What airlines like JetBlue teach us about building agentic customer experience

So good morning, CCW. Nice to see you guys again. I am Chris Arnold. I I've spent the last six years at ASAPP very much focused on AI deployments in the contact center. And prior to ASAPP, I spent twenty years at Verizon, started out as an agent on the phone many moons ago, and I think I did just about every job in the contact center. Today, spend most of my time with operators like Preya and operators like all of you guys really helping bring to life all of the AI technology we're gonna talk about today. Yeah. And hello, everyone. Thanks for for coming today. My name is Preya Rampertab, and I am the senior manager of our customer support standards and programs team at JetBlue. So I've actually started my career in a contact center space as an agent. That was about fifteen years ago, and unexpectedly grew my career within the contact center space, first as a training and development manager, and then as an operationals leader. And now, I'm overseeing our strategy for customer support and the policies and the processes that govern the customer experience holistically. So as you saw, a little technical difficulty with the mic, but like on or off, I'm excited about this topic. Can't wait to dive in and we're gonna have a great discussion. Now Preya, I wanna start with you because when we think of high stakes customer interactions, we know that they exist across all industries. Every industry out there has that moment where if you don't show up for your customers, if you don't deliver what they want, it can create a very negative experience. Now in airlines and in the travel space, that's very clear. You know, you miss your flight, you're going to something important and you can't get what you're looking for, that creates a lot of stress, a lot of frustration, a lot of worry. And the best brands, the best airlines, the best support systems, they help you overcome that concern. But unfortunately, sometimes agents themselves are facing concerns. If they're overworked, if they're stressed, they can't be what our customers need. And I wanna start by asking you, when you think about the types of issues your team handles, what are ones that maybe seem to overburden the staff? Be it emotionally, be it just in terms of focus, that are you know, on the radar right now. Yeah. So I think one of the biggest things is when there's high complexity, there's high emotion, and that can be I think some of you guys are probably wanting to go back home, whether it's the Northeast or Atlantic because of the the snowstorm that's coming. Right? When those type of issues arise, it is a strain on our crew members from an emotional perspective. They really do take the ownership of that customer to say, "you know what? I really wanna get you on the soonest flight I can, but these are just the options that I have." Right? And we have these, what we call, blue hero moments where our crew members are trying to go above and beyond, and they're trying to see like, you know what? "Let me give you a callback. Let me see what options are available." And then they, you know, they try to take it that that step further. I will tell you that with the traditional automation that we've done, you know, a few years ago, a couple years ago, it worked back then, but it was because some of those interactions were very linear. The interactions that our crew members are dealing with now are very multifaceted. Right? They're looking at rebooking or refunding or they're looking at, you know, helping the customer look at another option that's weeks away. Right? So the AI that we've looked at has looked at orchestrating different workflows to coming together to help provide the customer with these are your options, but these are the policies that govern what we can do and how we can help you. And I think additionally, helping our crew members through that, right, with additional training. I think, Brian, you talked about empathy. And empathy is such a big thing at the moment. But it's not just empathy. It's conflict resolution. It's de-escalation. And that's what's helping our crew members go through these paths and these journeys with our customers. Yeah. And the reality is, I hear a lot of talk about automating simple issues, eliminating simple tasks. And we have to remember that in the eyes of our customers, nothing is always simple. It may be like, thus require tons of critical thinking. It may not be the most intellectually rigorous question to answer, but to them it now. To them it's important. Totally. And to them there's gonna be a degree of nuance in context that we need to appreciate. Now Preya, you mentioned that traditional automation isn't sufficient for that. It doesn't think about the issue the way our customers do and therefore isn't affected. Now Chris, you come from the technology side, you go to the space, why is that? Why has the way we've approached AI automation not been enough for the reality of our customer issues? Yeah. I think if if for those of us who have been around a couple of decades, we've seen this movie before where new technology emerges. So you've got IVRs, you've got chatbots and CRMs. Everybody in here has a host of systems that have delivered on a few things. You know, the simple things like making payments, changing addresses. Some of these legacy technologies are quite good at handling the simple stuff and quite frankly that's the reason why agent, human agent jobs have only gotten harder over time because the easy stuff, I remember when I was an agent has been automated away. But those technologies are very limited because they're very brittle. You know, all of the complexity that Preya mentioned, a deterministic bot cannot handle anything that sort of branches off of the very norms. But the generative solutions, the agentic solutions that are in front of us today are very good at handling those nuances. Handling very individualized, going beyond personalization to an individualized when I need to have my flight rebooked or I'm looking for lost baggage that's very specific to me. The deterministic technologies of the past couldn't handle that because it didn't have access to data. It didn't have access to the systems required to actually get to resolution. That's the key difference. So show of hands, who here has heard sub variation of