Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | Android | TuneIn | RSS
At Booking.com, they’ve been helping travelers with their trip planning for many years.
The arrival of generative AI has given them new ways to help travelers with this business-critical task.
Over the past year, Chris Cameron has applied his UX writing and content strategy skills in ways both familiar and new to help build a new AI-powered Trip Planner tool that integrates with Booking.com’s travel-booking app.
We talked about:
- his work as a principal UX writer at Booking.com on their “writing system,” which is sort of like their version of a design system for UX writers
- his recruitment to a “tiger team” at Booking to develop a new travel-planning AI chatbot for their travel-booking app
- the key differences between his prior product work and his work on this AI product
- the new kinds of collaboration that have arisen in his work on a generative AI product, in particular his work with machine-learning engineers
- the transition from the prototype of the app to its current position as an established product
- the product-feedback mechanisms that are built into the Booking “Trip Planner”
- how to jump start your learning if you’re new to working on generative-AI tools
- how they were able to leverage components in their current design system to build the new Trip Planner app
- the prompt engineering skills he developed by creating an AI “story robot” for his three-year-old son
- his optimism about the employment prospects for UX writers
- how traditional content strategy practices like establishing voice and tone and consistent terminology manifest in AI product design
- how new AI practices are just as likely to show up as enterprise productivity improvements as in customer-facing products and features
Chris’s bio
Chris Cameron has over 13 years of professional writing experience across journalism, marketing, and UX. As a Principal UX Writer at Booking.com, Chris oversees UX Writing Systems, managing the tools and workflows that enable over 80 UX writers to efficiently create high-quality content localised into over 45 languages and dialects. Born in Boston and raised in Phoenix, Chris now lives in Amsterdam with his wife and son.
Connect with Chris online
Video
Here’s the video version of our conversation:
Podcast intro transcript
This is the Content and AI podcast, episode number 11. As generative-AI tools are introduced into consumer products and enterprise workflows, the core work of content designers and UX writers still feels familiar, but the context for the work and many of its details are evolving. Over the past year, at Booking.com, where he has been working on an AI-powered travel-planning app, Chris Cameron has seen first-hand how the traditional concerns of content strategy and UX writing manifest in the world of generative AI.
Interview transcript
Larry:
Hi, everyone. Welcome to Episode #11 of the Content + AI Podcast. I’m really happy today to welcome to the show Chris Cameron. Chris is a principal UX writer at Booking.com, the big travel booking agency based in Amsterdam. Welcome to the show, Chris. Tell the folks a little bit more about what you do there at Booking.
Chris:
Well, thanks, Larry, for having me. Yeah, I’ll give a bit of my background as well. Like yourself, I started in journalism and then got into copywriting. And after moving to Amsterdam from the US at a very young age, 25, I guess, I eventually joined Booking in 2016, a little over seven years ago. And back then, the role was actually called copywriting. There was about 25 of us. And over the years we sort of discovered that we were actually UX writers, and we’ve become now this community of over 80 UX writers.
And now, I am a principal UX writer, and the area I look after we call writing systems. And what that is is sort of like the writing version of design systems, but it’s not so much a system, it’s more like the tools and the workflows that we use to get our jobs done. So my role is to work on those tools and work on those workflows and make sure it’s easy for our writers to get their jobs done in an efficient and easy way so they can create high quality content. And more recently, one of the areas I’ve been interested in looking into is GenAI and how we might use that to improve our workflows.
Larry:
Yeah, that’s why I wanted to have you on the show. You told me about this product you developed, the Trip Planner, that’s based on AI. Can you tell us a little bit about how that project arose and how you got involved with it?
Chris:
Yeah, definitely. So my involvement with AI and GenAI in general started when ChatGPT came out. I think a lot of people took notice back then. That was late last year, 2022. And I started playing around with it. I’m always a bit of a nerd and early adopter of technology, so I started using it for different things. I have a toddler at home, so I was actually using it to create bedtime stories for him. I would say, “Let’s ask the story robot what kind of story you want to read tonight”, and he would just generate a story idea, and ChatGPT would help us with the rest. It was a lot of fun.
Chris:
But professionally, I started thinking, “Okay, how could this be useful for our writers at Booking or how Booking as a company could use it?” And early this year, 2023, the company was seriously looking at GenAI and thinking, “Okay, what are we going to do with this?” And because I was already exploring it, I got pulled into some early discussions, and I thought, “Okay, we’re going to have some brainstorms, some chats about how GenAI did,” but actually the company was already like, “Let’s go build a GenAI chatbot and put it in the app, and this is going to be the only thing you focus on for the next couple of months.” And I’m like, “Okay, let’s do it. Let’s roll.”
