Innovation Roadshow - Flyr Labs

Transcript

Matt Brown:It's a pleasure to be here at CAPA, to be a sponsor this year. I get the great pleasure over the next 10 minutes to just tell you a little bit about FLYR, our company, our vision, and most importantly, the work that we're doing with customers, airlines globally. A couple of things first. We're FLYR Labs, not FLYR or Fleur Airlines. So you're not going to hear, hear an airline story here. We are a company of about 250 now data scientists that are structured across the globe supporting customers with some of the technology that you, that you just witnessed on the video. So today I'm just going to walk you through a little bit of our vision and our technology and how we work with customers. To give you some context about FLYR. First, FLYR was founded roughly 9 years ago, and in that, in that process, we sort of recognized that the methodologies used today across a lot of different commercial decisions rely on past performance. They effectively ask commercial leaders to look in the rearview mirror to predict the future. And as we all know, over the last 3 years, that, that entire model has proven to be insufficient at best. And so our company was founded with really a key purpose, and that purpose was to take some of the most advanced technology used across other industries and effectively unlock it to help leaders like yourselves in the room make more informed, smarter, more optimized commercial decisions using a core set of technology. Key to that for us was artificial intelligence. Now, AI is certainly a buzzword, right? We hear it in almost every corner of industry that exists today. But when we think about AI in the context of true application, right, how is it being applied to effectively drive new business outcomes? There's very, very few examples, particularly very few examples in the aviation space. The previous panel before me all talked about revenue, revenue, revenue, right? It was unanimous up here on stage, and they're not alone. McKinsey's Global Institute believes that AI unlocks more value. It creates a potential for another half trillion dollars of value in the travel and transportation space. And what's even more interesting about that is 90% of that $500 million in value McKinsey believes comes in the form of sales and marketing. All of the issues that we're talking about today, can they be impacted and dramatically improve with the application of this technology? We certainly think so, and, and here's why. Because we rely on humans, some of the smartest people running some of the best airlines in the world have to make decisions. We have to make assumptions. Those assumptions are typically steeped in experience that drives specific decisions every single day. And by golly, we're doing the best we can, right? The interesting thing about AI, unlike other technologies that are implemented and then left alone with incremental improvement is AI is always learning. It's like when you train a new pilot or you train a new employee or a new revenue management analyst. There's that learning curve that takes quite a while and a whole bunch of experiences that we as humans have to understand to be able to apply in the future. AI is brilliant in the sense that it's constantly learning and it's learning so much faster than we as humans can possibly learn. And it's also capable of processing tremendously large volumes of data. And we all know that the aviation space is full of data, right? So we believe that AI is the key to unlocking this winner-take-all advantage for our airline partners. It took me a while. I'm not a data scientist. I'm in fact a pilot, so that the two could not be further apart, right? Um, but this helps me a little bit, and I hope it helps you think about the context of what we mean when we say AI. AI's been around for a long time, since the '60s. It's evolved to what's called machine learning, where machine learning models are capable of producing outputs that drive specific outcomes depending on what inputs are, are placed in it. And then more recently, a subset of that machine learning called deep learning has been applied, and that deep learning context is what uses a network of information to constantly evolve and arrive at much better decisions. We believe at FLYR that deep learning is really the key to helping our airline partners arrive at these decisions. So think of it like this. When you're, when you're driving or you're going from point A to point B, as a human, we have a tremendous amount of options For how we get from point A to point B, particularly in an urban setting. And so we have to understand how our different decisions— go right, go left, right, do a U-turn— impact the ability to get to that destination, probably in the shortest amount of time and without, you know, damaging your car or somebody else's vehicle, right? Deep learning is applied rapidly today to optimize that path. And it's that same technology that's used in navigation and even self-driving cars that we're applying today in our cloud-based solution to serve up forecasting and revenue optimization tools to our airline customers. And we're betting on it. We are all in on deep learning technology. And we feel like we're in pretty good company because across industry we see global leaders who are heavily invested in this technology to improve and unlock value in areas that quite simply we've never been able to solve before. So how do we do it? Well, this is just a really fancy eye chart to show you that we, we effectively are training and constantly training and improving machine learning models to generate context out of data. So we don't just look at the last year's performance or the last week's performance. To give us an idea of the forecast of the booking curve for the next 180 days in a market. We're looking at context and real-time data, data sources that have never been considered