Strategies for Adopting AI in Changing Airline Business Models
Today’s generation will become intimately familiar with AI, deploying it across teams and embracing its power to further business objectives and drive outcomes. This fireside chat discusses how airlines are adopting and implementing AI, including considerations in adoption, best use cases, strategies, and challenges. Watch as PROS President of Travel Surain Adyanthaya, Microsoft Global Leader of Travel Transportation and Logistics Julie Shainock and flydubai Senior Vice President of Revenue Management Ramesh Anantharaman discuss.
Full Transcript
Surain Adyanthaya: The topic is the application of AI and the changing airline business model. For this panel, I'd like to invite on stage Julie Shainock from Microsoft, and Ramesh Anantharaman from flyDubai.
[applause]
Surain Adyanthaya: Hey, Ramesh. Please have a seat. Thank you. So, to start off, could you introduce yourself, a little bit about your history and your AI background, with your jobs. [chuckle] ...
Julie Shainock: Do you want me to go to first?
Surain Adyanthaya: Please, Julie.
Julie Shainock: Okay. Hi everyone. I'm Julie Shainock, I'm the global leader for travel, transport and logistics within Microsoft, so that's on the enterprise commercial side where all the different industries sit. I've been with Microsoft for seven years, and I've been in the industry for over 30 years. So, and all I've done is travel and transportation. So, I've been involved with AI probably the past 20 years, 20 plus years, so that's me.
Surain Adyanthaya: Thank you.
Ramesh Anantharaman: Hi. So, my name is Ramesh Anantharaman, I've been in the airline industry for 35 plus years, and then revenue management for about 25 plus years. Currently, I'm heading the revenue management systems and all of commercial systems for flyDubai. I have been involved with Emirates and before that with Air India and I've been involved with PROS for over 25 years.
Surain Adyanthaya: Yes. [chuckle] Thank you. So, I guess to start off, AI's been such a buzzword and this entire event has been really focused around AI and the application of AI for pricing, revenue management and offer and order, but we've all been involved in some level of AI for decades actually, and PROS has, and your companies have, can you talk about when you first encountered the application of AI and your view on the evolution of this space?
Ramesh Anantharaman: I think, when I first encountered AI, we did not have the term AI. We had much more of analytics, and that was way back in the mid-90s when I started working in revenue management. And then of course with PROS and a few other systems, which we developed those days, it was called much more analytics rather than AI. And over time, we have seen huge leaps and bounds in analytics, and the latest one being the generative, and that's when the old one was called analytical AI. And the new one is being called the generative AI. So, it's been an amazing evolution, I would say for the past, at least from an airline industry perspective, over 30-plus years. It's been a good journey. And I think there is a lot which can be done.
Julie Shainock: And from my perspective, you know, certainly, AI is not new, and it really started in the late '50s, early '60s. And I go back, and I look at AI, machine learning, deep learning, and then we've got generative AI. So, as you look at generative AI, I look at the definition, and I'll read you the definition just so you have it, but generative AI is to create new written, visual and auditory content given prompts or existing data. So, as you look at the journey we've been on, not only in the airline industry, but overall, there's a lot of different things that we've taken into account. So, it's really how do you do different prompts and how do you have large language models and then be able to draw insights around all this data.
Julie Shainock: And I think that's the other piece of this, as you move into generative AI, there's a lot of data out there, and it really starts with data and laying a lot of the foundation from that standpoint. And that's the point of view that we look at from Microsoft as well as for the airline itself. And one other, I'll just say one other thing is, in a pre-pandemic world, there was a study done by McKinsey, and McKinsey ranked travel and then transport and logistics as a top two industries to take advantage of AI. And the reason for that was because there was a lot of data, whether it's avionics data, telematics data, had all this data coming from those environments, but then in a post-pandemic world, Accenture just recently did a study and released information, and in a post-pandemic world, we are, travel is fourth from the bottom, and the reason for that is the maturity of the data that we do have. So, we do look at data as one of those fundamental tenets and having that data in the right platform.
