Adobe Summit: Virgin Atlantic's AI Concierge and the Future of Travel

[00:00:00] - [Speaker 0]
Regular listeners will know I always say that I only partner with companies that align with my values and what I'm trying to build here at Tech Talks Network. And NordLayer fits that perfectly by helping businesses stay secure without adding unnecessary complexity. And every company today relies on SaaS tools, whether it's collaboration platforms, CRMs, or finance systems. The common denominator here is everything now happens in the browser. But here's the problem.

[00:00:29] - [Speaker 0]
Most browsers were never designed with business security in mind. This is one of the many reasons why NordLayer has introduced a new business browser that puts security, visibility, and control directly where work happens. And it helps protect against things like phishing, malicious sites, and unauthorized access, while also giving your teams a safer and more controlled way to work online. And the best part is it doesn't change how people work. It just makes what they're already doing more secure.

[00:01:02] - [Speaker 0]
So if you wanna learn more, please head over to nordlayer.com/browser and see how it fits into your workflow. But now it's time to get today's guest on. What does it actually look like when AI moves beyond experimentation and starts delivering real business value? This is a question that comes up in almost every conversation I'm having with business leaders right now. Yep.

[00:01:28] - [Speaker 0]
There is no shortage of ambition. There's no shortage of investment. But over the last few years, we've seen a gap between pilots and measurable outcomes. And it is that gap where many organizations are getting stuck. So today's conversation is coming to you straight from the show floor of Adobe Summit here in Las Vegas, where I get to sit down with Neil Letchford, vice president of digital engineering at Virgin Atlantic.

[00:01:56] - [Speaker 0]
And this is a company that has been quietly been doing all the hard work over the last four to five years, rebuilding its digital foundation, aligning teams, and creating the conditions where AI can actually deliver value and deliver it at scale. But this is not a story about chasing technology or the next shiny big thing. It's about solving very real problems from launching an AI concierge that helps customers navigate complex travel decisions to driving measurable revenue through personalization. Virgin Atlantic is currently showing what happens when you combine the right idea with the right operating model and the right mindset. So in our conversation today, I wanna unpack what it takes to move from proof of concept to production and understand how Adjenting AI is already shaping customer journeys and understand why leadership, autonomy, and clarity of purpose, how all these things matter just as much as the technology and the announcements themselves.

[00:03:01] - [Speaker 0]
So with that scene perfectly set, let me beam your ears directly to Las Vegas where you could sit down with two neals, one a tech podcaster and one the vice president of digital engineering at Virgin Atlantic. So a massive warm welcome to the show. You're joining me here at the Adobe Summit in Las Vegas. Tell me a little about who you are, what you do, and what brings you to the summit this year.

[00:03:25] - [Speaker 1]
Well, nice to meet another Neil in the technology industry, Neil. So I'm Neil Letchford. I am the vice president of digital engineering at Virgin Atlantic. So I look after all customer facing digital experiences at Virgin, and Adobe is a is a key partner in how we deliver those experiences. Basically, have the whole Adobe stack from analytics to AEM, AP, etcetera.

[00:03:50] - [Speaker 1]
So we're here now thinking about what's next in that stack. I'm really excited to hear what Adobe are gonna be talking about at Summit this week.

[00:03:58] - [Speaker 0]
And I know you're gonna be talking about a few things that you're gonna be revealing here today, and we'll get to that in a moment. But tell me a bit more about the journey that you've been on because I I've got some big stats here from, I think it was last year, where you'll the abandoned browse use case reportedly launched two months ahead of schedule, generated 1,200,000 in revenue all within twelve weeks. And the reason I wanted to start with this, I think whenever we talk about AI at a tech conference, the big problems that people talk about is stuck in pilot purgatory, ROI, and all that kind of stuff, and you've kind of been ahead of the game on this. So tell me where you've come from and the journey you've been on.

