×

Fill out this quick form for your free webinar

First Name
Last Name
Job Title
Company
Country
State
Phone Number
I agree that PROS may contact me. I can unsubscribe at any time.
By submitting this form, I agree to the storing and processing of personal data by PROS as described in our Terms of Use and Privacy Policies
Thank you!
Error - something went wrong!

AI No Longer Just a Buzzword for Airline Growth

Kevin May: (00:01) Hello there. Good morning, good afternoon, good evening wherever you are, or good day to you if you're listening to this on the replay. Welcome to another PhocusWire webinar. This one is produced with our friends at PROS and it goes under this wonderful title of AI, Artificial Intelligence: No Longer Just a Buzzword for Airline Growth. Us in the media are often guilty of using lots of buzzwords, but what we're going to do over the next 50, 55 minutes or so, is really dive into some key things to how it affects airlines. 

Kevin May: (00:35) A little bit of background first of all. Science and innovation are the foundation to many airline mission critical systems. Airlines are mission critical in more senses of the word, including revenue management. Airlines have been discussing AI for several decades, but only recently let's be fair, are many of them implementing the real life applications for AI to drive value to their business. 

Kevin May: (00:58) Now, this isn't pie in the sky stuff, real world applications are now being used today to optimize revenue and drive efficiency across airlines all around the world. So, what you're going to hear I said, over the next 55 minutes or so is leaders in innovation and R&D at a major airline. You're going to learn how major airlines are using AI and machine learning, and in particular you're going to learn how PROS is incorporating 30 years of science into mission critical airline systems.  

Kevin May: (01:28) So, little bit of housekeeping before we kick off. As always, we are recording everything today so if you have to dip out at any time, don't worry, you'll be sent a link to the replay later on today or first thing in the morning. We have a Q&A at the end. This is essentially rather than a presentation, we've done this one quite differently. It's more of a panel discussion between myself and our three experts who I'll introduce you to in a minute.  

Kevin May: (01:55) But again, if you want to pass this on to your friends, your enemies, your contacts, your mom, your dad, whoever, you'll be sent a link later on. Don't forget to send your questions in before the end, we'll get to those, but let's dive in.  

Kevin May: (02:12) Joining us today, we have myself obviously. I'm the moderator. I'm the Editor in Chief at PhocusWire. I'm here in the UK today. We're joined by Justin Jander, who's the manager for product management at PROS. His colleague Justin Silver, he's a senior scientist. This is a wonderful title I must say. He's at PROS as well. We're also joined, very welcome to Lucio Bustillo. He's the manager for Revenue Management, Science and Innovation. That's an equally brilliant title, at Air Canada.  

Kevin May: (02:42) The two Justins are joining us from Houston and Lucio I was in Montreal, in Canada. Thank you very much to our experts for joining us. Normally I'm fairly formal with these things but as we do have unusually two Justins joining us today, I will be referring them as Justin Jander and Justin Silver. Don't think I've suddenly gone very formal on everybody. Those of you that know me well would know I'm normally not that formal, but I think it's very helpful for today.  

Kevin May: (03:11) First of all, let's just do a little run around to our panel here. Justin Jander first of all, just give us, what is it you do and how does it work, and a little bit about your background first of all.  

Justin Jander: (03:26) Sure. As you said, my name is Justin Jander. I am the manager of Product Management for our revenue management products. You were joking about fun titles, mine is just very redundant it seems. I've been at PROS for about nine years now. I have a master's degree in statistics, which I feel like really lends itself to my role at PROS because of the focus on the data that we have and the way that we work with airlines around that.  

Justin Jander: (03:56) I think one of the things I'd like to just talk a little bit about why I love the airline industry and why I love working in the airline industry. One of the reasons is the diversity of people that we get to work with. There's people from all over the world, all trying to solve the same problem of flying people from A to B, but the way that they approach it and the way that they tackle that really varies.  

Justin Jander:(04:18) As a person in this industry, I get to see the different ways that people handle that and the different backgrounds that they have, and really get to understand how they take their, how they look at the problem and how they address it and we get to help out. As PROS, we get to help out sharing our expertise, and then working with them to implement that into software. I think that that bringing together a software problem against an industry problem and combining those two things together into a really interesting solution.  

Justin Jander: (04:48) The second reason I really love the airline industry is because I really like being around airlines, airports and flying and really anything related to travel. Just a fun little fact, I've tracked every flight that I've taken for the last 10 years, including the distance traveled, the aircraft that I was on, the aircraft type, all of the various details. I'm really as you can tell, probably a little bit too obsessive about the airline industry, but it's something that I really am interested in and it really fuels my interest in solving the problems that we have. That's a little bit about me.  

Kevin May: (05:22) Okay, Justin J, thank you very much. Unbounded enthusiasm, that's great. Lucio in Montreal, tell us a little bit about yourself, your background and what you do at Air Canada. 

Lucio Bustillo: (05:31) Okay. My name is Lucio Bustillo. I am a manager for Revenue Management, Science and Innovation at Air Canada. Our team is responsible for the calibration of the revenue management systems and providing ad hoc decision support for RM. My background is in business and economics with a healthy dose of computer science. I've been in the industry for the last 10 years, the last five years with Air Canada.  

Lucio Bustillo: (05:57) Like Justin, I'm a big lover of the industry not only because of the complexity of the problem, but what we deliver, which is an experience and really the opportunity to discover the world. It's something that I'm really passionate about. Not to the obsessive point of keeping track of the aircraft and the distances, but definitely I have my little collection of boarding passes going back to the last 10 years. Definitely a very dynamic industry with a lot of very interesting operations, research oriented problems to be resolved.  

