Transforming Hospitality with Real-Time Data and AI Agents

Summary

United Airlines and Chick-fil-A built AI and data strategies on Databricks around four pillars: AI-assisted data engineering, data infrastructure designed for AI agents, reliability and governance, and empowering their teams.
A connected data fabric that unifies booking systems, loyalty programs, contact centers, and operational platforms enables real-time AI agents to handle customer service at scale and reduce resolution times by 25 percent.
The anticipation engine architecture turns fragmented hospitality data into unified customer context, enabling proactive service such as automatic rebooking during disruptions before customers are even aware they are affected.

Transforming Hospitality with Real-Time Data and AI Agents

Building great hospitality experiences requires real-time understanding of customers across fragmented data sources. When customer information is siloed across booking systems, loyalty programs, contact centers, and operational platforms, hospitality brands can't deliver the personalized, anticipatory service that modern guests expect. Databricks' unified data platform enables teams to connect these fragmented sources into a single source of truth.
Hear from United Airlines and Chick-fil-A on how they built four core pillars for AI and data success: AI-assisted data engineering, data infrastructure designed for AI agents, reliability and governance, and empowering their teams. Learn how connected data fabric enables real-time AI support agents to handle customer service at scale, reduce resolution times by 25%, and restore trust during disruptions. Discover the architecture behind anticipation engines that turn unified data into customer loyalty.
🤝

Chapters

FAQs

What four pillars does United Airlines use for their AI and data strategy on Databricks?

United Airlines built their AI and data strategy around four core pillars: AI-assisted data engineering to accelerate pipeline development, data infrastructure designed specifically for AI agents, reliability and governance to ensure trustworthy outputs, and empowering their teams with the tools and skills to build and maintain AI-powered products.

How does a connected data fabric help hospitality companies deliver better guest experiences?

A connected data fabric unifies customer information siloed across booking systems, loyalty programs, contact centers, and operational platforms into a single source of truth. This unified view enables real-time AI agents to understand a customer's full context before and during interactions, powering proactive actions like rebooking flights before a customer is aware of a disruption.

How do real-time AI support agents reduce customer service resolution times in hospitality?

By grounding AI support agents in unified, real-time customer data, hospitality companies equip agents with the full context of a customer's bookings, loyalty status, and history before the interaction begins. This eliminates repetitive information gathering and allows agents to propose solutions immediately, contributing to the 25 percent reduction in resolution times described in this video.

What is an anticipation engine in the context of hospitality AI?

An anticipation engine is an AI system that uses unified customer data to predict what a guest needs before they ask, enabling the proactive service that defines memorable hospitality. Examples include automatically rebooking flights during disruptions, surfacing relevant loyalty offers at the right moment, and personalizing interactions based on past preferences—all powered by the connected data fabric built on Databricks.