[00:04:32] - [Speaker 1]
Yeah. I mean, over the past, I'd say, four to five years, we've been on a traditional digital transformation journey. So when so we talk about digital transformation, you think about legacy systems, you think about decoupling, and then you think about enabling in in your experience layer. And really that's the journey we've been on. So we've been building using Adobe in that experience layer to give us the flexibility to to build those personalized experiences for our customers.

[00:04:56] - [Speaker 1]
Abandoned brand browse is a is is a great example. But as I say, we use the stack across the board from AB testing for all our content, for analytics, everything. And really, this is the culmination of that, those four years work. This is where we are today, but now we're really looking to push on to what what does the next four or five years look like.

[00:05:16] - [Speaker 0]
And although we are sat at a tech conference, as a Virgin customer, think it's always known as delivering a very human and memorable customer experience. That always seems to be at the heart of everything. So when you look at digital transformation today, how do you make sure tech strengthens that emotional connection rather than making travel feel automated and almost transactional? Is that a tough balance?

[00:05:35] - [Speaker 1]
It isn't. It isn't. I mean, Virgin strives to be brilliantly different is how we talk about things, but at our heart, we are a human company. And when we look at our data points and we look at our NPS, we know that the the highest NPS and best touch points in our customer journey are when customers get to talk to our people. And it's our people that make that Virgin Atlantic difference.

[00:05:57] - [Speaker 1]
And I'm sure you probably you probably feel the same when you step on board that Virgin Atlantic aircraft, you know, you get that special it's a different feeling than flying on some of the other the other airlines. And so our challenge really is how do we bring that that experience or how do we bring that learning into our digital channels? And so we think about things, obviously, brand experiences, we think about imagery with typography, but but mainly things like tone of voice is really important as well and how we communicate with our customers, trying to

[00:06:24] - [Speaker 0]
bring that that virgin flare and virgin tone through into our digital channels. And Adobe describes personalization as Virgin Atlantic's North Star, but what does personalization actually mean inside the business? And one of the reasons I ask that is when I sit on the plane, I do this a lot for traveling to tech conferences, I'm the the business traveler, but I'm surrounded by families and people excited that have have saved all throughout the year to take that big trip and that once in a lifetime holiday. Again, personalization is different to different people there with your customers.

[00:06:53] - [Speaker 1]
Massively different. And also Virgin Atlantic is a long long haul only airline. We also not only sell flights, we sell holidays. And you're right, we have frequent travelers like yourself, you travel on on on board, but we have lots of people that travel maybe once every couple of years. And when they do travel with us, it's a big purchase.

[00:07:11] - [Speaker 1]
It's a big money purchase. And and especially when you're with your family, you know, it's about building experiences, personalized experiences for those segments of people to really smooth them through their journey all the way through from sort you know, booking that ticket, but certainly at the airport, onboard the aircraft, and in resort. So we try and segment all our customers like that, and then we try and build the right, you know, experiences for them based on that.

[00:07:33] - [Speaker 0]
And, of course, here at Adobe Summit, the big theme is AI and agentic AI. There's gonna be some big announcements, but I'd love to dig into the kind of difference AI and agentic AI is making to you. Tell me about your story with your journey with AI and Adobe.

[00:07:46] - [Speaker 1]
Yeah. So Virgin Atlantic at the beginning of the year, actually December, we launched our AI concierge. This is a multi agent agentic system to help our customers through the customer journey. We built it in over the summer last year and it was mainly focused as a proof of concept on could we help customers find help content and information easier and test it against our existing chatbot if you like to see what we could do. Super interesting and we found the results were transformative really.

[00:08:19] - [Speaker 1]
Customers could really find straight away information they were looking for. They were getting much more clear information and we were able to fill much more questions, much less dead ends, and it really opened our eyes to say actually we've got we've got something here that we we could build upon for our customers. It was also super interesting because it sort of surfaced some problems in our own area because we gave concierge access to our website to find information. And what we found is when people were asking concierge questions, it would go off and find press releases from, you know, ten years ago that weren't relevant. So very quickly we had to we came up with this idea of knowledge over data.

