Orchestrate all the Things - An AI-powered revenue operating system for aviation and beyond: FLYR Labs Lands $150 Million in Series C Funding. Featuring CEO / Founder Alex Mans

Episode Date: September 22, 2021

A multi-trillion dollar business in crisis, upending incumbents, unfettered ambition, and pragmatic deep learning. FLYR's story has it all. Article published on ZDNet

Welcome to the Orchestrate All The Things podcast. I'm George Amadiotis and we'll be connecting the dots together. A multi-trillion dollar business in crisis, appending incumbents, unfettered ambition and pragmatic deep learning. Flyer's story has it all. I hope you will enjoy the podcast. If you like my work, you can follow Link Data Orchestration on Twitter, LinkedIn and Facebook. So I was born and raised in the Netherlands. I've been writing software since I was about 10 years old. I luckily had parents that recognized my love for computers and they put me in a Montessori school and allowed me to really explore that passion.

I was able to finish with my primary school and high school early. So by the time I was 14, I would either have gone and studied physics or start a company and decided to do the latter. I started a company in network security, built a company until I was 19 to learn at a very young age what it is to start a company, to run a company. I moved to Amsterdam when I was 19, spent a few years in and around Amsterdam, starting to invest in startups. And then when I was 22, I moved to Silicon Valley with the idea that I want to do something on the intersection of two businesses, industries that I was passionate about. Aviation on one side. I'm the guy who writes down the tail number of the aircraft before getting on board. And then on the other side, AI and data. There was this incredible power becoming available in computing through AI and better use of data. And I felt like aviation was a great place to apply it.

Like many people at first, I recognized that prices are all over the place. Why is there so much inefficiency apparent in airline pricing then to bring it to the company the the first iteration of what we tried to do was trying to predict the price of airline tickets and actually arbitrage the airline but by basically trying to recommend to travelers how how when do you buy how do save money? But we realized that actually the opportunity was much greater on the enterprise SaaS side. There have always been only three companies in the world that provide forecasting and pricing technologies to airlines, Sabre, Amadeus, and Pros. And their triopoly has basically resulted in a lack of innovation.

So we recognize that if we're able to use data more effectively, if we're able to deploy deep learning, neural network-based AI for airlines, we can be better at forecasting the future and at pricing their airline tickets so that we can improve the results for the airlines. If we fast forward a little bit to today. What we do for our airline partners and customers is we integrate all of our commercial data into a standardized model. We're a SaaS company, so all of our customers want the same software. We don't rely on heavy professional services that you might find in some of the incumbent vendors. We integrate all of the data,

then we use deep neural networks to forecast the revenues, forecast the demand, and establish the revenue optimal pricing strategy that both creates opportunities for travelers, by being much more reactive to how we price, but also increases the revenue opportunity for the airline in total. And then lastly, we provide a user interface

that kind of bridges the gap between our automation and the analysts and gives them an ability to understand what's going on across the network. So integrate their data, forecast the future, price more intelligently, and then provide an interface that the airline can use to make better decisions, whether that is on pricing or whether that is for figuring out where to add or remove capacity, where to add a flight, where to spend their marketing budget, or just to understand what the quarter is going to look like revenue-wise.

So that's why we call our solution a revenue operating system. We either inform or automate all of these commercial decisions across the airline. And what COVID has done for us, it's made it very clear that airlines need better technology. And we were there at the right time, now helping all these airlines plan for the future in a very different environment from what it was a year ago and that's really accelerating our business culminating into our ability to raise the capital we just raised okay great well actually i got more than i asked for in your introduction because you covered a little bit of everything that i had in in my agenda but well we can go in a bit more depth i guess since we still have more time.

Even if we only had like five minutes, you would have covered everything. But well, luckily, we have more. Sounds good. No worries. Well, okay. So actually, the first question that I had was I was trying to, let's say, wrap my head around what exactly is this revenue operating system because, well, the name, on the one hand, the name implies that, you know, it could well be something like, you know, a SaaS for, I don't know, running your, you know, your revenue streams.

