CASE STUDY
can now delve into channel impact by week, analyse seasonal profiles, understand halo effects and even investigate cannibalisation.
We can also‘ roll the dice’ further, assessing the impact of a competitor’ s campaign on us, and then exploring potential defensive strategies to ensure we don’ t lose potential travelers. This level of conversation was never possible before and wasn’ t even the direct purpose of our initial AI implementation. But the availability of clean, well-structured, and governed data allows us to do so much more.
How do you make sure all that data is securely locked down?
While I’ m not a CISO, I can tell you that information security is embedded at every step. Whenever we build anything, from a small data table to a finished product, our leadership team includes technology experts, especially in information security. We have established technology design principles, agreed upon even before we started looking into a new cloud platform. Every iteration, every build, goes through a rigorous process to ensure its governed and secured, with people only having access to what they need. With platforms like Unity Catalog, data governance, control and management become much easier and more visible. We know who’ s using what and whether they should be. Crucially, we’ ve never had a breach.
In terms of the AI transformation, were there any major hurdles that stood out to you?
By far, the two biggest were building business trust and handling ongoing maintenance and data governance.
Starting with governance, when any organisation moves towards a cloud-first approach with data at its heart, data governance becomes paramount. This encompasses everything from master data management to data stewardship, cataloguing, labelling and taxonomy. The challenge isn’ t technical, but rather ensuring people understand its ongoing importance. It’ s not a“ fire and forget” task; you can’ t just label data once and be done. You need to stay on top of it, especially when data sources refresh or agencies change. If data isn’ t explainable by design – whether for an AI capability or not – it won’ t be built or progressed.
The second challenge was trust. The solution here is to turn your information customers into your partners. One benefit of using AI is that the language of data suddenly becomes accessible through natural language. This significantly lowers the barrier to entry. People who were typically just recipients of reports – dashboards, PowerPoint decks, data tables – are now co-creators. We bring them into our world, into platforms like Databricks, where they can learn and experiment with things they’ ve never been able to do before. Giving them a sense of agency and making them ambassadors for what we’ re doing builds that trust.
The closest thing to an‘ out of the box’ solution here is to give those ambassadors a platform to shine and let them take the credit. It’ s not always easy, as you might feel you put in the effort. But let them take the credit, because it will be more authentic and inspire others. If they can do it, others will think,‘ I can do it too.’ That’ s powerful; it helps bring more people in, making your efforts more trustworthy. Before you know it, you’ ve built a community that’ s eager to get going, and that’ s when you know you’ ve built something truly special.
Have you got any new products or projects on the horizon that you will be bringing within the next few months regarding your AI processes?
There are lots of projects underway. AI, as you can imagine, gives birth to many concepts; some may never see the light of day, but that’ s okay. I love that people are thinking and daring to do what they couldn’ t do before. I can’ t give you a definitive timeline, but the one thing I believe has the most appeal is the ability to have a conversation with your data. I think when that gets cracked, it will be a complete game-changer in how our sector – and any sector, for that matter – can interact with data. It will democratise insight in a way that goes beyond anything we’ ve ever seen. I can’ t wait for that to come.
When you say“ have a conversation with your data,” just clarify it for us?
A typical, very traditional way information is made available to the public or stakeholders is through a dashboard. Now, a dashboard is wonderful; it can look quite good, offering different options and people can effectively choose their own adventure when they start to explore it. However, it’ s limited in terms of what you can do and see, and sometimes that barrier can be a step too far.
Having a conversation with your data simply means that instead of pre-canned visuals, you have a prompt where people type in a question or problem. For example:“ Where do I think travellers are going to come from over the next six months?” or“ What are the key things that travellers from Asia are looking for?” or“ Which attractions were they likely to prefer over the last two months, compared to last year?” Whatever the question may be.
The system will then generate both a set of easy-to-understand visuals and a narrative explaining the insights. The level of data modelling, the schema of abstraction and all the technical specifics are immense. But again, this is where the secondary benefits of developing generative AI products really come into play, because you have that well-structured, well-governed and well-managed data, and it starts to feed that process.
As human beings, we are naturally inquisitive; we ask questions and have been doing so since childhood. People are very good at asking questions, and this connects quite well with how our brains think. So, we’ re flipping the model of insight-driven decision-making to instead become question-led insight.
In our view, this fulfills two purposes: Firstly, it gives people access to insight in a way that works for them – far more granular and personalised than ever before. Secondly, it gives us, based on knowing what people are asking for, valuable insight into what truly matters to them. This, in turn, allows our development pipeline to become much more nuanced. p
16 INTELLIGENTCIO EUROPE www. intelligentcio. com