Building a global travel tech powerhouse with Denis Lacroix

Pari: I wanted to ask a little bit about your experience. You have been with Amadeus for over 15 years. Can you elaborate on how the unique dynamics of the travel industry has changed? And how have they influenced your journey at Amadeus, particularly the impact of global events and evolving customer expectations?

Denis: I’ve actually been with Amadeus for over 35 years, much longer than you mentioned. I started in computer science and software engineering before joining Amadeus in 1989 when it was still a startup in the travel tech sector.
Amadeus has an interesting business model where we charge customers on a pay-per-use basis – for airlines when a passenger boards, and for travel agencies when a booking is made. This model made us very vulnerable during COVID-19.
When travel ground to a halt in early 2020, our revenues plummeted by 90%. The company was in deep trouble. Fortunately, we had enough cash reserves to keep operating and paying salaries during those dire 6 months. Recovery was slow but steady after that initial half-year.
Today, airline passenger volumes have rebounded to above 2019 levels across all major markets like North America, Europe and India. So while 2020-2021 was an extremely difficult period that threatened the company’s survival, we managed to weather the storm due to our financial strength and the eventual travel resurgence.

Pari: It’s amazing how resilient the company has been going through that tough time with COVID-19. Your resilience to even continue paying employee salaries during that very difficult period is impressive.

Denis: It was very difficult, but we were able to survive on our own financial strength. Interestingly, this crisis forced many industry players to reconsider their plans and business models.
For our commercial airline offerings, the airlines deeply impacted and running their own reservation systems realized they were saddled with fixed costs they couldn’t reduce. This allowed us to actually grow our market share during that period, which was unexpected.
The crisis prompted many potential customers who previously ran their own systems to embrace Amadeus products instead. By switching to our model, their previous fixed costs became variable, allowing them to realize the value of our solutions when their flight operations had stalled. We gained market share across all our businesses as the crisis forced reevaluations.

Pari: It’s interesting that you had a pay-per-use model even before Software-as-a-Service (SaaS) became popular. You were operating that way ahead of the curve.

Denis: Yes, it is quite interesting. These days, major companies like SAP and Adobe are trying to transition from licensed models to pay-per-use or something close to it, since SaaS has become all the rage.
But it’s funny because that’s how we’ve always operated at Amadeus, even before the term “SaaS” existed. We were essentially a SaaS player from the beginning, just without knowing that term.
Amadeus was born as a product company with a pure SaaS model. All the systems used by our customers are hosted by us – there is nothing running locally on their premises. We’ve been delivering software services this way from the start, before it became a widely adopted approach.

Pari: You also have deep experience in building global organizations and teams across multiple countries. Could you share your experience in setting up these global teams?

Denis: It’s very interesting. From its inception, Amadeus had an almost unfair multicultural advantage baked into its DNA. The company was created by four European airlines – Lufthansa, Air France, Iberia, and SAS from Scandinavia. So it was a multicultural setup from day one, not just a French or Spanish firm.
This created a very specific culture of working across nationalities and languages right from the start, even before I joined in the late 80s. We grew accustomed to operating in a multicultural environment early on.

Then for our distribution business to succeed globally, we needed to quickly blanket the world with our capability to aggregate travel content for agencies. This meant learning how to work across more than 150 countries – we joke that we operate in more nations than even McDonald’s!
So being a global, multicultural organization is fundamentally part of our DNA and core strengths. This allowed us to rapidly expand products and services worldwide, which would have been far more difficult as just a single-country player trying to go global later.

We cultivate this multicultural identity. At our Nice headquarters alone, we have over 4,000 employees from 92 nationalities. As a Frenchman myself, we all basically speak “broken English” since few are native speakers. This intrinsic multiculturalism across languages and backgrounds is a key reason for Amadeus’ success.