[00:08:59] - [Speaker 1]
So we we we think about, okay, what is the knowledge we want to provide for our concierge? And also then how is that knowledge used across the rest of the enterprise? So for example, you phone up the contact center, you wanna know that the the information you're getting there is the same as what's on the website, is the same as what concierge has given you. So that was kind of the the first key thing we learned. But then that really unlocked an idea about, okay, you know, now we've proven the technology broadly works, where do we take it from here?

[00:09:27] - [Speaker 1]
And again, we looked at our data points to see where were some high friction points that that that we could support our customers with. And there was two that stood out, flight searching and holiday booking, holiday searching. And holidays are actually really complex for customers to book. We actually have retail stores at Virgin Atlantic where customers can be can come and see a consultant to to book a holiday. Again, if you're going to one of your two year holidays away to Orlando, big money purchases, you wanna get it right.

[00:09:57] - [Speaker 1]
So we thought concierge might be able to play a role there. So we spent a lot of time looking at the data. Again, biggest basket value, best was coming from our our retail stores. So we sat sat in our retail, our digital teams went to the retail stores, sat sat with the the people there, and we sat in on interviews with people as they were booking the holiday. What we did is distilled the information down to seven key bits of knowledge that the agent was getting from the customer and we called this minimum verbal knowledge and what we then did is codified this as part of our agent so that our agent, when you talk to it about holiday booking, will try and find out those in conversation, those seven bits of information so it can recommend your holiday.

[00:10:38] - [Speaker 1]
And it's been incredibly successful. It is selling holidays, it is selling flights, and it's been a really surprising people really enjoy the holiday planner, which is what we were not expecting that as an outcome, but but super interesting.

[00:10:52] - [Speaker 0]
Wow. That is absolutely incredible. How long did it take you to go from a proof of concept to then getting out of pilot purgatory, get it out of there, getting it live?

[00:11:00] - [Speaker 1]
Yeah. I yeah. I mean, end to end, I think we got to the proof of concept done in September, took two months and we finished it in September. And then September to December was, you know, testing it and productionizing it, ready for deployment, and we deployed it over Christmas in December.

[00:11:18] - [Speaker 0]
Wow. And for business leaders listening that maybe wanna follow in your footsteps and learn how you work with Adobe and the different technology that you're using there, Any lessons learned along the way? And and what's the initial fear about but but what if this agent goes rogue and says something it shouldn't or hallucinates?

[00:11:33] - [Speaker 1]
That's a massive problem. Right? So, like, we think about one of the key things we considered, there was two elements. Right? One is observability as a big thing we worked on.

[00:11:43] - [Speaker 1]
So it's key to us that we could observe every step of an agent's interaction with our customer. So not just the conversation itself, but all the API calls it's making in the background to find knowledge, to call an API, to recommend a holiday. So we had full traceability and transparency of of of of the flow. That was super important to us from a support perspective. And then the other thing we work really hard on is is what's called evals.

[00:12:10] - [Speaker 1]
So then if you come across evals, it's a machine learning term for basically evaluating a model. So to determine if it changes over time, effectively, you just have a bunch of standardized questions and answers where you know the answer to, you can evaluate the model against your evals to determine how it's performing over time. Obviously, you know, small things like, you know, changing a prompt slight assistance prompt slightly in the background can have big impacts on your evals. Adding a new agent in the mix can have bigger impacts as well. And obviously, the model upgrades.

[00:12:40] - [Speaker 1]
I mean, one of the biggest learnings was the the model upgrades themselves increased the, you know, ability of the concierge over time. So as the models improve, we got a lot of brackets free intelligence off the back of that, which was which has been amazing to see. So for us, getting in early has paid dividends.