Kevin May: (06:34) Okay, thank you Lucio. And last but not means least, Justin Silver. 

Justin Silver: (06:40) Hi everyone, I'm Justin Silver. I'm a senior scientist here PROS and I work on our science and research team. Our team really focuses on the innovation of our science based solutions. We solve business problems across many industries. Of course, airline is a big industry for PROS and has been for a long time, but PROS solves problems in a number of other industries as well. 

Justin Silver: (07:01) My background is in statistics. I've worked for several years on developing solutions for price optimization, sales effectiveness, particularly in the B2B enterprise space. But more recently, I've taken my experience from working in B2B and started to apply it in the airline space. I really see that there's a lot of opportunity for airlines to really take their digital commerce experiences to the next level to modernize and improve those experiences. I'm excited to be working on really interesting use cases in this area and to continue to innovate on that. 

Kevin May: (07:35) Okay, thank you Justin, Justin and Lucio. We'll come back to you straight away there Justin Silver. AI and machine learning, just broadly, tell us why do you think everyone is talking about it? 

Justin Silver: (07:50) Sure. Well, the first thing I'll say is that, really there is no universally agreed upon single definition of AI. People ask, "Well, what is that definition of AI?" I'll give you a definition and a little bit, but I think one of the challenges with that is that what people consider to be AI is always changing, it's going to evolve. As computers have more and more capabilities, as the capabilities of technology changes, that will continue to evolve the definition of AI. 

Justin Silver: (08:18) Some people will even say that AI is whatever a human can do, but a computer can't do yet. That pretty much gives us the impression that it's going to change. But generally, what we can say is that AI is an evolving set of technologies that are out to mimic aspects of human behavior. There are two characteristics of an AI system really, automation being one of them where the system is really trying to replicate the way that a human would make decisions and take action. 

Justin Silver: (08:48) Then there's the aspect of learning, where the system is going to make adjustments based on some type of feedback mechanism and then improve its performance. I'll talk a little bit more about how that adjustment happens, but another thing I would say is that while there's definitely a lot of buzz around AI today, you'd be hard pressed to go anywhere and not hear someone talk about it, turn on the TV and see commercials about it. Everywhere you look and turn it's definitely a lot of buzz, but it's not a new concept. AI has been around for quite a long time. 

Justin Silver: (09:20) What's really changed in the more recent years is certainly the amount of data that has grown exponentially. The amount of data that has been and is continuing to be collected, and of course, also the scale and the affordability of computing. But also, the algorithms themselves have evolved as well.  

Justin Silver: (09:38) I think when people talk about AI, they often lump it together with machine learning, as if AI and machine learning are the same thing, as if those terms are interchangeable. Sometimes you'll even hear people say that machine learning is even a type of AI. The truth is that neither of these assessments is really particularly accurate. Machine learning you can think of as the process taking data and turning that data into some type of model. And hopefully, that model provides a way to reveal something about the data itself, and even more so, hopefully it actually produces something useful. 

Justin Silver: (10:12) But, what can be confusing is that these models, once you've turned the data into a model, what's confusing is that these models themselves are often called machine learning also. For example, if you think back to our favorite statistics class when we all learned about linear regression. Linear regression is a type of machine learning model, as is a neural network, which is something that a lot of people are talking about now as another type of machine learning model. 

Justin Silver: (10:36) But, would you say that just because you have a system that's using a linear regression model, or even if it's a neural network model, would you say that just based on that alone, that you have an AI system? You probably wouldn't. So then you might ask from that point, okay, well what makes the system and AI system? Well, it's not necessarily the model that you're using that matters as much, but it's really the process where that model is being updated based on new information or new data, and the fact that that updating is going to happen in an automated way. We call this the learning loop. 

Justin Silver: (11:10) There are many different applications of AI, and we see many of them around us in our everyday lives. At a high level we can categorize them into these categories here. By no means am I saying that this is the only way to categorize AI applications, but generally you can think of it as Perceptual AI, where you're talking about chat bots and digital assistants. Think of Amazon Alexa and Siri, those types of interactions. 

Justin Silver: (11:33) Internet AI, where you're getting recommendations from say like Netflix, Amazon, Spotify. Business AI, where it's more about business decision support, things like algorithmic trading, fraud detection and automation. Then of course, something that we're seeing more of is Autonomous AI, robotics, the self driving car, et cetera. 

Justin Silver: (11:53) Regardless of what area of application we're talking about, you see how that learning loop in the middle, it's really that learning loop that's the key aspect for the system to be able to drive value. For example, that's the way that Netflix is going to get better at recommending you movies or Spotify is going to get better at recommended you songs. The more that you use those system, it's going to get better and better recommending those things to you. This aspect is really a crucial part of providing a good customer experience, which is just one of the value propositions that AI can provide. 

Justin Silver: (12:27) The value case for AI depends on the area of application. For Business AI, particularly with the solutions that PROS provides is often about maximizing revenue or maximizing margin. That's a common objective for airlines, and also for B2B enterprise businesses as well. But also, improving process efficiency is another major area of application for AI. The idea there is that AI is not necessarily replacing the analyst. A good business analyst is not going to be replaced by an AI, it's augmenting what that analyst is able to do. Automating those more repetitive, predictable tasks and freeing that analyst up to be able to focus more on the tasks that are of higher value, the more cerebral past. The tasks that really an analyst is really suited for. 