Surain Adyanthaya: Yeah, I completely agree with that, Julie. In fact, if you look at the history of AI, when Deep Blue beat Kasparov, the pivotal moment was when it was possible to capture the last 10,000 Grandmaster games and digitize it to train the algorithm. The algorithms were there from the '70s, and same with the Google translation services, when they broke a billion tokens of translation, that's when that became possible. It seems like data has really driven the possibility to use algorithms that have actually been around for decades in some cases, and now are coming of age. Absolutely. Ramesh, you have a very agile team. Your team, you guys move really fast, can you talk about some of the AI initiatives you have going on, use cases, value propositions that you're finding are the best to approach first?
Ramesh Anantharaman: Okay. And for us, the biggest advantage is we are still relatively a young organization. So it's easy for us to change, and we definitely have implemented a few use cases, lot on the ancillary offering area, ensuring that whatever ancillaries we offer, we target the right people, there is a lot of initiatives going on and lot of it is centered towards customer experience, how to make sure that we provide the right kind of customer experience, lot on the predictive analytics, even though it's kind of, I'm talking off the cuff in terms of what we are planning to do is the ability to identify passengers who are likely to report late at the gate, and how do we get to them? How do we proactively tell them, okay, you're going to miss your flight, coming across, that is some of the use cases and a lot more heavily centered towards customer experience.
Surain Adyanthaya: Interesting. Julie, from your perspective, you've worked with many different companies?
Julie Shainock: Yeah. We've worked with many different airlines around the globe, and I think we have many examples. I'll just highlight a few of the examples that we do have, probably the first example, and you may all have heard about it, it's more on the operations side and it's all around the turn times. And so, we've worked with American Airlines and other airlines as well to deliver a turn time solution that is really based on Teams, which is a communication collaboration platform. So, we started, this was a project that started in a pre-pandemic environment and the only project that remained. And so, this project really looked at reevaluating and re-engineering the overall process of an airline at the airport, for the gates at the airport. So, the gate manager and then the gate agent, etcetera.
Julie Shainock: Been highly successful. They've implemented this at 264 airports today, so really looking good. The next thing you look at is you've got above and below the wing, everything, all coordinated. The next thing is you can start within the knock in the sock, as you look at the channels all during the day, you can really start to apply AI to determine which flights might be impacted during the day. And so, then you can actually try to resolve some of those issues so that you don't have the impact during the day itself. Air India is doing a similar kind of functionality. They also have an AI chat bot that they built with Copilot and I'll talk a little bit more about Copilot. But then Indigo also has a Copilot that they've built for reservations and they've seen, with a soft launch, they've seen a 75% reduction in the call center agents.
Julie Shainock: So that's a huge opportunity as well. And I think the other piece of this is we look at the AI brain, like for operations. So, customers want to know, as you talked about, customer experience, customers really want to know if you have a delay, they want to know they're going to be able to get to their next destination, so they're very concerned with some of that. Then if there's an actual cancellation, they really want you to handle all the different, you know, canceling the flight, getting them a hotel, getting them meal vouchers, etcetera, etcetera. So, I think those are the kinds of things that we're working on with airlines around the globe to try to handle some of that. Some of them are doing it very, very well and with their high value customers, but it's looking at every single passenger on a flight when you do have a disruption.
Julie Shainock: So those are some of the things that we're seeing. And then we're seeing also Copilots using some of the natural language search. But I think one of the most important things as we look at some of the new AI, the Gen AI, is being able to, it's the prompt that you're asking, you have to be very, you have to train the models and the large language models, but you also have to train yourself on how you're going to actually ask a prompt. The prompts are very important and really determine how you're going to get the data. So, those are a lot of the things that we're working on with the airlines.
Ramesh Anantharaman: Yeah, and if I can add a little bit more on that, Julie, even though I did speak about customer experience being on the forefront, there are a lot of things which are behind the scene operations where we are looking at applications specifically in the operations area, ensuring we have an on time performance, and we are also looking at kind of a predictive maintenance especially on the aircraft side, which are all the aircraft which is likely to have fault. And the amount of data, what we have available, especially on the operation side, is humongous, and ability to harness all of that data and ensure that you have a very reliable operation, maximizing your utilization, those are some of the areas where I think there is a lot of potential. A lot of airlines have done it, but I think there's still a lot to go.