[00:13:00] - [Speaker 0]
And we hear a lot at tech conferences the phrase human in the loop. If people do need to speak to a human, is there something in place to guide them straight to that human as well?

[00:13:09] - [Speaker 1]
Yeah. So that's something we've implemented recently. So that's integrating with our contact center. So our contact center have a a web chat feature. So you can hand over to the contact center and they can pick up a chat.

[00:13:21] - [Speaker 1]
So we've launched that integration two weeks ago, and again, some really interesting insight there. So the LLM or our agent will detect will infer how the customer's feeling, and it will use a number of indicators. So for example, it might be sentiment is the language, or it might be the number of turns in the conversation. So longer conversations usually end up with a customer wanting to talk. You know, we're not answering solving the problem.

[00:13:45] - [Speaker 1]
We should probably hand over. And also, you know, repeated, you know, I wanna talk to a human. I wanna talk to a human. We'll we'll hand you over. So we have LLM inference basically deciding when to do that, and then it hands over to the contact center with a full context of that conversation.

[00:14:02] - [Speaker 1]
So the human then can very, very quickly answer the question. We've seen it super successful. We measure ourselves on the ability to what we call channel shifts, so not hand out our phone number because, you know, contact centers get very busy, so we want to try and solve problems in the digital channels. And we've seen we are seeing a shift. We're seeing the green shoots of a shift from people phoning to people using concierge to solve questions.

[00:14:28] - [Speaker 1]
As a technologist, what's super interesting is now I have the data on the handovers. So I now know the answers and the questions that the humans are in the contact center applying to the customer. And actually looking at that data, we are now have like a feedback loop to improve Concierge so we can sort of get it answering more and more questions over time. It's been incredibly valuable, really successful.

[00:14:50] - [Speaker 0]
And you used the word there context, and context is everything in those kind of conversations. How accurate and how pleased have you been with the the context and understanding of those queries?

[00:14:59] - [Speaker 1]
Yeah. So so context and understanding, you know, everything about the customer you can is really important. The first versions of Concierge, we deliberately made a decision not to bring in customer aware or booking aware data. We were very we that was a conscious decision from a data privacy perspective not to do that initially. I think now that we've proven the value in the system of of concierge and we are seeing usage, like good usage within it, within our mobile app and our website, we are now the next phases of concierge will be extending context to include customer and booking information, opt in of course, we're not just gonna do delivery.

[00:15:42] - [Speaker 1]
The the benefit to the customer is like for you now, I I know that you're a frequent flyer gold member and therefore I could pass concierge will personalize its responses based on that for you. So I think again that will be another step change in in in concierge usage for Virgin Atlantic.

[00:15:57] - [Speaker 0]
I think it's it's very interesting as well how so many different people use it in different ways, and it's not just solving problem solving pain points, it's almost also delivering value. You mentioned there about the trip planner. Tell me about that feature. And if I'm going on the Virgin concierge now and I I open it up, how would it help me plan a trip, for example, just to bring it to life for people listening?

[00:16:19] - [Speaker 1]
Well, yeah. It's I mean, it's imagine just going into a travel agent and asking the question, I wanna go on holiday. Where would you like to go? What you're interested in? You know, I I do you wanna go somewhere sunny?

[00:16:29] - [Speaker 1]
It will ask you. Do you wanna go somewhere sunny? Somewhere warm? Or do want to do you want a a city break or something like that? And and it will work with you on understanding where you would like to go.

[00:16:38] - [Speaker 1]
It'll then try and understand your how many people are going, when you would like to go, and start refining it down. I mean I mean, you should really try it. It is it is just

[00:16:48] - [Speaker 0]
like being working with a travel agent determining those questions. Wow. I'll definitely check that out. And I think we're we're talking more and more less about the technology and even less about the tools and more about return on investment and business outcomes, customer value, etcetera. What what are the the things there that that you've noticed that you've picked up on that make you happy about return on investment providing value, etcetera?