Kevin May: (13:18) Okay, wow. That's probably one of the best overviews I think I've heard for a while. I mean, it's one of those things where you ask one person you get a different answer, but that was terrific. Let's talk about how this applies to travel and the airline industry. Justin Jander, give us your overview and some of the typical problems that we're trying to solve, or you're trying to solve and the airline industry is trying to solve today? 

Justin Jander: (13:48) Yes, for sure. I think there's a couple of different areas where you see that AI can be applied in both the broader travel industry, as well as specifically on the airline side. I think the first one is around personalization and the customer experience. Often you look at it, and we always see the numbers for airlines in this aggregate form, millions of passengers boarded each year, the number of total hotel rooms sold. We look at it at the ... Its presented to us in the totals. Whenever an airline does their annual report, we get to see the kind of totals. 

Justin Jander: (14:25) But, what's interesting about the airline business is that the margins are really, really razor thin, so every single passenger is important to the airline business. And so, being able to work directly with the passenger and communicate with them in a way that encourages them to come back to you is really important in the airline space. That's really where you see the personalization become important. 

Justin Jander: (14:51) What we mean by personalization though, is this idea of when I'm interacting with a passenger on a website, I know something about them so I help them along. It's not necessarily this idea that we change the price for them or we do something that's trying to make us more money, it's really about a bigger picture for the passenger in the airline, so it's that overall passenger value that we get. 

Justin Jander: (15:16) One example I really like to use on this is something I personally experienced. I'm not going to name the airline that it happens on, but I fly with a particular airline a lot and I have a level of status on them that affords me certain benefits. One of the things I've noticed is that oftentimes whenever I'm checking in from my flight on the mobile app, it has the opportunity to present me with the opportunity to choose early boarding, and to choose free check back, or to choose a check bag and choose one other ancillary at a bundled price, which sounds really great, right? 

Justin Jander: (15:53) They're having the opportunity to communicate with me as the passenger, and so I have the opportunity to buy something else that is beneficial to me at a price that helps me achieve that. That's really great from the sense that it's helping me as the passenger. The downside is that in this particular case, all three of the benefits that are bundled together are something that I get with my status for free. 

Justin Jander: (16:19) This is where the idea of personalization comes in. The airline has the opportunity for a touch point with the passenger that enables them to really be able to interact and get something else, and encourage the passenger to get a better experience and work with them in the future. But yet in this case, they're not aware of who I am, specifically in this interaction so they can't tailor that that offering to me specifically. 

Justin Jander: (16:45) That's really where you see the opportunity for personalization to come in, specifically around artificial intelligence, we really can see the opportunity to say, "Let's learn about these people and learn about the behavior of those people as we see over time and help change what we offer them to really suit what works best for that passengers." And again, it's not something that's really negative, this isn't the robots are tracking all of the things that I do or something like that. 

Justin Jander: (17:16) This is more about the passenger's experience so that they say, "Whenever I fly this particular airline, they know who I am and they give me the options that make the most sense for me and that's something that encourages me to continue flying with that airline." That's one example of where we see problems being solved with AI. Another interesting one is around the operational side of the airlines. 

Justin Jander: (17:44) On the operation side, this is something that obviously we don't deal with at PROS, but it's something that the industry is looking at different opportunities as well. One of the interesting ideas here is the idea that the airlines have hundreds planes in the air at any given time, and one plane being pulled out of service during the day unexpectedly can cause a massive amount of disruption for passengers across their entire network. That's a real problem. 

Justin Jander: (18:13) One of the interesting ways that AI is being used is looking forward and trying to see, when will mechanical issues happen on an airplane, that will cause that plane to have to come out of service? In doing so proactively, you can actually pull that plane out ahead of time and adjust your network, and your schedule, and your fleet planning ahead of time so that you don't run into these situations. 

Justin Jander: (18:38) These are the types of things that Boeing is doing, Airbus, Rolls Royce on their engines is tracking this and they're working with partners to help see, let's track the intricate pieces of the information that are going on within the engine or within the different mechanical pieces, and sending that information back to the airline and really giving the ... Sending that information back to these algorithms that can help predict what's going to happen, learn from them when it does, so that it can then continue giving you that information. 

Justin Jander: (19:10) That's just one example of where we can see from an operation side. It's also really true from the crew planning and scheduling, all these different applications around the airline. What's really cool about the airline industry is that there's just massive amounts of data that's just waiting to be explored and airlines are really good partners in terms of saving that data, which is an excellent opportunity of really what powers AI. Because AI doesn't just make things up on its own, it really uses the information that it has out there and again, the richness of data that the airline has is really valuable from that. 

Kevin May: (19:49) There's one more here Justin. I mean, we're talking about revenue optimization. Can you give us a sense really of, how does AI and the science of it play into revenue optimization? And also, explain to us how and why did revenue management science within revenue management emerge? 

Justin Jander: (20:09) Yes, it's a great question and sort of buries the lede. Of course, I gave all these examples of other uses of AI, but the one that's out there is really what PROS does and that's in the revenue management and revenue optimization space. From my perspective of studying the history of revenue management, PROS has been in the business for over 30 years and has really brought the idea of revenue management from each individual small airlines, broadening out the perspective in doing it. 

Justin Jander: (20:39) PROS's approach has always been the science and really ... Not just PROS, but also just the RM space in general was really at the forefront of taking some of this key information that the airline has and using it to understand, what is the best way to maximize revenue for that airline? Just a quick overview for those that aren't familiar in the audience today, revenue management is really, the classical definition is selling and offering the right seat, at the right price, at the right time, to the right passenger. If you do all of those four things, you ultimately maximize your revenue. 