Full transcript

[00:08] Good afternoon everyone. Are we having a great time at Summit? There we go. They said, "Hey, do you want to talk at 5:20 in the evening?" And I said, "Sure, everybody wants to listen to Lorraine Bacon." So, I am Lorraine Bacon. I lead our uh retail, consumer goods, travel
[00:26] and hospitality industry for our go to market ecosystem. And really what that means is I get to talk to a lot of people about a lot of things. And I get to talk today about two of my very very favorite brands. So let's start with a simple observation, right?
[00:42] The best hospitality has always been about anticipation. I want you to think about that for a minute. Where do you get your best feelings for hospitality? Right? Long before AI, the brands that we remembered were the ones that understood us well enough to anticipate
[00:58] what we needed next. Think about when you travel, right? It's that hotel that remembers your preferences. Do you want those fluffy pillows or not? It's the airline that will rebook your flight for you when the disruption happens before you even know it's happening. It's the
[01:15] restaurant that knew what mattered to you before you had to ask. And at its core, hospitality is about making people feel known. It's one of the last real uh areas where people can feel like they matter to every person they interact
[01:30] with. And that's really what the point is, right? It's the preferences, it's context, it's intent as you deal with people. Now, we see that through different ways. We see it through personalized offers. Have anybody ever had a conversation in the car and then
[01:47] all of a sudden your algorithm is popping up with all of your new personalized offers? It's real-time decisions. It's seamless experiences, but really overall the question is how do you get from the person to the context and make that that special
[02:02] individualized experience? And so this is where we're really seeing a huge shift in AI right now. It's going from assistants to agents and a lot of places are looking at it in the back end, right? How do we handle it in the back office? What is happening there and
[02:20] really where does AI create the most value and what does it have to do with the data that's underneath to work for it all? I didn't know that the words were going up there. So a lot of convers a lot of
[02:35] conversations are happening with vendors with suppliers with companies that are focused on conservative answers to the AI question. Okay, it's talking about supply chain optimization. Anybody who gets to talk to me oneoff, I'll talk to you all day about this. I love supply
[02:51] chain optimization. I love demand forecasting. Right? These are safe places to actually do this back office automation. And it's really about discipline there. And we want to show how we can demonstrate value in those controlled environments and really talk
[03:08] about building governance before scaling. Governance is something you're going to hear about over and over again this week and you probably hear about it over and over again every single day, right? But really, it's about proving their reliability before putting AI in
[03:23] front of a customer. That's what it is. And for most industries, that's the right approach. But I challenge you. Travel and hospitality is different. Travel and hospitality is about the human, the person behind the personalization. And this is one of the first industries
[03:40] where we've really seen AI move from the back office and become part of that customer experience itself. So that every person in this room, you probably have had an AI mediated experience while traveling just here, right, to this event. Uh there's a chatbot that might
[03:56] rebook your flight. Uh We have a lot of flight delays lately with the different airports and it's the airports. It's not the airlines, guys. Uh there is a recommendation engine that says, "Hey, would you like to take this hotel offer instead of that one?" It's
[04:11] it's really where this particular sector has moved AI to the front of the house and it's faster than any other industry. And it's not a prediction. This is what customers reality is. This is what customers demand. They want to be known and they want to be serviced very very
[04:28] quickly. And some of those experiences have been fantastic. They're remarkable. They're wonderful. And then there's the other experiences that we're not going to talk about today where we get frustrated and you know representative representative representative only now we're typing it representative into the
[04:44] chatbot. So it's signals. These are all signals and they're telling us what we're missing. And so AI is really giving us that surfacing of the customer experience and transactional AI is really handling that request that
[05:00] arrives, right? So it's understanding enough context to address the need before the request is made. And it's a difference in how you're leveraging that data. So you can't personalize an experience when customer information is fragmented across the different loyalty
[05:18] programs, reservation systems, mobile applications, marketing perform platforms, operational systems, and conra contact centers. It's a data infrastructure problem before it's an AI problem. And so really
[05:33] taking a look at that real-time visibility is what matters. And so today's conversation is key. It's about using data and AI to move from understanding what happened to determining what should happen next. How are we going to go forward in that next
[05:49] step with customer journey? And United Airlines and Chick-fil-A are totally different, right? They operate in different environments, but they're the same. They're all about the customer. They're all about the experience. They're all how does this matter to each individual's life. So they're going to