Surain Adyanthaya: Yep. I think customer experience will be a key differentiator between airlines in the future. I can tell you, there's one, there's certain airline that I fly all the time and they always offer me a free bag and I haven't checked in a bag in 25 years [chuckle], and I have three free bags on any flight if I did want it. So, it's clear they don't understand me, and it should be a minor thing, but it really bothers me that they don't know even the basics about who I am as a traveler. And I think the new generation of traveler, the millennial especially, expect you to know who they are. I'm not a millennial, but they expect you to know who they are.
[laughter]
Ramesh Anantharaman: I don't think any of us are.
Julie Shainock: None of us are.
[laughter]
Surain Adyanthaya: And to have an understanding of what their needs are, and they will go to another vendor who they believe really appreciates who they are and will fundamentally provide a better offering. But given what you're saying, when you look at these AI initiatives, how do you look at the value? Do you look at them as value-generating or cost-saving, or is it a combination of the two? Because you have to create a business case and validation to do it.
Ramesh Anantharaman: I would say both value generation, cost saving, as well as providing lot of non-quantifiable benefits. For example, if you are providing a best customer service and consistent customer service. And what AI is going to come, especially in the call center area, is not just foster customer service, a consistent customer service because you're no longer dependent on individual people's skillset to provide the best service to the customer. The machine learns and supplements or augments the response from the call center agent. So definitely a revenue generation cost, and I think a lot of non-quantifiable benefits, a lot of it. And I think if I want to go a stake, take a step back. And we say that even though there are a lot of things which benefits are there, the airlines and everywhere, every industry has to invest a lot of time and having, building the right foundations, as far as I'm concerned, the right foundation for any kind of AI initiative is having the amount of right data and at least a minimum amount of data to ensure that whatever predictions you are doing are meaningful. And I think that kind of preparatory work is crucial.
Surain Adyanthaya: Excellent point. Thank you. So, I guess moving along, Julie, Microsoft's been at the forefront of generative AI, your partnership and investment in openAI has been massive and some of the recent news about investment and new processors and things, it sounds fascinating. Can you speak to some of the new innovations that have just come out and what can we expect in the future?
Julie Shainock: Okay, yeah, I mean, I can't speak to everything in the future, but I'll give you some ideas. Certainly, I think last week there was one big announcement around ChatGPT-4o, and I think the most significant announcement there has been with the voice, the voice aspect of things that are now included there. So just, that's going to be a huge play going forward. I think the other thing I would tell you as you look at AI itself, there's been an IDC global study, it's from September of 2023, we know that for $1 invested, there's a $3.50 return on investment but 5% of organizations are receiving an $8 return on investment. We know that 92% of solutions that are deployed with AI take less than 12 months.
Julie Shainock: 40% of organizations have implemented these types of solutions, the AI solutions, in less than six months. And, we know that their ROI for investing with AI is on average less than 14 months total. So I think there's a significant return on investment that everyone's looking at, and I think as you look at the Copilots themselves, you can use Copilots to do a lot of different things, you were talking about maintenance, so you can actually do a Copilot around TechOps and all the line people can be using a Copilot to query information. You can do the same thing, you know, you might have somebody asking about fleet and within the knock and the sock on day of operations, you can have something front-end that as well. I think operations are seeing a significant amount.
Julie Shainock: And then as you look at customer experience, so looking at different environments where they're pulling all that data together, you were talking about they really don't know you, that's because they only have probably a single piece of data around you itself, and it's probably around your loyalty piece. But when you start to go out and pull in all different kinds of data sets, then you're able to really do that Customer 360 environment, know me, value me, serve me over my channel of choice. So, I think these are the kinds of things that you're going to start to see. And as you mentioned, I really do see we finally can get to this 360-view going forward. The other piece is we've got personalization, hyper-personalization, certainly there's a lot of enhancements that you all are bringing to the market as well.
Julie Shainock: And so, we look at the digital ecosystems, and this is sharing information among the players that need to share information. We look at the digital decentralized identity. So that's me, I'm going to opt in and share information. Then I don't have any of my PII data sitting out there and you're not, you as an airline are not liable for that as well. So, I think there's a lot of different things that are coming and we still have to have the foundations in place, but these are the kinds of things that are out there. So, I mentioned the multi-data sharing of information. And then, I think the other piece, as we look at the AI, the other piece is really all around the security itself. So, Microsoft, that's a major tenet for Microsoft, and we are very active in that space, and we have solutions across all the different areas and all the different pillars, but it is a key tenet within Microsoft around this whole security aspect, the bad actors, the threats and everything else. And we will continue to invest. So, we've invested 5 billion and we're going to invest another 20 billion over the next five years. So those are stated areas where Microsoft is going to make some significant investments. So, that's the other piece of the whole AI piece is the security aspect of everything.
