential to refining the experience and making it feel polished and intuitive.

For instance, during the development of Smart Filters, we discovered that while the AI was excellent at understanding specific queries, users sometimes struggled with phrasing their requests. This insight led us to refine the user interface and provide subtle guidance, improving the overall experience without sacrificing its simplicity.

4. AI is a teammate, not a tool

When done right, AI feels less like a tool you have to operate and more like a teammate quietly supporting you in the background. This distinction is critical for designing experiences that feel natural and human.

A great example of this is in how AI-powered assistants suggest relevant content, streamline workflows, or proactively offer solutions before the user even realises they need them. These “teammate” moments build trust and make users feel supported rather than burdened by the technology.

Designing AI to behave like a helpful teammate requires a shift in perspective. It’s not about showing off what the AI can do; it’s about meeting users where they are and helping them achieve their goals effortlessly. In every project, I aim to design interactions where users feel the AI is working for them, not demanding their attention.

5. It takes a village to build good AI

Designing AI features is an inherently collaborative process, often more so than traditional product development. It requires input from a wide range of disciplines — from UX writers and researchers to data scientists and machine learning engineers. Each team brings a unique perspective that shapes the final experience.

This cross-disciplinary collaboration is essential because AI systems are rarely linear. The back-and-forth nature of refining algorithms, testing designs, and balancing technical feasibility means that communication and alignment are critical.

Where we go from here

Building AI products has taught me that the most impactful innovation often happens in quiet ways. While the tech world buzzes about the latest AI capabilities, the real work is in the subtle details — how a search result anticipates a user’s next question, how a recommendation feels personally relevant rather than algorithmic, how complex technology fades into the background of a natural user experience.

The travel industry gives us a unique lens on this challenge, but it applies to any industry. Users come to your platform not to experience AI, but to achieve a certain goal. Our job isn’t to build impressive technology — it’s to understand the needs of our users deeply enough that we can make technology truly helpful. Sometimes, the smartest solution is the simplest one, and the best AI can be the one users never notice.