Pari: It’s interesting how Nice is considered a tourist destination, but you’re running a very diverse technology team there. The company’s multicultural DNA allowed you to seamlessly build global teams, including in India and other locations.
Diving into the hot topic now – how AI is enabling developers to be more productive. Every customer I speak to has huge questions around measuring engineering productivity across teams. We haven’t solved productivity measurement for engineers in 50 years, but there are many questions like how do I compare productivity between more junior teams in India versus slightly more senior teams in Eastern Europe?

Denis: You’re right, the discussion around measuring software engineering productivity has been ongoing since software was invented. A couple of key points:
First, it starts with the caliber of talent you can attract. There is a famous quote that productivity can differ by 10x between engineers, though that may be an exaggeration. But as a high-profile product company, we differentiate ourselves by hiring top talent – much stronger engineers than average.
Interestingly, we haven’t gone so far as properly testing AI tools like GitHub Copilot by having two teams do the same work, with and without the tool, and measuring the difference. That would be expensive and mechanically challenging to set up.

However, we are deploying AI assistants across our engineering base, not just for coding but for tasks like creating test harnesses that engineers often dislike. If it helps them invest more time in proper testing, we’ll gain from higher quality code in the long run.
The productivity gain numbers quoted publicly by the tool vendors themselves need to be taken with a pinch of salt, as they are trying to sell you something. I’ve used Copilot myself for home automation projects in languages like PHP that I hate. It helped me do things I wouldn’t have touched, without making me proficient.

AI also helped explain 300 lines of convoluted PHP code when I didn’t understand what was going on. So there are benefits, but whether it’s 1%, 5% or 10% gain, nobody really knows yet. We’ve made the gamble to adopt it because our 10,000+ engineers weren’t threatened and were pretty cool with it.
Ultimately, the productivity boost remains to be seen. But we think the potential upsides are worth going ahead with AI assistance for our engineering teams.

Pari: One thing we’ve found from benchmarks is that AI is making global software development easier. As you mentioned, it helps understand code written by other teams, which aids in transitioning work between locations. Teams working across different sites seem more confident they can understand each other’s codebases when using AI assistance.

Denis: I completely believe this is the case based on my own experience trying to make sense of thousands of lines of unfamiliar PHP code – the AI was really helpful for understanding and explaining it.
When we opened our R&D center in Bangalore, one of the biggest challenges was knowledge transfer – getting the new team up to speed on codebases owned by other sites like London. At that time, the only way we could manage it was by relocating some engineers temporarily to work side-by-side. It worked but was expensive and slow.
So it’s quite possible, and I believe, that generative AI can significantly help us transition knowledge and codebases between locations much more efficiently than physical relocation. If it avoids that relocation overhead, it would provide a big boost to our ability to work as a distributed global team.

Pari: The travel industry is going through massive changes post-COVID, with rapid growth, new technologies and standards emerging. How do you continue innovating and creating value for customers who also went through tough times?

Denis: It boils down to truly listening and understanding customer needs. A big trend is airlines wanting to behave more like retailers – knowing customers better, selling ancillaries beyond just flights. New industry standards like NDC, IATA Order and Offer are enabling this retail transformation.
Online travel agencies have been moving in this direction longer, but there’s still work to do on personalization. As an elite flyer, I’d expect the airline to provide a differentiated, personalized experience when buying from them. Implementing this level of retail behavior and personalization requires fundamental changes that customers now demand.
Airlines may never be exactly like Amazon, but that’s the direction they want to move towards, which we are supporting through our product evolution.

Pari: An interesting benefit of being in a location like Bangalore is having large retail companies driving global innovation, allowing knowledge sharing from that ecosystem.

Denis: Absolutely, and there are unique innovations happening in India that we don’t see elsewhere yet. Mobile payments like the government’s solution are very powerful, practical, and could be a sign of things to come globally. We’ve had to adapt our systems accordingly.
Having an early window into these types of future trends in India gives us another advantage in anticipating what may spread to other markets. Just like the Chinese super-apps, we think similar mobile payment models will proliferate worldwide.