we're competing on the customer experience or customer centricity matters? Good enough amount of hands, I'm assuming everyone else is just tired of the long night Orlando. But these are pretty common cliches of our space and they're great, right? They rally our community to say, hey, we matter. The agents we hire, the people we train, we have an impact. We drive business results. But they also have the consequence of really creating expectations for customers. Customers now can say, I don't have to settle for the worst experience. I don't have to settle for what's generic. So you know, this can go to either of you, but when we think about how customer expectations are changing, what do they look for now, especially in the airline space? What does that mean for the way we have to approach automation and the way we design our journeys? Yeah. I think, first and foremost, I think our customers are looking for reliable and caring service. And I think there there is a level of "I need something I need something quick. I need a quick resolution. Don't pass me off to fifteen million people. I need you to help me solve my problem." And then there's a bit of transparency, right? "What is happening? Why is it happening? And and what options do you have available to me?" And I think sometimes we underestimate that why, right? I think that why is so important to say listen, this is what the impact is. I unfortunately cannot control mother nature and this snowstorm that's coming, we have to work around it, and these are the options that we have available to you. And then the resolution aspect. Yes, we absolutely wanna resolve their problem, but there's that reassurance that I'm gonna help you through that journey, and I'm gonna help you get to where you need to be, and we just need to work together towards that. Right? And I think that's what good looks like. And it's using the data. Right? Using the data to say, not everything is gonna be AI related. Some of it is governed by our policies and our processes. So do we need to change that to enhance that to help complement that customer journey? Yeah. And I think that human element, that reassurance, that belief that get why this matters to them Yes. And you're responding accordingly, that's certainly important. You know, the reality is, it may seem like a five minute delay, but that five minute delay can be the difference between making an important job interview, being there for the family, spending time with kids that you don't get anymore, and that's gonna matter. And if you don't show through your actions, your responses, your dialogue that it does matter to you, that customer is gonna start to become entitled. They're gonna start to be argumentative. They're gonna start to create an unfortunate environment for your agents, your team members, your crew. And Brian, you mentioned something earlier about the ability that, you know, what's small for us is not small for that customer. And I think oftentimes we tend to put our own biases into, you know, this is the this is the the Oh, this is a small issue. That's okay. They'll they'll be fine. And it's like, no, it's a small issue for us. But it's a broader issue for that customer, and and that's what we need to take care of in the moment. And then what makes it even more challenging is that accepting it's a big issue with real stakes, and yet, Chris, they still want it done quickly. They still need it done conveniently on their terms. And I imagine that is why we're having this conversation, right? Because we have to do this at scale, we have to do it faster than ever, but we have to do it in a way that feels personalized and tailored to that. That's right. You know, I would argue that customer expectations haven't really changed a lot. I mean, we as consumers, we want what we pay for, right? And we choose to do business with companies and we expect loyalty and we expect when things don't go right that there's a rapid response and I don't have to contact you multiple times. I don't have to repeat myself every time I call and that is really the charter and honestly I remember a day when I used to be compared to my competitors. You know, I think about my days at Verizon, we would have outages. That was the urgent response required in my world. Preya's is even more urgent when I'm trying to get a customer to destination, right? And so how do we respond in these critical moments, the moments that matter? And it's how do we use their data to create a very individualized experience that feels easy, low effort because I'm no longer comparing myself to my competitors. I'm comparing myself to every experience I have. So I think about the Amazons of the world, Chick fil A where I live, you know, sets a very high bar. You know, and for those of you lucky enough to stay at a Ritz Carlton, you know, the bar is very high. And that's what we compare ourselves to. Yeah, and I think one of the things that we don't wanna send people down the wrong path here is that we're talking about the need for human connections and human reassurance that this is certainly not trust VA, don't use AI versus data interactions rely only on human workers. In fact, that we've actually seen, Preya, in your environment, there have been many cases where you've been able to use AI effectively to give customers what they want and create a better operation. Absolutely. When we think about, you know, when customers are calling in or what they're seeking is the the core details of my trip, right? What time am I flying? When am I departing? What is my boarding time? Like, you know, did I put bags in? Did I put my seats in? Those are core details for a customer at the moment. And then we transitioned into looking at, okay, customers having questions about their carry on luggage, and they have questions about, well, what are the TSA options that I have? Making sure my, you know, my TSA pre check is in there. Right? If I'm traveling with my pets, you know, where do I put my pet? Am I allowed to put them on the seat? Am I allowed to put them under the seat? So there's a variety of different use cases that we've used where AI has helped us, and those are the easy ones. So those are the ones that customers are constantly asking about. So it allowed our crew members to focus on some of the complex situations that that, you know, AI really can't help us with. Yeah, think, you