[00:17:11] - [Speaker 0]
Because I think that this is there's a real appetite for this now, isn't it? We've we've gone past the shiny tech syndrome and it's about business outcome.

[00:17:17] - [Speaker 1]
Well, I think I think actually we've been really quite lucky at Virgin Antic. We are a challenger brand and the nature of our company is to push push forward and try something new and try something different. And I think actually, you know, we went quite early on our journey here. We partnered with OpenAI and Adobe to deliver this. And I think in doing so, we were given free rein, you know, with a small team to build out a proof of concept knowing it might might lead to something, might not test and learn if you like.

[00:17:50] - [Speaker 1]
Yeah. Actually, you know, based on the timelines I gave you, the actual the actual, you know, investment to produce this was was not a lot. The the hardest bits were talent, you know, getting the right talent, finding the right partners to support you, but the actual system build and technology itself has been, you know, from a software engineering point of view, relatively straightforward.

[00:18:09] - [Speaker 0]
And I love busting myths and misconceptions on this podcast, and when many people listening think of Adobe, they'll think of the creative suite and so many different things there, but not necessarily what we're talking about now. So tell me about the the role that Adobe's playing in in what you've achieved here.

[00:18:23] - [Speaker 1]
Yeah. I mean, as I say, Adobe supports the high the whole customer experience front end for us, you know, everything from hosting and running our website, the content management systems that sit behind it, support us with all our analytics, our decisioning, all our AB testing. I mean, it is the whole suite that we have from from Adobe supporting the whole customer journey. That is the whole thing we got from them.

[00:18:43] - [Speaker 0]
And many business leaders listening to our conversation, they'll be struggling with the same challenge, getting teams aligned around AI rather than letting it sit in one department. What advice would you give to somebody listening that wanna create that same rhythm of change across their own organizations? You probably learned a lot of lessons. We've made your journey sound very easy. There was probably a lot of things you learned along the way, things you wouldn't do differently, but what advice would you pass down?

[00:19:04] - [Speaker 1]
Firmly strong leadership, you need c level buying. Okay. It only works coming down from the top. We were lucky, I mean, with our previous CEO, Shy, who who saw the value in this and gave us the permission to go after it. So that c level buying was key.

[00:19:18] - [Speaker 1]
And then my bit of advice after that would was we set up a very specific AI software engineering team just dedicated on solving this problem, and they had full autonomy. And that was a cross functional team of some AI software engineers, some front end developers, and some product designers. We pulled that team together and just gave them problem to go and solve. We didn't give them any of the any of the corporate overhead and governance overhead that we'd have to do elsewhere. We just said, right, go ahead.

[00:19:47] - [Speaker 1]
Here's your tools. Let's get some sprints up and running and deliver the backlog. And I think, yeah, that that level of autonomy really helped that team push forward at pace.

[00:19:55] - [Speaker 0]
And later today, you're doing a fireside chat here at the summit as well. I tried to book myself in, but it's sold out. So for everybody listening that can't attend that, tell me more about that.

[00:20:03] - [Speaker 1]
Well, there is a I think it's online. You can watch it online if you you want to. No. So so at that session, we're going to be talking about an Adobe report on consumer trends mainly focused on AI and AI adoption with with within brands. We we're gonna be talking about there's a couple of elements of AI like how the the traffic funnel is changing for how we attract and retain customers on websites.

[00:20:33] - [Speaker 1]
You know, we're seeing more people use AI such as ChatGPT or or Claude to do research and find information to access your website. So whereas traditionally you might have gone straight to a website and maybe end up on the homepage, we're now seeing people come much further down the funnel is what we call that flow. So that's that's one element. So there's a big question about, okay, well, if that's the case, how do we show up in those systems? How do we optimize our marketing spend?