Surain Adyanthaya: Great. Thank you. That's a lot. It's a lot going on. Exciting to see it. One thing that I feel like we didn't hear too much of at this event was the organizational impact of some of these algorithms and the upcoming AI. Will it cause a restructuring of these organizations or rethinking the role of the human, the analyst? What's your opinion on how all that evolves?
Ramesh Anantharaman: I definitely believe that there will be a total change, but I do not recommend, I do not see a radical change overnight in terms of organization structure or the functions. I believe it has to be kind of an incremental steps because you jump into AI without proper analysis, proper thinking, you could do much more damage than the benefits, what it can create and it's going to bring down the morale of the staff. And we always believe that our biggest assets are our people and it's very important to take everybody along and saying that AI is there not to replace you, AI is there to help you do a better job, help you to say that, to take care of all the rudimentary tasks, everything, all the routine, even from a frontline perspective saying that all the routine questions, what you answer, you get the system to do it, you get the machine to do it, and you step in where the real expertise comes in.
Ramesh Anantharaman: So, you use this as an opportunity to upskill everybody, to elevate themselves, to get them out of the mundane work, and from that perspective, and it's not just with the frontline staff or the call center or technically not technical staff. I think even from technical staff, there is a lot of apprehension saying that even right from developers, am I going to have a job, engineers, am I going to have a job if machine is going to be doing everything? So, it's very important that we take the team along. We say that AI is there not to replace you, is there to supplement your work and your kind of satisfaction is going to get better by using AI.
Surain Adyanthaya: Absolutely. Completely agree. What do you think, Julie?
Julie Shainock: Well, from a Microsoft perspective, I think you mentioned frontline workers. I mean, this is a huge area. So, when we started to do some of the work with... And we were still doing the work with the airlines, the frontline workers, that's your first face to your passengers, or your customers, or your guests. So, I think it's so important to give the tools directly to the frontline workers and have them utilize the tools just much more efficiently. There's no reason for a gate agent to have to walk down the jetway, they can handle everything from their phone, from an iPad, whatever, open the flight, close the flight, handle duplicate seats, handle catering issues as well. And then the bins themselves. Well, in the US all the bins get filled really quick because everybody carries everything in the US, but outside the US, people tend to check their bags a little bit more.
Julie Shainock: But it's amazing to watch and someone will say, "Don't check any more bags." All they do now is we have an index, think of an index card and it's like an adaptive card within the Teams platform and they say, 90% full, start checking bags now. So, these are the kinds of things that speed efficiency, turn times. But these frontline workers, the worst thing that can happen to them is they don't know the industry's adapting. There's a delay or something's going on and they don't have the information. So having the information improves the customer experience, the employee's experience for them and the customer experience for all the different passengers, customers and guests out there. So, I think that's part of it. I look at it at the frontline and I also agree with you 100%, the goal is not to put people out of work, it's to upskill people to handle other things. So, I look at AI as augmenting and helping, you know, have you do your job better and more efficiently and have better information.
Surain Adyanthaya: Great. So, I guess the last question. We're all in different stages of our AI adoption journey. What would you... Any advice you'd suggest on how to get started for folks who are early in the adoption?
Ramesh Anantharaman: My most important advice is invest sufficient time in building the foundation, building the data. We have been working on what we call as a big data project to ensure we have a common place where the entire airline's data is there, be it operations, finances, commercial, customer, everything is there in the commonplace, it might look very easy, but there are huge amount of challenges. But because especially in an airline, the same information, you can get it from multiple systems and they're not always the same. So, you need to understand, and you need to clearly define which is going to be your single source of truth. And possibly it'll take a few years before you can integrate all of those. I think that's the first step you need to do. And take small baby steps, look for opportunities where you can have quick wins to get the team, rest of the team on board. Because in spite of the fact that there is a lot of buzz in the entire world, everybody switch on TV, everybody talks about AI and everything.