But on the other hand, the fact that, as I already know from your press release, you're employing machine learning and predictive models, you know, it means that it doesn't just do that. And you also actually touched upon a number of things. So the first question I was going to ask you is what kind of data do you integrate and whether actually integration is, well, I know that integrating them is a prerequisite, but I wonder whether you actually use all of the data that you integrate or, I mean, for your predictive models,

or some of them are potentially, you know, just for informational purposes or dashboards or whatnot. Yeah, good question. So I think on the data side, the primary sources of data come from the airline, right? So whether that is their schedule or their inventory or their bookings or their searches or their revenue accounting data

that shows us kind of what are people buying. Most of that data comes from the airline. But we also ingest data about competitor schedules, competitor pricing, events, weather, kind of market capacity, total market capacity. So you've got two sources. There is the airline's data, and then there is supporting external data. And at the end of the day, all these sources that we ingest are feeding into our models.

And this is kind of where the model architecture comes into play. Unlike what traditional regression-based forecasting systems that you've seen historically in this industry, we rely on these deep neural networks. And what they allow us to do, they allow us to always look at all the historical data, but not to try and draw a line from recent past into the future. What we're trying to do is we're trying to understand context. We try to understand that the Wednesday flight from Seattle to Boston is very similar to the Thursday flight from Los Angeles to New York, as an example.

So our architecture is trying to understand context. And once you understand the context across the whole network, you can intelligently subsidize your forecast with data other than for the flight to the market you're trying to forecast. And this is where the secret sauce is, right? We can very accurately forecast future

revenue, future load factor, the revenue optimal pricing strategy, even for flights of markets or markets that have very little data or very noisy data or markets that you never flew before. And that's a superpower that airlines need now more than ever because COVID has kind of turned the industry upside down where the network is different, the competition is different, demand profiles are different, pricing expectations are different. And so between all that data that we kind of ingest and manage and standardize and the architecture of our product

that focuses on context, instead of just drawing out the line, we're able to very, very accurately perform in this very challenging environment. And that's really driving why airlines are switching to our solutions in this market. Okay, well, you mentioned context and you also mentioned, well, deep neural networks. And one of the things that deep neural networks are actually famously bad at, I would say, are actually providing, you know, in a way, providing context and providing justification as to why a certain prediction was made. And I wonder if this is something that you come across. So do you have airline executives, when you go to them with your predictions and telling them, for example, well, you said, I don't know, double your capacity for this flight or whatever.

Do they ask you, well, why? And are you able to answer that question? That's a good point. So I think at the end of the day, what matters most is if our forecast of the future is accurate, right? Because if your forecast of the future is accurate, then it can serve as an accurate baseline, an accurate tracker of how you're trending. is if our forecast of the future is accurate, right? Because if your forecast of the future is accurate,

then it can serve as an accurate baseline, an accurate tracker of how you're trending. So the way you can think about it is one of the reasons why we forecast future demand and revenue is in part to kind of track whether we're trending well. So the way we've built a system is we track our forecast accuracy. We use the forecast as kind of a baseline or a build curve, as it's sometimes referred to. And then we can track our actual performance through time against it. And we actively escalate to the airline, hey, there's

something happening in this market where we're underperforming or exceptionally overperforming the expectation, helping them prioritize what they focus on. And then there, we give them all the tools, all the information, competitors and capacity and pricing and bookings and searches that they can use to dig in deeper. I would argue that given the complexity of an airline network and given that context reaches across

so many variables, it is very difficult, you are correct, in a deep learning system to pinpoint exactly what is driving it. But we give the airline the tools to do the exploration, and we help them focus in the areas where they should focus. So instead of what happened historically, where an analyst kind of goes through that market, tries to find issues and makes interventions based on subjective interpretation, we actually let the system run 95% plus of all the pricing decisions where it is doing well compared to its expectation. And then we escalate potential concerns to the analyst and focus their attention on that 5%. And what this basically means is that we're seeing our airline customers

go from half of all flights being intervened on, meaning a rule-based control that intervenes on pricing, to just 5% or less, right? Letting the system manage the rest. So we found this kind of symbiosis between the automation and the analyst through the forecast, through the confidence in the forecast and by escalating the exceptions against it. Yeah, well that makes sense. I mean that's a broader industry trend that I would say that I see that people don't always necessarily you know want to automate everything. They're kind of trying to apply a recipe similar to what you described. So kind of escalate what needs to be escalated and just leave the rest, you know, to be taken care of by the system, basically.