Justin Jander: (21:16) What we have to do that is the system, or any revenue management system, the goal is to accurately predict who will show up to buy a ticket and what that passenger is willing to pay if they do show up. Once we have those two critical pieces of information, so who wants to fly, as well as how much they're willing to pay to fly. Once we have that combined with the network that the airline has created, we can piece those things together and say, "It's the best option to take this passenger who's willing to pay this much, and in doing so we maximize the revenue. 

Justin Jander: (21:56) What the RM systems are all about is defining these controls inside of the RM system, that didn't get put out on to websites that actually dictate what the price you see in the end is. And of course, it's all about holding out particular seats for high paying travelers. Of course, the classic scenario for revenue management is segmenting between leisure passengers and business travelers. The reason that you do that segmentation is because leisure travelers typically have a lower willingness to pay than business travelers, and business travelers typically book late, close to departure. The airline knows that if those passengers are going to show up, it's in their best interest to hold that last seat for somebody who's willing to pay a lot. 

Justin Jander: (22:50) That's the quick overview of what revenue management is. One of the things that you hear throughout that whole description is, knowing about what the passengers are ... Who's going to book, knowing how much they're going to pay, knowing when they're going to book. All of these different facts and knowing about, are they business? Are they leisure? Are they flying from across my network from long haul to short haul? Are they flying just long haul? All of these different facts really build together to create this opportunity that's right a system, and it's also right for artificial intelligence. 

Justin Jander: (23:27) And again, from the artificial intelligence side, what we're talking about here is automating the process. Instead of having humans just sit down and attempt to come up with, "I want 10 seats at this price point and I want 20 seats at this price point," we begin automating the process, so understanding through forecasting, how many passengers wants to show up, we understand how much those passengers are willing to pay. 

Justin Jander: (23:50) Then once we put that out into the market, the second part of artificial intelligence really comes in, which is the learning process. Really understanding, what happens when we put those out, and how does the system learn and react to what we saw, and understanding how we can adjust that for the future. Continuing to learn, continuing to make the forecast better, making the estimation of how much passenger's willingness to pay better, and continue growing that and making the system even better as it learns over time. 

Justin Jander: (24:22) It doesn't really just stop there. In addition to our system from an automated in the backend perspective of these forecasts and the optimization side, we also have users that are responsible for really adjusting the inputs and the outputs of the system and understanding the business side of things. Through that, there's also an opportunity there to implement things that are repetitive things that the user is doing today and really let the analyst focus on being an analyst. 

Kevin May: (24:54) Okay, us journalists are often slow on the uptake with various things like this. I think I'm finally getting up to speed. Thank you very much. I think it's about time we probably heard from our airline representative here. Lucio, Air Canada in Montreal today, could you just give us a sense of, at a basic level, could revenue management be done without the "science"? 

Lucio Bustillo: (25:20) Well, I would probably put it this way, revenue management by definition contains a concept of optimization and that concept really comes from the characteristics of the product that you're selling. On the airline side, you're dealing with perishable goods, fixed capacity on the short term, volatile demand, and very seasonal demand. So, I would very much say that if the airline industry had to go back to a pre-science state, a lot of airlines would go broke. 

Lucio Bustillo: (25:50) Actually, it is this science that has enabled the airline industry to offer its products to so many passengers. As you might notice on how level trends of airline fares, the fares have done overtime nothing but come down and travel is more accessible than ever today. This capability of being able to scientifically determine the willingness to pay, yes, there's a significant portion of the consumer surplus that is moved to airline, but actually this increases the offer so we are able to reach a broader audience and make the travel experience more affordable for a broader range of people. 

Kevin May: (26:32) Certainly the democratization of air travel and air fares going down is something that most people would applaud me. Tell me Lucio, what would you say are the benefits of establishing some degree of consistency and efficiency in these kind of processes, and adding insight overtime to your decision making? How does that work? 

Lucio Bustillo: (26:53) Well, I would say that the profitability of a company really can be understood as the sum total of its decision. By making your decisions more granular and more accurate, you're able to increase that sum totals. It really as much as you're able to automate or rigorously address those decisions, you're able to focus your team really in the most complex cases and focusing on the overall strategy, which is what analysts should actually be doing, instead of taking a lot of micro-decisions constantly. 

Kevin May: (27:32) Okay, and then there are modern revenue management departments. I mean, they're looking at AI and science is driving initiatives. How are you and how do you sense the industry is doing that? 

Lucio Bustillo: (27:46) With any new technology, you always have to build a proper roadmap to apply it. I would say that this varies from company to company, but the usual approach is to build a short term roadmap and a long term roadmap. Your short term roadmap, it makes sense that you get the quick wins first in terms of figuring out which activities you do today could be enhanced by this technology. Or, which activities could be that you're not using any technology, or very rudimentary technology could be on boarded by this type of technology. Most notably, the forecasting that airlines do in terms of demand is one of the big areas where AI technology can be leveraged. 

Lucio Bustillo: (28:31) Also, another area where this can be leveraged is understanding the value of the network, especially the additional data sources that you're able to leverage with the scale of AI allows you to have a broader vision of your network and react a little bit more quickly and more accurately when you're dealing with the problems of operations that an airline actually deals with. Actually, it also allows you to very short term at forecasting, or an optimization aspect to stuff that seems rather mundane that you might not have a specialized system to do. With the affordability of AI technology, you can rein in processes that they are inefficient or manual very quickly and make them into more profitable processes. 