before when we're looking at, at forecasting and optimizing price points along the booking curve. And what that allows us to do is to create through these neural networks unique context that nobody else is able to provide our customers. And really what that has to translate into is business results. We, we've coined our system the Cirrus Revenue Operating System because it's not just an application, it's a platform that's 100% cloud-based. It ingests terabytes of data. That data is constantly managed and, and formed to make decisions on a real-time continuous basis in the cloud. There's no heavy infrastructure to implement the technology, and it creates really ultra-accurate forecasts of load and revenue that allow us to make tremendously unique, uh, and impactful commercial decisions across the airline commercial functions. At least this is what we're finding. It's really a function of taking data, like I said, any data that we can, applying the deep learning, machine learning models, to create ultra-accurate forecasting. That forecasting allows us to inform pricing throughout the booking, uh, curve. And then of course create a tremendous amount of end-to-end intelligence capability that our users and our customers are now taking to make other decisions across the airline. For us though, it doesn't stop at simply pricing a fare because if you now have a platform that looks at all the data necessary to inform the pricing of a seat. If you start to then consider the trade-offs of seat pricing and ancillary pricing to optimize for total revenue and push that through to the offer state, you've got a really compelling solution that leverages the same set of AI technology under one single pane of glass, giving you sort of a single source of truth for the activity and the pricing optimization for your airline. That's exactly the platform that FLYR has built and is in use with customers around the globe today of multiple shapes and sizes with respect to airlines. So don't just take my word for it though, 'cause it's really good to talk about this stuff. It's really good to have hypotheticals, but it's even more important to actually have results. FLYR is now managing about $15 billion of revenue literally thousands of markets on behalf of our airline customers with the revenue operating system. The results over the last couple of years, even during pandemic times, and frankly prior to, prior to the pandemic, in nearly every instance we average 5 to 7% revenue uplift against existing tools. So whether you have a legacy provider or you are managing revenue on your own, The markets indicate a significant advantage using the technology. That translates to about a 4-point load uplift. Really, really compelling is the fact that on a mean average percentage error basis, our forecast error is reduced by up to 10x, and that's across almost the complete booking curve, not just 7 days prior. And even more compelling, I think, is the fact that The new technology and methods give the analysts or the users of the tool an ability to start to learn and trust the system, resulting in the fact that they are reducing the amount of times they influence or override the system by up to 90% in most of our use cases. Here's the data. We don't just do this through, through theory. We actually compare our performance. We've compared our performance against all the major providers of revenue management systems. And what you'll see in this is sort of test set A, that'll be to your left, and then our test set on the right against a baseline and revenue over a series, a period of time. Sorry, it's a little bit blurry on this screen here, but that's the 5 to 7% revenue uplift. And then from forecast perspective, you can see sort of legacy across the top. The days out of the booking curve along the x-axis at the bottom. And you can see that forecast error is dramatically reduced as well, 5 to 10x. We also believe, and we're seeing with our customers in use cases where now that you can forecast loads and revenue with some more accuracy, the ability to take those and translate them to other commercial decisions across other commercial operations. Network planning, scheduling, air cargo, marketing becomes really, really valuable. I'm gonna get the hook off the stage here in about 15 seconds, but as somebody who is responsible for commercial growth, I'd be remiss if I didn't tell you a little bit about our commercial model. First and foremost, we do not charge any airline customer of ours a single penny until we produce value. So just by having a system, unless that system performs better, much better than the existing set of tools, we don't get paid. We don't charge our airline customers to test it. We pay for all implementation costs, including any additional resource costs for those customers. So it's quite a compelling offer. It's a 12 to 16 month implementation. Since it's cloud-based, all of our users get incremental updates weekly into the system as the community improves. the software process. And we believe that that process is much more of a partner than vendor. It's not about providing you with technology, wiping our hands, and wishing you the best. It really is an ongoing evolution. With that, we'd love to talk to you. In 2021, the company raised $150 million in financing alone, with more coming this year that we'll be excited to announce soon. And we'd love to share more of our customer stories with you. So thank you again for the opportunity to be here.

Copyright policy: All transcripts on this site are the copyright of CAPA - Centre for Aviation. Our reproduction policy is as follows: you may quote up to 400 words of any transcript on the condition that you attribute the transcript to CAPA - Centre for Aviation and link to the original video page. All other use is prohibited. While we aim for 100% accuracy in the transcript, there may be some minor transcribing errors.