know, the idea is to be one, by virtue of streamlining workflow, getting the right tasks to the right, to an actual employee when they're required, we create more opportunities for them to be human. But we also want to give them more capabilities to be human and that's by surfacing the right information, pulling across disparate data systems, feeding that employee what they need in the right moment. That facilitates that emotional connection. So we can say, okay, AI can handle these issues. If we're trying to really compete not only with our competitors, but with the Ritz Carlton, with the Disney's of the world, we know that we have to do more than just be good enough for these channels, answer the question. We wanna deliver something exceptional, something that's really on their terms. So Preya, we're come back to you for a second because what is the customer trying to get done? What does good look like? And why can AI satisfy that in many cases? Yeah. I think, you know, again, I think it goes back to customers just want that real time support. And what we've seen is, you know, when a customer is is being disrupted and we have to send out cancellation emails or delay emails because of the weather, customers are very much, okay, well, are my options? How quickly can you give me my options? And when a customer feels like they are being disrupted, the first thing we want to do within our space is to give them some control back. And we do that within AI and the self-service technology that we have say, listen, we've disrupted you. You know, unfortunately, we can't control the weather events, but this is what we can control. And this is what we're gonna do. We're gonna give you some control back for you to select the options that work best for you. And when they're coming through our AI bots, they're also looking at customers are also looking at, okay, well, what what am I qualified for? Or where do I go to be able to do that? And our bot not only tells them and shows them what to do, but it gives them the policies that govern their specific situation. Absolutely. It should be working in their interest. Correct. And I think too often, we assume it's only it's working against their interest. It's trying to keep their complaints to a minimal, keep the cost of serving them down, and that can't be the mindset. Because when that's the mindset, you could have the best trained AI in the world, the easiest to use bot in the world, customers are gonna feel disappointed. And to overcome that though, we want our bots to do more than just what yesterday's bots do. We don't want it to be an expensive fancy MTQ page. We want it to take real action, make real connections. Now Chris, this is really where the agentic AI part of when conversation comes in. How are you seeing that evolution? What impact it can make in use cases like what Preya was talking about? Yeah. If you think about the journey we've been on for those two decades I mentioned earlier with IVRs and chatbots, All of these solutions, I kind of bucket them into the sort of infrastructure category. They're all good at doing a thing. You know, a CRM has a very specific purpose but what we're talking about here is not the incremental benefits of these legacy technologies. We're talking about this leapfrog step forward in terms of technology and how it can handle the complexity that Preya is talking about. Because now these agentic systems which to differentiate agentic from generative, generative is very good at handling a conversation. Okay, the bot can handle a conversation with a human in a very human way but it's not enough to drive business outcomes. So generative is an important component from a conversation perspective but agentic is where you integrate with data sources like knowledge bases, FAQs. You integrate with systems like CRMs and all the other tools that you know are part of your probably complex tech stack and that's where you facilitate transactions. This is where the AI can reason. It understands the context of that customer's journey because it's not just this call or this chat that is important, it's what has transpired in the history of this customer's experience and now we can stitch all of this together because most customer journeys are very fragmented. You think about chat is responsible for one thing, voice is often a very separate channel and what happens to customers? They end up having to repeat themselves. They come into the IVR and they say agent, agent, agent because these technologies haven't historically resolved the issues. With agentic AI, with generative AI, you finally have all of the context in real time to respond to the urgent issues that are very complex that the deterministic solutions of the past were so brittle that you would end up in what we would call a doom loop or I would have to abandon that channel and go to the voice channel and say agent, agent, agent. That's the key difference. So as AI technology matures though, we know it's gonna be expected to handle more, it's gonna be expected to do more and to be accountable for. And that makes thinking about the human in the loop that much more critical. Because when you're talking about making high stakes decisions in real time, when you're talking about handling more emotionally charged conversations in real time, you do wanna know that there's a human involved in some capacity, right? So what is your mindset? What is your thought process of where we're going as it relates to having you? Human involvement is absolutely critical. And you know, I'll speak to the ASAPP CXP Customer Experience Platform. It has to be integrated into the systems that you use, your agents use today. So I often run into Salesforce's CRM. You want to have embedded within the CRM an ability for the human agent to interact with the AI. And so if I go back five years, I think about we were so focused on agent assist. We wanted the AI to assist the agent in facilitating the conversation. Well that has flipped. You know, particularly in the last two years, instead of the AI assisting the human, we're now seeing the humans assisting the AI. Because again, the AI is very capable of having a great human conversation and with the right access via APIs and knowledge base through the agentic AI, it can actually facilitate functions, it can facilitate transactions, it can make changes but you