Lucio Bustillo: (29:26) On the long term of course, you would be talking of really reengineering or retooling the core processes of any company. In the long term I would say, nobody has an exact line of where it's going to end with AI technology, but definitely making operations more efficient and addressing the situations that would require a massive change in your infrastructure might be the next target for whatever strategy you have when implementing AI. 

Kevin May: (30:03) Okay, and just before we bring one of the Justins back in, Lucio could you provide us some other examples of airlines using data and analysts to bring value to the overall business?  

Lucio Bustillo: (30:14) Well, definitely RM is one of the ways that I'm most familiar with. It really is at the core of taking data, transforming it into a forecast and later transforming it into a decision. There's a lot of areas in the company, the difference is just the scale of the data you use and the depth of complexity. Aside from revenue management, probably I would cite airport operations, flight operations, and anything that manages the capacity that the airline has. 

Lucio Bustillo: (30:45) One of the very interesting things that at some point people have started looking on is for the operations of airlines on the coordination of airport agents; managing shifts, supporting the fact that they're constantly changing position. Definitely you can save multiple kilometers from a person walking from end, to end, to airport by smartly organizing their work schedule. Any item that generates a heavy footprint when it comes to data and that its defined by very small decisions that require a small amount of data is target to be enhanced by any of the AI processes discussed during the previous minutes. 

Kevin May: (31:36) Let's bring Justin Jander in here. What do you sense is next in the world of revenue optimization in the broadest possible phrase that we can use? 

Justin Jander: (31:49) It's an interesting question, because we listened to Lucio and this isn't the first time we've talked to Lucio. Working with his team and working with airlines like Air Canada, we really get the idea of what they're interested in the industry, and really understanding where it's going to go next. Some of the key things are, of course, on that back end prediction of how many passengers want to show up and what the amount they're willing to pay is, that's a key focus and bringing in new data sources that aren't really traditional for the RM space.  

Justin Jander: (32:25) Beyond just the bookings and the booking class, and the fair paid, we want to get more into the details of what are the drivers of the willingness to pay? But, it goes beyond that to. It really gets into the user side, because what we've seen is that we can continue iterating on the science and that's really important, but we really also need to make sure we're focused on the analysts, because they're responsible for really managing the network and really managing things that are outside of the purview of the system itself. 

Justin Jander: (32:56) What we want to do is take the processes that they have that are more repetitive and that can be automated, and let the analyst focus on being the analyst. One good example of that is around the holidays and special events. Those are things that really are where money is driven in the airline business, around your peak seasons, and your holidays and special events. What we want to do is take the guesswork out of when those things are happening, and let the analyst focus on the output once they have those particular scenarios. So, really focused on letting like I said, letting the analyst be the analyst, rather than being just focused on data input. We want them to really analyze data, and bring their knowledge of the market to the system rather than focus on just data input like I said. 

Kevin May: (33:47) Okay, thank you, thank you. This is the bit that I quite like, is when you can start comparing it to other sectors and how it's being used. Tell me Justin Silver, leveraging examples of AI and science from other industries, how could the travel industry particular learn from other industries in the adoption and application of AI? 

Justin Silver: (34:09) We've seen AI being increasingly use to basically extract insight from large amounts of data. AI is able to spot patterns and trends in the data, and really respond much more quickly than a human could. In that regard, Al is really well suited to give prescriptive guidance on things like how to price products, how to identify future trends and usage, personalizing offers for groups of customers as well as for individuals. 

Justin Silver: (34:38) At PROS, we've done this successfully across a number of industries; manufacturing, distribution, services, logistics and transportation just to name a few. Really, it's about finding ways to support the commerce experience, and that's both on the buyer side as well as the seller side. For the seller, it's really the system helping identify what are the right opportunities for which products to offer, to whom, at what time and at the right price. That sounds familiar based on what my colleague Justin Jander was saying in the revenue management problem. 

Justin Silver: (35:09) On the buying side, it's really trying to, the system's trying to surface the right recommendations at the right price to try to create a positive buying experience. But regardless of whether it's the buyer or the seller perspective, it's really important that we have a scalable system. That has to be able to support a large number of these customer interactions quickly, and very precisely. 

Justin Silver: (35:34) Travelers, they're using Amazon and Google and the Alibabas of the world, they're used to these streamlined and tailored online experiences. And airlines can can provide a similar type of retail experience. Applying AI to merchandising, and providing recommendations based on cross sell and upsell, getting access to a passenger's second wallet and increasing their spend, while also providing a good customer experience. 

Justin Silver: (35:59) But I'll say generally when it comes to infusing AI in business applications, whether it's for an airline, or a B2B enterprise, or some other type of business, when it comes to these business applications, just putting even AI aside for a second, there has to be a very clear definition of what is the problem that you're trying to solve. I think sometimes, there's a tendency among business leaders to take a use case and look at that use case and say, "Can I apply AI to this?" Often you'll hear that question? 

Justin Silver: (36:28) That may not be the right question to be asking, because if you can fully digitize a process, you can technically apply to anything that can be fully digitized. I think the real question to ask is not can I apply AI to this but, what's the value of applying AI to this use case? Then in order to answer that question, you have to establish, well, what is the measurement of success? In other words, have some type of KPI or key performance indicator? 