[06:05] talk to you about their different customer journeys, their different operational realities, but at the foundational layer, they have made the same decisions. They have decided to build the data architecture that makes anticipation of what people need possible. And what you're going to hear
[06:21] today, it's not a road map, right? It's how they have unified the powers that the data that powers the customer understanding and how they're extending that human touch through data, through context and intelligent action because
[06:37] ultimately the challenge is the same for every single person in here for the hospitality trends, right? How do you make millions of customers feel like they're your only customer? So now that's enough out of me. Let's go ahead and hear from the people who are
[06:53] actually building the answer to that. I'd like to go ahead and welcome up our United Airlines team.
[07:55] Thanks Lauren for that wonderful introduction and setting the stage. You might have noticed 32 billion. Isn't it a great investment that United is looking for? A great flight is not just about going from point A to B. The real journey begins the moment you start planning
[08:11] your trip. Let me paint a picture. I'm a mother traveling with family. And I'm a business executive trying to maximize every minute possible. My priority is simple. I want day time flight. I want to make travel easy for
[08:26] kids, no connections and of course easy on budget. And I'm thinking do I have direct routes? Can I work uninterrupted? Are there any premium seatings available? Yes, that's where our network, expanded network and new routes comes in.
[08:44] Now on board, a quick show of hands. How many of you have experienced our inflight Wi-Fi entertainment system? With Starink, you can now stream, browse, stay connected all in the air just like at home. So no more need to
[09:02] download Wi-Fi like load videos. Wi-Fi is all set for you. Exactly. And I can follow my favorite FIFA matches right from the air. Along with this, we are also investing and exploring on bettering our physical
[09:17] experiences. Polaris premium cabins, seatback screens for every customer and large overhead bins. So, no more lastm minute gate check-ins. And even the smallest thing matters. Bringing back customer favorites. Troop waffle. We
[09:34] call it fleet modernization. But our customers say the flight just felt a little better today and that's the goal. No isolated improvement. We are talking about one flawless customer journey.
[09:53] So we are here today to share our experience how we using data and AI to transform customer experience. I'm Charuta. I'm Anka Garval. I'm Rea Bata. We will walk you through our vision, challenges, and our road map.
[10:11] Numbers talk for themselves, but still to give some context, we are global airline serving six continents, more than 150 million travelers every year. With 100,000 strong employee team, we are committed to give the top tier
[10:28] customer experience. Now that we have reached cruising altitude, sit back, relax and enjoy this data journey with us. Whatever we just saw is not by chance. It is with deliberate AI strategy.
[10:46] In our industry, conditions change by a minute. If data is delayed, we might take wrong decisions. That's why real time and trustworthy data is our foundation. Data alone is not useful. We have to implement our strategies to have
[11:04] the best possible tools. Self-service is the way forward. In this world, we cannot really take security as afterthought. It is very much part of our development cycle. We repeat this cycle again and again so that we can
[11:19] provide easytouse tools for our employees and even for customers. Okay. Thanks Aruta. So let me just take a step back for a moment. So if you think about the last wave of transformation, so it was all
[11:36] about the automation, making processes faster, reducing the cost and improving the overall efficiency of the system. But the talk of the town is agent A right now. So the system that do not just follow
[11:52] rules but go beyond the automation. system that can understand the context from your data, reason it and then take an action accordingly. So I'm really excited to share about the data and AI strategy uh of ours. So f it's built on four key pillars. So first is AI
[12:09] assisted data engineering. So by embedding AI everywhere and using agentic coding tools, we are simplifying and accelerating how do we do data engineering stuff all together with respect to how do we do pipeline development, how do we create an
[12:25] observability, how do we deploy our pipelines and you know how do we resolve our pipeline failure in the earlier stage of the pipeline development itself. So at the same time we are also using the similar momentum using AI to
[12:40] modernize our data platform towards the datab bricks lakehouse architecture. The second pillar is data infrastructure built for AI. So we started this journey couple of years back when we started preparing data for AI. So the challenge
[12:57] was at that time how AI can help us to you know to join the data in real time to do the aggregation in real time but at the time LLMs were not matured enough to able to do these kind of a stuff.
[13:13] Hence we started with the concept called connected data fabric that what that means is we connect the data which lies in the domain or across the domain together and building a context layer on top of that so that AI can understand the context in a much better way and can
[13:30] provide the accurate answers. So this is all about creating a a data for AI. The third aspect is reliability. So in our operation we support our customers 24 cross 7.
[13:48] So the reliability and the trustworthy data is the most critical aspect of the platform. Hence we started enabling few of the in-house tools we call it anomaly or detection into the place so that we are not compromising on the data quality
[14:05] aspect of the platform. The fourth pillar is obviously people. So we are we are empowering our people to upcale themsel uh into AI and other tact tooling so that they can innovate and build faster as well. So if we stitch