don't want the AI to always act fully autonomously. So I think of this as sort of a spectrum and historically we've thought very binaryly. It's either all agent or all automation. And now there's this spectrum where a hundred percent of your interactions can have some degree of automation. There are moments that matter. There are critical thinking moments. There are empathetic moments where a human needs to weigh in on a suit situation. They don't have to fully take over the conversation. They can provide decisions on credits or you know upgrades you know in in the case of JetBlue. So all of these critical moments where critical thinking and empathy are important, I can get in and provide a thirty second decision and allow the AI to go back and finish that conversation. And that's a completely turned upside down way of using AI and technology to handle the vast majority of a conversation. So we all know the cliche or a variation of it, you know, the idea of trying to change a tire and drive at the same time. That, you know, our customer demands don't shut off because we wanna explore new technology. People don't stop taking flights because we wanna highlight a new AI solution for customer service. And so Preya, from your perspective, what have you thought about as it relates to deploying AI without disrupting the need to continue operation to continue to meet customer standards? Yeah. I think the first thing is we positioned AI as the ability to support crew members. You know, there's this preconceived notion that, oh, AI is coming. That means my job is going away. And and that's not necessarily true. I think what we focused on with our crew members is that AI is helping you handle some of these other tasks so that you can focus on these other tasks. And we, as your leaders, are gonna help you navigate through these complex situations through additional training, through additional role playing techniques, right? And that allows our crew members to utilize these skills within their professional life, as well as their personal life. So we positioned it first as a tool of support, enablement for self-service. Then we looked at real life use cases where there was a high impact, there's high volume, but there's low risk. And we had to look at that to look at, okay, we're gonna identify what is that end to end journey for that customer. What is that gonna look like? And we looked at the data behind it. I think the data is is such an integral piece in this in this evolution, because it allows us to look at this is this is where we are. This is where we're gonna go. And we defined several different success metrics from CSAT to FCR, and also looking at the the customer adoption rate. Because I think the trust that our customers have in these solutions is also equally important. So those are some of the things that we did to deploy our AI models without disrupting. And I love that last part because I think we are very vulnerable to what I call mirage metrics in this space. The idea that, okay, containment's up, so we must be doing a good job forgetting that maybe it's because we got rid of the phone number on the website. So that's why more people are using digital chat bot. Or we're not giving them the option to escalate so more things are being resolved in self-service without thinking about whether the customer's happy. Without thinking about whether it's driving lifetime value. Without thinking about what it means for the type of work the agent now has to handle. So I think adapting metrics and really rethinking what this means for your processes, that is so key. But the reality is it's not just about deploying AI, deploying technology, and what it means for customer and employee adaptations. There's also the fact that we're not deploying it in an imperfect, we're in a perfect system. We know that our databases are not always properly integrated. We know we may have a very complex tech stack. We may not have the ideal framework. So Chris, from your perspective, what do we do to sort of simplify this process and ensure that all those fracture points, all those breaks are not gonna ruin the ROI to regenerate from this tech? Yeah. I think, you know, the six years I've been at ASAPP, we've been very focused on meeting customers, our customers, largely enterprises where they are. There's always a complicated tech stack. There's always complicated, sometimes broken processes. It's not enough to deploy great technology into a bad process. There's examples of that everywhere and you will not realize the ROI if you apply great technology to a broken process. So I think it's taken a more holistic view of the complexity of the all of CX. You know, let's recognize the the capability of the AI of the new technology, let's also recognize the business reality that we have to look at the processes. We have to look at the people factor, you know, and how can we really map out those customer journeys in a clear and honest way to say, here's where the fragmentation and the friction exists in our business today. How might technology address that? But also how might processes need to change? Yeah, and I think having an organization, a component of your organization that looks directly at the impact of AI is so critical. Because even if you have the perfectly trained model that does do what it's supposed to do on the front lines, it does absorb certain issues, that is changing the dynamic. It's changing the types of workflows you're gonna have. It's changing the way customers behave. And if you don't have a part of your organization that's constantly looking at that, constantly getting rallying stakeholders together, building the right collaboration and transforming accordingly, then yeah, you're gonna have a great initial pilot phase, but you're not gonna have that great long term impact. You're not gonna evolve over time. You're not gonna turn the technology into everything it can be. And that's why it's so critical to have a part of your team really zeroing in on what's working and what's not, and then democratizing them across the organization. Now, we wrap up here, is so important to remember that whether we're talking about a new policy, whether we're talking about a new technology, there have to be guiding principles. There have to be commitments you make to your employees and your customers that are not wavering, that