Justin Silver: (36:58) For example, we were talking about the offers of ancillary items like Justin Jander was talking about. You might say, well, am I looking to increase my conversion rate on ancillary offers like extra baggage or seay selection or lounge access, et cetera. But whatever it is that the KPI is whatever or multiple KPIs, those are what are going to drive that learning loop. We were talking about the AI learning loop before, which is going to adjust what the system is doing to improve this performance. 

Justin Silver: (37:26) From our experience, we found that it's really important at the outset when you're embarking on this type of an AI initiative, that businesses really define what are those key performance indicators, those KPIs at the beginning. That's what's going to make the difference between just doing science experiments in a lab versus actually delivering a science based system that's going to drive business value. 

Kevin May: (37:50) Okay, thank you. Here's the question that always comes up, and I think it would be really good to get your perspective. Do you think that AI will replace humans and is this good or bad depending on your answer? 

Justin Silver: (38:09) There's a quote from Vinda Souza of Bullhorn. He said, "Robots won't put humans out of business, they'll give them a promotion." When it comes to AI, we think about it as really reducing or eliminating those time consuming monotonous tasks, bringing people to do those more cerebral, higher value tasks. Moving up the opportunity to automate, to simplify processes, to scale processes and I think ultimately leading to more productivity from employees, rather than taking away from it. Gartner as is shown here is estimating that in 2021, AI augmentation is going to generate $2.9 trillion in business value, and recover 6.2 billion hours of worker productivity. 

Kevin May: (38:53) I guess that's the answer, isn't it? Just on that one Justin Silver, do you sense that that kind of attitude, which I argue is very progressive and forward thinking is mirrored around the industry and airlines in particular? 

Justin Jander: (39:16) This is Justin Jander. I'm going to jump in on that one, I think. I'd like to get Lucio's opinion on it to after I give mine. As Justin Silver said, you can turn on the TV and you hear people talking about AI. I think there's a rapper that's even doing a commercial for IBM Watson these days. You see these things about how many dunks the happened in a given basketball game, and AI was used to power that. 

Justin Jander: (39:46) Because of the emphasis on this in the public mainstream area, it makes people think about things in a different way. Now, of course there's the initial kind of scared part where okay, well am I going to lose my job? Are the robots taking over the world? It really balloons out from there and there's a lot of really good jokes that come with that, of course, but I think that when we settle in on to what it is, it's not ... 

Justin Jander: (40:15) Where I think that airlines in particular will converge to that we're just taking data and doing more with it. We have better computing power, we have better algorithms, and we have more opportunities to replicate the human actions to take away the again, I've said it a few times, to takeaway that repetitive nature of the humans jobs and let the human be the human. Let them actually work on the things that require human analysis, and human interaction, and discussion and things that a computer can't handle. 

Justin Jander: (40:49) Really, that's where I think this Gartner quote of the 6.2 billion hours of worker productivity. That's part, I mean, of course, the money is important to, but that's the part that I think is really interesting. It's something that allows the analyst or the person, the humans to be more productive with their day and take out some of those things. Lucio, I'd be interested in your opinion as well, and how you see the airline space adopting AI, given the scary nature of it. 

Lucio Bustillo: (41:20) I would answer it, and this is more going from a historical perspective. Probably if you go back to the 1800s, 1900s, it would be unthinkable to say that only 2% of the population would participate in agriculture. As you [inaudible 00:41:38] multiple rounds of industrial revolutions, there's always been this fear, "Oh, what am I going to do next? I'm going to lose my job completely," et cetera. But, I would say that society evolves and adapts. 

Lucio Bustillo: (41:52) As our technology gets better, we find new things to do with ourselves and new ways of being productive and this, what it actually does as you guys mentioned, is really enhance the productivity of the human. Take away the repetitive aspect of it and really allow the human to focus on the aspects where true intelligence is needed, where strategy is needed on those big decisions that are going to be taking just once and there's going to be a lot of uncertainty around them. 

Lucio Bustillo: (42:23) The AI is really going to take over potentially, on the stuff that are very well documented decisions that are extremely repetitive. I would say, and this is more from a self preservation kind of logic is, whenever you're asked as a kid you're supposed to say, "Oh, what do you want to be when you grow up?" I think that the answer of the question might need to change and instead of being a single answer that you give out, you might want to give out multiple answers first thing. 

Lucio Bustillo: (42:53) Second, you might want to answer a field of knowledge where you're going to be focused rather than a specific audience patient, because the change is only going to accelerate in the near future when it comes to AI and you're going to have to multiple times, redo your career. That's just a fact. 

Kevin May: (43:12) I have to ask you at this stage Lucio, what did you want to be when you grow up? 

Lucio Bustillo: (43:22) Certainly, airline pilot was somewhere in there, but I would say that revenue management was not very high on the list. It's something that I discovered along the way. 

Kevin May: (43:34) Okay. Just as we kind of got to the end of this little section, audience questions are starting to come in. If you want to send us a question in our last five or 10 minutes, please do so. You can do that by hitting the button on the live studio that you can see in front of you. That's for the audience. I'm just going to repeat that quote again because I thought it was terrific. Robots won't put humans out of business, they'll just give them a promotion. That's from Vinda Souza of Bullhorn. That's a great quote. 

Kevin May: (44:02) So, our panel, we've got a couple of questions come in and will be more coming in shortly. I guess this is a fairly technical one, so maybe one of the Justins here. Just a quick question. Our audience member asks, for new airlines or startup organizations, how to create proxy data set to choose the correct algorithm. Who wants to jump in on that one? 