You also mentioned in your introduction, well, the obvious, you know, the elephant in the room, let's say. So COVID and how it has disrupted the industry. And that's something I was also wondering about, again, in relation to your specific methods of predicting, let's say, that underlies the revenue operating system. So one of the things that, again, machine learning in general, not just deep learning,

is not really so good at is, well, kind of, you know, obviously predicting this, precisely this type of disruptive event for which you know there is no precedent and you have no signals and so on so i was wondering how well were you able to fare in that respect and well how did it turn out for you uh market wise i guess probably i kind of know at least part of the answer it can be too bad because otherwise you wouldn't be racing but i'm just wondering yes so um yes the environment has completely changed at the end of the day what we measure ourselves against and what the airline measures ourselves against is how does the flyer solution

perform against what i used to have with pro saber and m&s and when we launch with a customer we actually run effectively an ab test right There are three ways by which we can prove that we are doing a better job. The first one is we can objectively compare our forecasts to reality. We can compare how much better they are. Second, we can observe how the system is pricing. Think of it as like paper trading.

We can see what decisions we're making and what decision the legacy system is making, and we can compare. The analyst can look at that and say, that's a good decision, that's a a bad decision and we can do those comparisons even before we launch for the customer and then the third one is a true ab test we simply run 20 30 40 percent of all their flights the legacy solution continues managing the rest and we do an effective ab test i'm simplifying it here it's's very complex, but we do an effective A-B test to measure revenue uplift, load uplift, et cetera. And at the end of the day, what matters is how

much we outperform, right? How much we outperform. And because airlines are trying to maximize their revenue, reduce their costs, increase operational efficiency. And across all those measures, we are outperforming all these legacy solutions. So if you look at our commercial model, we actually go to the airline and we say, hey, we will implement in 12 to 14 weeks. We'll do it for free. We won't charge you a penny until we prove this outperformance. And when we've proven it, we'll take a couple of percent of the uplift. So for the airline, there's no upfront cost. They can see the results before they have to commit and when they commit they're only paying a fraction of the gain and then they have a huge roi on working with us incrementally um uh to what they had before okay well i would say that's that's a good way to

well to get your your foot in the door in a way so uh and you you mentioned customers a few times already. So I was going to ask you if you can actually share some names, if that's possible at all. The only public name I can mention to you is Air New Zealand, because they've done some press with us. Typically we're fairly like, at least right now, we're still fairly conscious of not releasing customer names, but we expect to

all vote about one customer per month as of january and we're exiting this year with about 10 or so yeah okay and these are all major airlines like the biggest of the biggest so what we're looking to be managing about 14 billion dollars in revenue on behalf of airlines, right? So setting the prices for annualized $14 billion in revenue by the end of this year. Okay. Well, yeah, I mean, I guess the numbers and the business model has to look convincing because you just got quite a substantial fundraise, actually.

And that's kind of the occasion why we're having this conversation. So I'm going to ask you to just share a few words about the fundraising process. Who's in it? What's the idea, actually? Why are you doing that? I mean, if you're going that well,

what are you going to use that cash for? Couldn't you just do it on your own as you have been doing so far? Yep. So right now as a company, we are doubling our staff every six months. And as you grow as a startup going through a growth phase, the biggest challenge, the biggest risk is that growth.

How do you not break the system? How do you not break the foundation of the business and the team? So one of the areas that are really important for a series C is bringing on board the right investors. You've seen this in the release. We put on board Lauren Stosie from from West gap, right? Great operator, amazing operators on their team. People like Jeffrey Katzenberg, uh, and, and Sujay Jaswal from, um, uh, WonderCo also great operators.

And then we've ordered people around the business as well. So we're really trying to surround ourselves with the right people to make those decisions and help us scale that business, scale our business. That's the first piece. In terms of allocation of capital, it's going to a couple of places. Our demand is outstripping our supply right now. So we need to scale up our delivery capacity so that we can meet the demand of all these airline customers and beyond. Second, we are actively investing very heavily in our product so that we can serve the broader airline organization. So how can we make sure we can best serve decisions, not just in revenue management and pricing, but also in planning, marketing, leadership, cargo planning, ancillary pricing, you name it.