Ramesh Anantharaman: A lot of people are really not aware of the core aspects of AI. Everybody talks about AI, but if you look at the understanding of a normal human being, a common man, the understanding is very poor. All they know is saying that, okay, it's a magic box comes and going to do something, at the same time there is a lot of fear. Just make sure you eliminate the fear out of the people. That is by providing some quick wins, how AI, these solutions are going to help each and every person. I would say that's number two. And number three, always look for opportunities where AI can be implemented. As I was sitting here, I saw an amazing use case came and we know we are missing one of our panel members, Claudio. And possibly in a couple of years time, if you see that one person is not there, you'll be able to switch on the AI. And you ask the questions to the machine and the machine has got all the information about what Claudio spoke in the past, including the previous conferences, whatever all is posed and possibly able to give you a 90% accurate response for Claudio.
Surain Adyanthaya: It's possible none of us have to be here in the future.
[laughter]
Ramesh Anantharaman: True. And I was thinking that this next use case and saying that, okay, Surain is going to tell me saying that Ramesh, I'm going to ask you this question and AI tells me this is going to be a standard answer. Now add something more onto that. I think that's very important, we understand to say that what is it we are going to do? We are not going to be doing the routine things. We always look for something better. I think, always look for those incremental opportunities. Keep your eyes and minds open. I think minds open is very important to look for use cases, how you can make a difference.
Surain Adyanthaya: Would you agree that, I think sometimes analysis paralysis happens, people want to have massive projects that you take small steps that fund the next small steps.
Ramesh Anantharaman: It's very, very important we do it. For example, what we are working on is this Customer 360. That project has been going on for some time and we hope to get it to fruition. And what we are looking at, even before we complete all of that, we are going to provide some incremental values across to various touch points and saying that, "Okay, this is the added benefit you're going to get. This is the added benefit you're going to get." And that's very important people get used to it and saying that and more than a central team pushing solutions across, you have to get the business to say that, "Okay, I want more solutions from here. I've got this challenge. Can you help me?" I think if you're able to get to that stage, I guess that's going to be a successful implementation.
Surain Adyanthaya: Great. Thank you. Julie, would you like to wrap it up with some thoughts on this?
Julie Shainock: Sure. I'll wrap it up. I agree 100%. I mean, the first thing is the data platform. The second thing is look for quick wins, low-hanging fruit of opportunities that are in your organization or across your organization. And then, the third area is we're looking from a Microsoft perspective where each one of the industries is looking at what are those top use cases around AI. So, like for us, call centers within airlines, call center might be one of those. Or call center analytics and how you're dealing with customers, customer service requests, all those kinds of things. Search and itinerary planning can be one of those where you can use natural language in that environment. Marketing data. So, there's a lot of different things going on. And so, as you look at all the different data out there and the data sources out there, so like Microsoft owns Xandr that they bought from AT&T.
Julie Shainock: We're working with other companies where they've got gobs and gobs of characteristic data that we're starting to utilize and monetize. So, there's a lot of different things that we can use from a data monetization standpoint. And then the connected aircraft. So, that would be some kind of new different IFE in that environment. And then, I guess, the last area might be the maintenance aspect of things where there's lots of different things where we can do from a remote assist and you can use chatbots in those areas. So, I would look at, and you can actually go onto a Copilot and ask what are the top use cases for airlines and that kind of stuff. And, some of these may come up, some not, but we are actively working with airlines across the globe in these kinds of environments. So, I would encourage you, I guess the other thing I would just say, just to end it is, go ahead and get started, but at least follow the steps that were outlined because they're very true and it'll make you much more successful in the marketplace.
Surain Adyanthaya: And I'm hoping that Copilot would say dynamic ancillary pricing.
Julie Shainock: Yes, that's it.
[laughter]
Surain Adyanthaya: Request specific pricing, willingness to pay, and a few other things that we think are a natural next steps too.
Julie Shainock: Absolutely.
Surain Adyanthaya: Cool. Thank you. Well, with that, thank you so much.
Julie Shainock: Thank you.
Surain Adyanthaya: I really appreciate your time and we hope you had a great conference and see you next year in Vegas.
Ramesh Anantharaman: Thank you. Thank you.
Julie Shainock: Okay. Thank you.
[applause]