Justin Jander: (44:28) I'll take that. This is Justin Jander. Yeah. You're right, it gets into the technicalities. I think that what we do from a PROS side is, we often get the ... First off is getting the data in the right formats and understanding what it is that you want to do. I think before you even get into creating the proxy data sets, what's really important is to establish, what question are you trying to answer? 

Justin Jander: (44:55) Once you've established what question you're trying to answer, then you want to decide, what sorts of data would drive the answer to that question? I think that the time of soccer games, the football games for those international people, the time of football game dictates the demand on a particular flight. This is just a random thing. But if that's what you believe, then you have to start collecting up the data that drives the particular outcome that you're thinking. 

Justin Jander: (45:28) In that case, what typically we do is, we start by creating the data, create the associations of the data. If you say that one thing drives the demand on a flight, you need to associate the demand with the driver and once you have that, you can do it on a small scale, and then start to work your way up from there to where you get to the airline sized data of hundreds of thousands of things that you have to process through. 

Kevin May: (45:54) Okay, thank you. Justin Silver, here's a good question. What does it take for us, an SME travel business to get started with an AI application? That's a really broad question, but it's a good one. What would you say Justin Silver? 

Justin Silver: (46:11) I'd hate to echo what Justin Jander just said, but I think a lot of what Justin Jander was just saying could apply here, really in first being clear about what is the use case that you're trying to address? What is that problem you're trying to solve? I'm a big, still stand by being a proponent of the scientific method, which we probably all learned in science class in elementary school, but it really does start with formulate a hypothesis and collect data, do some type of experiments to either refute or confirm that that hypothesis, then continue to iterate on that process. 

Justin Silver:(46:49) So I think it's, think about what are the pain points? What are the challenges that you're facing as a business that you're trying to address? Then think about, okay what data could I gather, what data do I have available, and start to evolve from there. The modeling aspect really comes in later on once you've gotten to the point of figuring out what problem to address, have the data to test it out, et cetera. 

Kevin May: (47:13) Okay, while we've got you Justin Silver, there's this one referencing the talk earlier about personalization. With regards to business travel, how does corporate policy broadly play into personalization? 

Justin Silver: (47:31) That's a really good question. In our B2B applications, we've definitely seen this because we're trying to provide personalized recommendations or offers, when it's not necessarily an individual that's on the other end, it's more of a business representative. It's really about being able to understand the profile. So, having information about the fact that this is a corporate login, having some information around that, which would connect to understanding about what types of offers would make sense for that type of traveler.  

Justin Silver: (48:03) Then understanding that if there's enough information about that particular company's policies, then potentially the offers could be in line with that. That might be a little bit of a dream, but things like that, that we would take into account more of the interactions based on the fact that it's a business profile. If I'm an individual and I shop for myself, but I also, maybe I shop on behalf of my business, I would have different profiles and different logins to reflect that. I think that would be a way that we could potentially address that as we've done in some of our B2B applications. 

Kevin May: (48:41) The [inaudible 00:48:42] world of cargo, we've actually had a question about that. Are there any recent advancements of applications around AI in cargo revenue management? 

Justin Jander: (48:53) I can take that one. At PROS, we don't limit our revenue management products exclusively to passengers, we do have some cargo side business. What we've done there is more around the pricing side of the cargo business. It actually is a similar answer to what Justin just provided on the corporate side, but it's more about doing a sort of configure price quote type of deal, rather than the actual revenue management, but more on the pricing side of cargo. 

Justin Jander: (49:28) We have some innovation that's been going on on that side, and that's something that we can provide more details about kind of through other channels. It is something that we're working on right now, and there's been a lot there to use the same type of technology that we have on the passenger RM side and extending that to the cargo side as well. 

Lucio Bustillo: (49:51) I would add to that statement that cargo business is let's say, a little bit messy in the sense that it deals a lot with regulation and repetition. Just to emphasize the point that Justin made, it is a classical operations research problem so many of these forecasting and optimization algorithms are applicable to it. If I had to draw stuff or items that could be improved on cargo with the application of AI, definitely the forecasting of demand, the forecasting of capacity available and of course, the optimization of the price itself, and the negotiation of contracts that could be improved or enhanced by these technologies. 

Kevin May: (50:37) Okay, thank you very much Lucio. We've had some really diverse questions coming, diverse in terms of where they're coming from. There's one with from an airport that we'll come to in a moment, but one from actually a tourism board, which I think is really interesting. Since in the modern age where customers are becoming increasingly aware of the data that organizations hold on them, what risks do you see around privacy and customer rights with regard to AI? Maybe be one of the Justins can help us on this one? 

Justin Jander: (51:12) I think that's one of the important things, and I don't want to speak just as the PROS vendor side, but the first thing I will say is that we take security very, very seriously from our perspective. Obviously things like GDPR are very, very important. Those are the types of things that we really focus on to make sure that any type of personal data that we do have is very, very secure. 

Justin Jander: (51:38) On top of that, often what we're doing isn't really about the fact that I'm Justin Jander at the airline's website, it's more about someone like Justin Jander. They don't know necessarily, we don't save the very specific information about the passenger, but we're saving more information around things that are like that passengers so that we can help again, make the experience better for them. It's again, not as much of this tracking all of their movements and understanding all about them, but more about understanding a way to make the experience better. 

Justin Jander: (52:16) Those are even things that you see, like disruptions happen and you see that typically if I knew that every time a disruption happened to this type of passenger, I should read book them on this type of flight. That's really the kind of things that we're looking at. It's not really about knowing that it's Justin Silver and I'm going to do something very specific to him. 