So second is product. The third one is really around M&A. There's a lot of opportunity for us right now to go faster and provide capability to airlines by acquiring and bolting on other companies. Because we have this data infrastructure. We have the system integrations. We've already got the relationships. So we can actually acquire companies through M&A and kind of scale much faster by taking those products to market with our existing customers and integrations.

And then the last piece I would also raise is, in addition to airlines asking us to come to their rescue during COVID, we also got a lot of inbound interest from auto travel and transport verticals, rental car, cruise lines, railway companies, freight companies, hospitality, events, you name it that's similar challenges and the data is similar the problem is similar so one thing we're investing in starting now is we're spinning up effectively the initial efforts around taking our core technology and deploying it into those verticals it's still a SaaS platform it's still the same core technology but we're tweaking it to fit the needs of those industries. Because our vision is to provide all these insights and automation for all of travel and transportation, not just

airlines. And that's kind of, I would say, the roadshow narrative that we're building and that we're proving out over the next few years. And this fundraise enables us to do and invest in all those areas well yeah i mean having heard that yes i mean you certainly do not like ambition so i would say that you know for this kind of ambition maybe you know that fundraise is probably even small you you'll you'll need another one pretty soon if things go your way we'll give you a call as soon as we do. Cool. All right. And I think you also kind of touched upon, you know, where the money is going.

And I was wondering more specifically about, you know, the departments, the areas in your company. So is it going, you know, to engineering or marketing or, I don't know, operations? Yeah. Yeah. or marketing or other operations? Yeah, so the beauty of our solution is, and maybe Matt can speak to this, but about 75% of our company staff is engineering and data science, right? So we're technology first. We work with our customers.

We work very closely with our customers. But it's technology and the performance outcome that drives the result. And that eventually builds a bigger and bigger moat compared to either legacy incumbents or new entrants. So we're investing very heavily in the technology piece. Technology, engineering, data science is split between the teams that work with our customers to deploy and launch our technology and the teams that build up the products.

So I would say at a high level, about a quarter of the company is our delivery and implementation teams that work with our customers. About half of the company is product engineering and development. And about a quarter of the company is sales, marketing, people in operations, analysts, M&A, things like that. Okay. Okay, cool. Right, and then the other thing,

the other area, let's say, that I was wondering about, and again, you kind of touched upon it at least partially, was whether you see yourself expanding to other domains, which you already answered. So I guess the follow-up question to that is, well, how easy do you think that's going to be and it also touches upon i guess well in terms of your technical infrastructure i guess there's like two main components let's say one is like the your data ingestion pipelines and these you will

you will have to adapt either your specific data sources and then there's the actual predictive models and I wonder how much you know adjustment let's say you will have to do to those and I also wonder by the way I don't think you mentioned whether you have any IP on those like patents or anything. Well so I think the the way let me speak to the way we're going to enter neighboring bases, neighboring verticals. So similar to how we did with airlines, the first thing we need is you need a partner customer that is willing to give access to the behind the scenes. What are the operational needs? What is the data? Where does the data sit?

So the very first thing we'll do when we enter a neighboring space is we sign a strategic agreement with such a partner. So whether that is rental car, whether that is cruise lines, or whether that is railway, we will start, and we are starting, we can't announce these yet, but we are starting with one or two partners in each of those spaces. So these are basically early adopter customers who give us access to all the knowledge,

the data, and allow our teams to port the technology into their space. So that's kind of critical step number one. We are happy to co-invest in kind of getting things deployed because we realize that, you know, they're taking an early adopter bet on us. So we're equally making an investment into them to make it successful.

