Sylvain Roy has spent most of his career in technology roles at Amadeus, joining as a contractor in 2003 and rising to senior vice president of technology for Platform Engineering in 2019, before becoming Chief Technology Officer in 2023.
Roy spoke to PhocusWire about the company’s technology priorities, its move to the cloud, and how it is using generative artificial intelligence both internally and externally.
The interview has been edited for brevity and clarity.
Given your role within Amadeus, what are your priorities for the next six months?
Moving to cloud, no question about it. It’s a big program and it’s transformational. Different people think differently about moving to cloud. It’s a case of quickly lifting and shifting everything to cloud, some companies just move their applications and don’t re-architect them. That way they don’t get much out of the cloud, but that’s not what we’re doing. It’s a transformation in terms of platform and ways of working. It’s not going to be done in the next six months, but it’s my number one priority for the next six months. We’re making good progress.
Generative artificial intelligence is the second thing we’re working on, and we believe this technology will impact the world, the market and Amadeus in two ways: firstly, internally, and secondly, how this technology will take our product to the next level.
In my job, there’s always a third priority, which is all about system stability and making sure that we’re delivering the level of service that our customers expect. This work never ends. If you ask me the same question five years from now, it’s still a priority.
How do you set priorities?
We are not adopting technology for technology’s sake, it’s about what is the business need and what do we need to do to better serve our travelers and our customers?
Get digital travel tips delivered to your inbox every day
Subscribe to our newsletter below
The other aspect is around technology. Technology is evolving faster and faster, and there’s more and more stuff coming out. We have less time to really choose the right technology. Our technology strategy is informed by two things: internally, what the business and customer needs are, and externally, what’s happening in terms of technology. What can we leverage, and which technologies are less important? We have to find a way between the two.
The move to cloud is a big part of it. All the new capabilities we provide to our customers, for example the ability to deploy our software anywhere in the world. Cost optimization is a business need. When you go to the next level, there’s another need: system stability. So I truly believe that our technology strategy serves our business.
So, given those priorities, what do you see as the biggest challenge?
The cloud migration is the largest program, the largest IT program in our organization right now. Every application in our system right now is either on the cloud or is involved in the migration process. So, just the scale of it is the biggest challenge because we have a lot of developers around the world working on this. It’s a transformation and it’s a big challenge to make the transformation successful. There’s a technology side to this, the platform that we’re building with Microsoft, but the other part is the operating model. There’s a change in responsibilities that are empowering developers.
Let’s talk about some of the interesting use cases for AI both internally and externally, and where you see future developments heading.
Internally, we’ve already deployed GitHub Copilot for our developers, and (Microsoft) Office Copilot across the organization, which means we’re leveraging AI to do code generation to generate better code. That’s working really well. And we’re also deploying our own large-scale language models to improve access to knowledge within the company.
Another thing we’re doing internally is using generative AI for testing. We want to generate tests for ourselves and for our business, so we can better adapt our systems, and when something goes wrong, we want to figure out what the problem is. Generative AI is here to help us do that.
What do you think is going to happen next? I think what we’ve learned about agent design is that the idea is to take a generic model and then specialize it. So instead of talking to one generic model that does everything, you have agents that are generic models that are specialized for specific topics, and they all talk together in unison. So the future is probably going to be this model where you can do more. Not because one agent is going to be smarter, but because if you want to deploy software, you have one agent for the software, one for testing, one for deployment, and they’re all going to work together. I think the whole industry should move in that direction.
Moving to the outside world, there are several dimensions. The first dimension is that generative AI generates content. The second dimension is that it’s a conversational interface, almost a new channel. The third dimension is all the problems that AI can solve, the solutions that AI can find for those tedious tasks.
If you think about the first aspect, think about shopping. We provide websites for airlines. We need to provide landing pages, and those pages are generated as we look for inspiration. Airlines need content for that, but it’s hard to get that content, and hard to build those pages with content. We can do that for them in a completely iterative way.
You may have learned about agent design that takes generic models and specializes them. So instead of interacting with one generic model that does everything, you have agents that are generic models that specialize in specific topics, and they all interact together.
Sylvain Roy – Amadeus
In conversational interfaces, Cytric Easy has generative AI models that provide corporate travelers with a very natural interface in Microsoft Teams. Essentially, the traveler is talking to a virtual travel agent who answers all the questions the traveler asks and makes the booking on their behalf.
The third is digital assistants. Our products are complex. Some of them are complex for users to use because the problems they solve are complex. How about a digital assistant that gives you insights into what you’re doing now, insights into the data and results of what you’ve done in the past few years, and helps you set the two in the current year? We believe this can make many of our solutions more accessible.
The industry keeps pushing the mantras: as travel retailers, we want to create connected travel, we want to create the perfect trip. We talk about being customer-centric, but we don’t actually do that. How far are we from a better trip?
Clearly there is a lot to do, and much of the complexity comes from the number of actors. No single actor can solve the problem.
If you look at the whole journey, you’re in contact with travel agents, airlines, airports, hotels. Then you have other players like vendors, and then you have local governments and regulations. So it gets incredibly complicated because no one player can make it better. I truly believe that Amadeus is one of those companies that has a role to play because they’re at so many touchpoints. If you’re shopping (for travel) on a website, it’s likely Amadeus. Between the airline and the travel agent website, it’s going to be an Amadeus touchpoint or right behind it. When you book, Amadeus is probably in charge of that, too. Then you check in at the airport, you board, and all of that is done by Amadeus, so they have a role to play throughout the whole journey.
How long will it take to actually improve this? I think we’re seeing more and more ability to really align the sector. From a technology perspective, what we’re doing is introducing event-driven architecture. Right now, we have a huge service-oriented architecture, we have thousands of microservices. We’re complementing that with an event-driven architecture.
Why? Because events allow you to monitor the entire journey and react in real-time. Unlike services where two applications call each other, events flow into the system and make your applications aware of what’s happening. This allows you to track travelers and have a more responsive and supportive application in real-time.