Kevin May: (52:38) Okay, thank you. This is our question rom our airport audience member. They ask, "Have airlines noticed an increase in airports using AI for operational efficiencies? Any suggestions in particular on how to work better together?" Lucio, you're with an airline, you travel to lots of airports around the world, what do you think? 

Lucio Bustillo: (52:59) I would say that when it comes to adoption of new technology, the distinction between public and private is very visible. I would say that that makes a big difference on who adopts what. Definitely I would say that airlines have a vested interest in supplying a certain amount of data to airports and airports to airlines. There's a lot of room and opportunity for both to exchange, and a quick example. 

Lucio Bustillo: (53:27) If airlines shared their demand data and passenger flow with airports and you establish a good deal with of cooperation, which speaking from our case, particularly when we do with some airports, the airports can do a better job at planning their capacity when it comes to the different security checkpoints that you have on the airport. So, definitely being able to share the data from one to another is one of the big enablers that would allow the airports to get on the train as well, because I would say that airports don't necessarily have access to all the wealth of data and airlines might sometimes have. 

Kevin May: (54:09) Okay, thank you very much. We're always conscious of everybody's time and we're up to the top of the hour almost, so we're just going to do a wrap up question. Rest assured, we've had some really good questions that have come in from various airline around the world, which is terrific. What we're going to do is, we take a note of all these questions and we'll pass them on to the team of PROS, and they'll come back to you one-to-one so don't worry, your question has not been ignored. They'll just come back to you directly, just simply because we've run out of time. 

Kevin May: (54:39) It's always good when you're wrapping things up, then to go round the table and just ask people to give us a sense of where this is all going. We talked very much about the here and now and some of the operational things, but let's start with you Justin Jander. I mean, if we were having this same conversation in a couple of years time, what do you sense that some of the people would be talking about, whether it's audience questions and what you would be talking about? 

Justin Jander:(55:06) I think personally, it feels like the airline industry is at an opportunity to be a leader in doing AI for a real business application that goes beyond some of these common things. I think in a couple years, it's going to be more about seeing these real applications and people will be asking about how the airline industry was able to do those things. That's not to say that it won't be without its challenges and those sorts of things, but the message that we'll continue to put out is that this is about putting the power of analyzing back into the analyst's hands, and really giving that. That's really what I see as the future for that. 

Kevin May: (55:45) Okay, Lucio what do you reckon? 

Lucio Bustillo: (55:53) Sorry guys, lost the connection for a moment there. 

Kevin May: (55:57) It's okay. We were just wrapping up. If we were having this same kind of panel in a couple of years time, what do you sense would be some of the the issues with relation to AI and revenue management that we would be talking about do you think? 

Lucio Bustillo: (56:13) Probably, and this might be from the ongoing discussions. Definitely, your forecast accuracy two years from now will have significantly improved. You might be working on more models oriented towards willingness to pay, so definitely there might be talk about those and how to optimize those. Then probably the conversation would go to the integration between the e-commerce platform and the revenue management platform, and how to synergize the most between those two platforms to maintain airlines profitable and deliver a reasonable product to the customers. 

Kevin May: (56:58) Okay, thank you. The last word goes you Justin Silver. What do you think? 

Justin Silver: (57:02) Yeah, just actually continuing on what Lucio was just saying, really about providing those better customer experiences through e-commerce for the travelers, while also taking into account the revenue impacts of what's being presented to the travelers on the RM side. I think really the marriage of those two angles of the RM, as well as the e-commerce recommendation systems and another personalization aspects, I think those things coming together, I really see that. 

Kevin May: (57:32) Okay, that's great. Unfortunately, we are at the time. I sense given the number of questions that came in, in those last couple of minutes, that we could've probably gone on for another 15, 20 minutes. We've had questions for more airlines, and online travel agencies and all sorts. It's great, a very engaged audience which is always a good sign. First of all, thank you very much to our panel. That's Justin J, I'm now going to call you Justin S, and Lucio from Air Canada. Thank you to the team for taking us through this so expertly, thank you. 

Justin Silver: (58:08) Thank you very much. 

Justin Jander: (58:08) Thanks for hosting. 

Kevin May: (58:12) Last of all just to remind everybody, there is a replay. You'll get a link to that coming up. And all the questions, we did take a note of. They'll get back to, the guys are PROS with some one-to-one answers there just simply because we ran out of time. And just a reminder, we've been doing some other work with PROS here at PhocusWire. We've got a specific report that was published today as well to coincide with this, The Reality of Modern Airline Retail: When the Sky Isn't the Limit. 

Kevin May: (58:40) It's a very lengthy white paper that we've been working on with PROS for a couple of months now. Please look out for that, it's on phocuswire.com. It only goes for me to say thank you very much to Air Canada for joining us on the panel and in particular, thank you very much to the team at PROS for working with us on this PhocusWire webinar, and most of all the audience for tuning in today. We hope you have a great day. Thanks so much and goodbye. 

Lucio Bustillo: (59:06) Thank you very much and goodbye. 

Previous Video
How Distributors are Managing Online vs. Offline Pricing Challenges
How Distributors are Managing Online vs. Offline Pricing Challenges

Customers rely more on distributor websites for customer data and purchasing. As a result, it is becoming m...

Next Video
The DNA of Today’s A+ Sales Organization
The DNA of Today’s A+ Sales Organization

The selling process has gotten increasingly complex. Learn more about the challenges facing your sales team...