[14:21] all of this together that comes to the foundation which is our datab bricks lakehouse foundation. So I'll just quickly walk you through our about our data architecture. So here we have we are ingesting data streaming data from Kafka batch data from different
[14:37] databases. So the data sitting in our operational data stores like the data coming from booking system the reservation system crew data passenger data the check-in information the baggage related information so on and so forth. So we are bringing all of this
[14:53] data together into our unified platform which is database data and AI platform. We are processing this data, transforming this data and creating an open data format with data lake and Apache iceberg based uh data layer. The
[15:09] data finally gets stored in Amazon S3. So all the way from raw to curated uh to create a connected data fabric. All of this have happening inside the data bricks platform. So we are using uh spark structured streaming to connect
[15:25] the data in real time with the batch data what we have we're also ensuring that bringing unity catalog into the picture we are having a proper data observability data governance built into the platform itself so not we so that we
[15:40] are not compromising on onto the trusted data. So making sure that while we build while we transform we are creating a trusted data out of the platform and on the right hand side the connected data will be consumed through our different applications like contact center uh uh
[15:57] agent applications. The same data is being fired to our reporting and an analytical use cases and the same data is being fired to AI and ML models so that they can train on the data itself. Great. Thanks for that architecture
[16:14] review. Now let's see how this actually works in real use case. Remember last time you traveled maybe planning your trip to Hawaii. Your journey is a series of interconnected events which leads to one memorable experience. Your journey
[16:31] begins way before maybe sitting on your couch and browsing through different trip options. Next you are at airport handling bags and security lines. Then you want a seamless transition to flight. You want to sit back, relax and
[16:47] enjoy your journey. Finally, you grab your luggage and off to your destination. Every step should feel like perfect 10 out of 10 customer satisfaction. A single mishap at any touch point is going to ruin your
[17:03] experience. For you, it's one story. But behind the scene our data is fragmented in all these different different operational applications. For customer experience we can't allow our data to be in isolation.
[17:21] Data engineering team breaks this fragmented data and builds one connected layer a clear view. We call it data fabric. Now we can power our applications with real time and crossdain data.
[17:43] Whenever disruptions happen, people generally want to call up to the contact center as soon as possible. At that time, they don't think which channel do I choose. They just think I need some help. It has al also happened that people call up from their phone, chat from another device and are constantly
[18:00] posting on social media from another. Now this at United scale becomes 80,000 calls, 30,000 messages and 20,000 emails every single day. Now with this amount of data and these
[18:15] amount of loads that are coming in, what our customers generally expect is fast resolution. They want their answers quick. They want they don't want to repeat themselves over and over. They want a full context driven agent sitting
[18:31] right to help them up. And for our employees, the agent is looking for full context. They want to understand what what's the problem going on with the customer. They want to investigate as quickly as possible. This is where our tools come in handy. They have full
[18:49] context. They know what was planned, what went wrong. This way, they're able to help the customers much sooner. So having more channels is not really the agenda. The agenda is having a
[19:04] having a multi-dimensional unified intelligent data platform that helps them answer to questions as quickly as possible. Heart of customer satisfaction is one thing, reliability. No delays, no
[19:22] cancellations. We call it blue sky day. Everything goes as per plan but behind the our operation is orchestration of continuously moving parts. Flights change. Weather impacts. There is a crew
[19:37] rostering. Network is a massive puzzle. Flights fly from city to city. A single delay is going to cause a domino effect. Same goes for our crew members. They have limitations on the flying
[19:54] hours. The legal bounding that also needs to be considered. The most important piece are passengers. 40% of our passengers are on connecting flights. They have to make to next destination.
[20:09] Just think about planning a road trip. Family of four getting bags packed, car is fueled, finding the perfect route through traffic needs lot of planning and coordination. Now just imagine doing the same on a global scale for airline
[20:27] industry with thousands of flights with hundreds of passengers on each. Very rightly said Charata. It is indeed a massive puzzle to solve. 3,500 flights on a daily basis and 400,000 passengers
[20:43] every day are carried. Now there are chances that disruptions happen. So let's introduce the evil to the room. We call it IROPS, the irregular operations. Now there could be a situation where
[20:58] weather hits at a particular hub. The ATC is holding on to flights. At the same time, a geopolitical situation arises. Now with all of this happening there, we could also run into maintenance issues and crew availability
[21:13] constraints. Now imagine a snowstorm happening at NEWAK. Now this is not just a problem for Nwak. It becomes a cascading problem throughout the network. Now these delays that are caused can