don't break because we're in experimental phase. So Preya, starting with you, what do you see as those foundational principles that really should guide any tech decision we make here? Yeah. I think first and foremost is understanding the customer's problem. I think we need to identify what is the actual problem and not start with the technology. Technology is a piece of it, but it's not the whole it's not the whole pie. So we really need to look at what is that customer problem and and how do we go about looking at within our means of fixing that. And again, just like Chris mentioned, it's our policies that govern it. It's our processes, but it's also the people. How are we training our people to be able to help those those customers navigate through those journeys? And I think the last thing is just building trust with our crew member, our our customers and crew members. From a crew member perspective, making sure that they understand that this is a tool for support. This is a tool for assistance. But for our customers customers, making sure they understand this this AI model is going to help you get what you need in the speed, the transparency that you're looking for, and the resolution that you want. This tool is gonna be able to help you do that, and I think that goes such a long ways with our customers. And I'll tell you right now, with our model that we have, our customers actually believe that the AI model that they're speaking with is an actual human being. So I think that the work that we've done to enable our AI model to learn from top performing crew members has helped our journey with our customers feeling that trust and feeling that support. Yeah. I know that there's gonna be a lot of debate about whether, know, you need to be transparent about whether they're talking to AI, whether you don't. There's a potential down the road. We could be looking at policies that affect this. But I think what's really key there is whether your bot is literally mirroring human conversation can be mistaken for a human or not, what is important is that they don't feel like they're trading a great experience to talk to that AI based team. Absolutely. That they don't feel like they're sacrificing a personalized resolution, what they're looking for in that moment. And when they believe it's just as accountable, just as faithful to the brand's mission, just as much an advocate for them, they will be successful, they will trust it, and you will get the adoption you're looking for, and then all the employee facing benefits come to fruition as well. Chris, same question to you. Any guiding principles that our audience here should go home with and say, no matter where I am on my AI journey, I know that it's taking me closer to a powerful connection with customers. Yeah. I think, you know, I work with companies of all sizes across virtually every industry and the commonality I see and probably the mantra I would use is go slow to go fast. We don't like to do that you know and think it's so important where we're at this inflection moment to look back you know, everything we've talked about with the legacy technologies. Why haven't they delivered on the promise? Why are we still making these huge investments in CX? The customer satisfaction hasn't moved, you know, in a positive direction along with it. If we look honestly back at that, there's still so much fragmentation and friction in the customer journey. Customers don't get first contact resolution. There is no omni channel despite, know, we've been talking omni channel for two decades. You know, why haven't those things come to fruition? And so if we look at those things honestly, those honest realities, I would offer three things. I would say first, as an organization, I would have stakeholders from every part of your business, particularly the customer facing stakeholders. You know, there needs to be sort of what I would call an AI governance council because you don't want the contact center with their own AI, marketing with their own AI, sales with their own AI. Every business unit has a different AI solution. Your customer fragmentation will not go away. You'll just have a modern version of customer fragmentation. And so how do you get all of these key stakeholders in your business beginning with the customer journey in mind, backing into the second thing, out those customer journeys, being real honest about where things are breaking and then deciding is that a technology fix, is it a process fix, How do I create the most seamless, wonderful customer experience possible using the technology but not ignoring the processes that need an overhaul? Our workflows can be automated but there's a lot of other factors. And then the people side of it. Preya mentioned it. Don't ignore the people. We need to talk to people about at every level what is our AI strategy. We want everybody rowing in the same direction. This is from, you know, we talk a lot about frontline agents and human in the loop and, you know, enabling them to supervise the AI. But think about every other job that is going to change when all of us are using AI. You know, with the ASAPP CXP, you have the ability to create flows, lower no code tooling. So you don't have to have software engineers to get started with AI. You can start building out these generative flows now with the people that you have. I think about the QA analysts that were part of the last discussion. We have an opportunity to allow them. Nobody knows the business. Nobody knows the customer experience better than those QA analysts. Why can't we have them running simulations before you put automation in front of the customer? Use an AI platform to simulate what that customer experience will be like. Make sure you have that thing nailed down and creating a fantastic resolution before you put it into production. All of this is capable today but we really need to think about the people side of the equation and how will jobs evolve less than jobs be eliminated. How do we change the conversation and bring everybody along with us? Amazing insights. You know, will say JetBlue was the very first My very first flight ever was on JetBlue. I get to work with ASAPP a lot collaboratively. Great organizations, Great people here. Check out the ASAPP booth. Check out what your customers want and build AI accordingly. Let's give it up for our speakers here.