That's kind of step one. The second piece that's really important is investment in the neighboring spaces while your core business is growing cannot cause a distraction to the core organization so we're very very conscious about setting up the organization in a way where these neighboring verticals that we're going to enter are not going to distract or income like reduce the opportunity on the core airline business where we're growing very very

quickly so that's kind of what we're all very conscious of right now. I think the last piece is really important is if I look at the company in four or five years, we would love to build a public company. We believe that the public markets are ready for a company like us in a couple of years to provide a counterbalance to the few public companies like Sabre, Pros, and Amadeus that have been out there for decades. In order to get there, we need to make sure that we correctly diversify the business, which is exactly what we're doing, by taking our technology into all these neighboring

travel and transportation verticals, where we're going to become the biggest across data, forecasting, pricing, and business intelligence. That's really what we're trying to do. And we're doing it together with our partners instead of forcing it onto them. Okay. And well, to kind of go full circle in a way, because we started the conversation

and you mentioned COVID and what a big influence it had on your trajectory, let's say. Let's go full circle and wrap up with COVID again in the term in the sense of how do you see that you know influencing the the industry at large going forward so would you say that maybe you know that there's going to be a scaled down industry but the ones that remain are going to be more efficient sorry I think you broke up the first game thank you we face a question sorry yes yeah I was wondering what you think the long mid to long term Sorry, I think you broke up there first. Can you rephrase the question? Sorry, re-ask the question.

Yeah, I was wondering what you think the mid to long-term industry effect of COVID is going to be on the industry. Maybe there's going to be fewer players, but more effective. So I think on one hand, COVID has enabled airlines to take a good look at their operation and restructure their cost base.

So I think in the long-term, airlines will, they've removed look at their operation and restructure their cost base. So I think in the long term airlines will if removed bloat in their organization, if any on the cost side, I think that that's a good long term change. Even as they rehire their staff and the crews and their fleet. The industry is always going to grow if you look at the last decade over airline hospitality or anything in travel, travel and transportation will always be a growing industry. So we will be back at 2018, 2019 numbers in the next three years. Some markets in the U.S. already hit some TSA-based milestones this year.

They're going to travel on any given day or a particular day. So if there's a capacity and volume, the market's going to recover. The companies running it will be more efficient if I ignore the fact that they've had to take on a lot of debt, right, to finance the business. So it is possible that the efficiency they've gained on the operational side now goes to like interest payments and principal payments on the debts. I have no visibility in that, but that's an expectation. But by dealing with a growing industry,

like the world is getting more connected every single day, not just online, but also offline. People want to travel. There's a lot of pent up demand. So we are looking at our customers as a five to 10 year or beyond relationship. And we're very confident that all of those customers will be far greater in five years than they were pre-COVID.

So overall, the industry will continue to increase. I think if you look at travel and transportation, including freight, effectively, it's a $15 trillion global industry. That's the industry we serve. And especially in a world where our commercial model is tied to lifting their revenue and taking a small piece of it, the opportunity for our business is huge. We consider our total addressable marketing at tens of billions of dollars,

much of which today doesn't have a vendor that meets the need. Well, I would actually take that last part a little bit um you know further or actually you know take the the contrarian view if you will like okay sure the world is getting more connected but you know on the on the other hand this whole you know couple of years let's say has also shown to people that well you can actually have business meetings without traveling and you know you can do lots of things without traveling and plus you have you know this you know climate crisis looming and that's going to have an effect on traveling as well so it may go your way or it may i would say you i think ready player one hasn't quite happened yet so you can't yet go on to vacation on zoom so i think i think when it

comes to leisure, that's going to come back and has already been coming back really fast. So I think I'm not worried there at all. You are correct. Business travel will look different. It depends a little bit on the type of business, a little bit on the distance. So business travel, domestic, for example,

trans, con, US, like short, medium haul business travel is coming back or has come back really fast. It is the-haul business travel is coming back or has come back really fast, right? It is the long-haul business travel and the ultra-long-haul leisure travel that is still TBD. So I think that's... But as a percentage of total passenger volumes

in the industry, that is not that much of a percentage. Yeah. Okay. Well, either way, you know, what you seem... does indeed either way, you know, what you seem does indeed seems to,

you know, raise the efficiency of whatever players remain. So, overall, I think it's a

good thing. My only, you know, doubt is, you know, whether there's actually going to be that many

players or whether the pie is going to be that big, but that remains to be seen.

And I guess, you know, your pitch was good enough for your investors as well, at least. Again, I'll go back to like,

we're going after a $15 trillion market. I don't think, like, we're not concerned about the size of a $15 trillion global travel and transportation market at all. Thank you. I hope you enjoyed the podcast. If you like my work, you can follow Link Data Orchestration

on Twitter, LinkedIn, and Facebook.