[00:21:02] - [Speaker 1]
How do we optimize our websites and and assets to appear in in in these models. So so that's one element. How how do we optimize there? And then the other element we're talking about is is, yeah, the the rise of agentic experiences, this idea of agent to agent interactions and what that means. So for example, in the future, you know, Neil, you may have your own AI agent on your phone that you send off to go and complete a task for you.

[00:21:28] - [Speaker 1]
How how does that your agent interact with Virgin Atlantic, for example? How would it understand the context, and how would it understand your needs and requirements and book success on Virgin Atlantic to to for your journey or trip?

[00:21:41] - [Speaker 0]
And finally, how long do you think we are from everything you're talking about here becoming mainstream, and what do you think will separate companies that genuinely operationalize AI and agentic experiences from those that are still stuck in experimental mode, playing it safe? Is it coming much quicker than a lot of people really?

[00:21:58] - [Speaker 1]
It's here now. Yeah. It's here now. It's genuinely here now. And I have the stuff we're doing with Adobe on edge delivery services and experimenting there, utilizing their skills and tools with our agents to produce content, you know, we we've we've been able to fully have agents, you know, do research, do translations, build content, and publish out onto the Adobe systems.

[00:22:23] - [Speaker 1]
It's there now. And the way Adobe are exposing their systems works perfectly with the the way our concierge platform works. It's been a real eye opener and I think it's accelerating incredibly quickly. I think enterprises have a choice. You either wait till everything's ready in the market the market's there or you you go early and get that first move for advantage.

[00:22:44] - [Speaker 1]
For Virgin Atlantic, we went early and we're really pleased we did.

[00:22:48] - [Speaker 0]
I think that's a powerful moment to end. I'll include links to your LinkedIn, etcetera, so people can connect with you. But for anyone missing, wanna maybe play with some of the things that we've talked about today. Where where should they go? Where would you like to point them?

[00:22:59] - [Speaker 1]
Yeah. Virginatlantic.com. You can go there, and if you go into the help section, you'll be able to see Concierge in the bottom right hand corner, or download our new app, which we launched the just now in March, and that also has Concierge built in, and we will be extending that integration into the app for the rest of the year, building a native experience. Just to to add to that, it's not just text, it does voice as well. So we spent a lot of time on voice, and you can you can talk to it now as well, which is quite cool.

[00:23:24] - [Speaker 0]
Incredibly cool. And I appreciate just how busy you are today. I've seen you racing around sites, so thank you for taking the time to sit with me today.

[00:23:31] - [Speaker 1]
Great. Thanks very much, Neil.

[00:23:32] - [Speaker 0]
I think if there's one takeaway from this conversation today, it's that the future many organizations are still debating is already being built. What Neil and his team at Virgin Atlantic have demonstrated today is that AI does not need to be a multiyear science project. Instead, with the right focus, and the willingness to start small and learn fast, these are the things that make it possible to move from idea to live deployment in just a matter of months. But if I zoom out, I think there's an even deeper message here. The real advantage is not just adopting AI.

[00:24:11] - [Speaker 0]
It's how you operationalize it. And that means aligning leadership, empowering teams, building feedback loops, and constantly refining how systems learn and improve. And it also means understanding that customer experience is still the North Star, but technology simply becomes the enabler. And perhaps the most striking point here is, according to Neil, this is not something that is coming in the next few years. His big wake up call at the end of our conversation today is it's already here.

[00:24:44] - [Speaker 0]
And the organizations that move now are the ones shaping how these experiences evolve, the ones that wait while they might find themselves reacting to a market that has already moved on. So the question I'll leave you with is simple. Are you still experimenting with AI and agentic AI, or are you building something that is already delivering value today? Love to hear your thoughts as always, please. Drop by techtalksnetwork.com.

[00:25:13] - [Speaker 0]
You can leave me an audio message and browse through 4,000 interviews over there, but I'd love to hear from you. But right now, it's time for me to hit the chauffeur again and find myself some more guests. So I will speak with you all again bright and early tomorrow. Thanks for listening, everyone. Bye for now.