[21:28] happen in two ways. They can either be controllable where we could have intervened and cured but there are some um delays which are uncontrollable and these when issues arise needs to need to be solved and need to be solved quick.
[21:50] Do you remember your experience during the winter storm you just told me about? You really had to remind me about that. Uh I was returning from DC with my family and a flight was abruptly cancelled. Though I've been working in airline industry, I do understand
[22:06] weather is uncontrollable event. But as a standard passenger, no logic works. I was just frustrated mother with very cranky kid and yes impatient husband. I'm sure you all must have experienced
[22:21] the same. If you have a teenager or a pet, I'm I'm sure the complexity would be more. And this is where the decision making comes in most important. This is where we make tools. We make use of tools such as connection saver. This helps us track
[22:39] whether do whether we hold a flight for a few extra minutes so that our passengers can catch on to their next connecting flights. How do we strategically cancel flights to curb the rolling effect that's happening in the system? Now, with all of this, we also focus on how we can push in transparent
[22:56] notifications to our passengers so they're not left in the dark. With this, we also have a few more features for our passengers such as agent on demand. With that, a passenger when stuck in a case of disruption, they
[23:11] can directly call an agent right from their phone live. We also, as she said, we also try and wave off the change fees. so that rebookings become easier.
[23:28] So when I was actually stuck, I was just looking at the fellow pastors. Someone might be missing the most memorable meeting or might be the once in a-lifetime moment. Someone forced to stay overnight completely unplanned. That time I realized something important. I ops is not a metrics. It is
[23:46] about those human moments that matters. We are not recovering operations. We want to restore the trust. Our leadership has very strong principle which says it might be someone else's problem but it is our responsibility to
[24:02] fix it. These are the moments where we can do the difference. These are the times where our data and AI tools can really help. Okay. So the idea is uh here is very simple. create a AIdriven
[24:19] data products which connects data in real time and break the silos. So on the left hand side we are powering our data to contact center agents or the customer engagement tools all together and so that they can do quicker resolution in a
[24:35] more personal personalized manner. On the right hand side we are using the same data to in the analytics to be able to understand the trends the patterns the customer interaction in the past so that you know we can understand what is
[24:52] working right or not so that we can improve our data products and that further helps in improving our customer experience altogether. When a customer does a booking, there are three things that can happen. They
[25:08] can either have a happy path and they can fly as it was planned. They can run into a cancellation or they can rebook their flight. Now, let's understand what is a line of flight. Let's say you booked a flight from Boston to Dallas via Chicago. That becomes your onward
[25:26] flight. And when you're flying back to Boston, that becomes your return flight. Now if your journey goes as per plan then what whatever was intended to fly gets flown actually but let's say the weather hits and now you have to fly why
[25:42] Houston instead now this becomes disruption this is where uh your actual journey differs from your intended journey.
[25:58] Now imagine an agent having to go to multiple places to understand this situation. They get one data point from one place, another data point from another. This is where connected data fabric comes in and they it helps to give a centralized platform to understand the entire situation.
[26:14] Without connected data, they have to visit multiple different locations to understand this. It delays their decision making and the passenger has to repeat their story every single time. But with connected data they get to know what was planned, what actually
[26:31] happened, what was the delay like and what was the overall impact. Now instant clarity leads to instant resolution. Now let me bring this down to what matters the most and those are the outcomes. We have seen a 25% reduction
[26:49] in case resolution time. This means the agents are able to access everything quickly. The customers are able to get their answers at a much quicker pace. We also see a 20% reduction in automated case 20% increase in automated case
[27:06] resolution. Now, this does not mean we are replacing agents with AI. We're just letting AI handle the repetitive work and we're focusing our agents on more empathetic solutioning. And most importantly, this runs 24/7. As
[27:23] we all know, disruptions don't really follow business hours. So, this really helps. It's any time of the day we get our answers at United Scale. This trans this translates into faster resolution, less customer effort, more resilient
[27:41] operations during peak demand. So the takeaway is simple. When you connect real-time data with AI, you just don't drive efficiency, you enable speed, scale, and much better human experience.
[28:00] Ultimately, goal is simple. We want to reduce friction and preserve our brand loyalty. We have our data. We know how to join it. We have built AI tools which are in action. But that's just the beginning. We are collaborating with data bricks to
[28:16] improve further and transform our customer experience. Most of you here must have traveled United to join this conference. We really appreciate your business and next trip do plan to fly United. Make most of it.
[28:33] And every time you fly, remember behind the scene data and AI is working for you 24 by7. Thank you. Thank you for your time. Thanks.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.