Chris:
And so basically, a task force was formed within the company, sometimes called a tiger team, we called it sometime, and it was representatives, multiple people from writing, design, research, product, and then also machine learning, our iOS and Android engineers, of course, data science, and marketing and legal. It was a big team. In the end, it was almost like having a little startup within the company, it was about 70 people. And the UX work stream was sort of one half of it, and the other half was all the machine learning and the engineering that was going on.
Chris:
And this sort of kicked off in mid-April when we started this, and two months later, we were able to launch the AI Trip Planner in June. Just so people understand what it is we built, we built basically an AI chatbot into the Booking app, and people can chat with it and ask their travel questions, and it can help them get inspiration for where to go or what hotel to stay at or build an itinerary, these sorts of things. And it integrates some of the traditional booking experience, like with carousels and images and property ratings and things like that, right into the chat so it feels a bit more natural. And then if they tap on a property, they can go straight into the booking process and make a reservation.
Chris:
And so a lot of it uses some of our existing machine learning knowledge we’ve built up at the company over the years and then relies a bit on OpenAI ChatGPT to do that generative AI piece and really create a nice conversation. So if people want to try it out, if they’re in the US or they can VPN to the US, they can sign into the Booking app on iOS and Android and make sure their language is set to English, and they should see the AI Trip Planner right on the home screen.
Larry:
That sort of gets at some of the complexity around this, because I know you localize into 50 languages and cultures.
Chris:
Yeah, I think 45. Yeah.
Larry:
And so right now it’s just English only and in the US, so that’s interesting. And really, as you talk about that, I’m wondering from a user perspective, it’s almost like just a UI thing. For an end user, you could almost perceive it that way. “Oh, another way to interact with this thing and do my trip planning.” But on the backend, like you said, it’s 70 people on this tiger team that put that together. How similar was it to other products, because as a principal you’ve worked on a lot of different projects probably at this scale, how much of it was familiar and similar and how much of it was new? Tell me a little bit about that.
Chris:
Yeah, definitely. There was a lot that was familiar to just a normal building a product, but there were some key differences. For example, for working with a GenAI product specifically, it’s such a new thing that there’s not a lot of existing research. So if you’re going to go to your researcher and say, “Okay, what do we know about GenAI?” It’s like, well, they’re still learning too. So a lot of that was involved, looking at what is out there in the market, what competitors are doing, but then we were also able to combine that with the existing understanding of user needs, because essentially this is a search experience that we’ve been dealing with for a long time at Booking. So we know a lot about what the user’s looking for in that moment when they come to the app. So those needs didn’t change, but the way they were expressing those needs is the whole new thing.
Chris:
And in the early stages, when we were trying to test something, it’s not that easy to build a GenAI prototype. If you’re building a prototype in Figma, you can’t really insert the AI part in there very easily. Maybe soon that will be a thing. So we had to wait until we actually had a working build of the tool where we could play with it internally, and that’s when we started actually doing a lot of the understanding of, “Okay, what’s working, what’s not?” that sort of thing. So there was that challenge.
Chris:
But from a content and writing perspective, the biggest difference when working on a GenAI product was that you kind of have to let go of control over the words and the content that’s coming out of this thing, because when AI is generating the text and as a writer you’re seeing, “Oh, this isn’t great. We need to improve something,” then the only way to do that is to actually go to the source and dive into the code and the prompts and start making improvements right there.
Chris:
And that was a new thing for us, because we’re used to working in Figma with designers and then handing off a design to the engineering team and say, “Okay, go build it with the words I wrote,” and instead, it’s now, “I need to go talk to the engineer and get into their code or the prompts,” as it were, because they’re just English in the end, “and sit down with them and make improvements.” And that’s where we actually made some of our biggest gains was sitting next to an engineer or virtually via Zoom and saying, “Actually, let’s rephrase that prompt this way. Well, if we insert this instruction, it’ll talk more naturally,” that sort of thing.
Chris:
So we have always had a bit of a technical piece to our role here at Booking, because we do a lot of that content management of all the translations that we manage in all those languages. So we’ve traditionally worked very closely with engineers to make sure that everything’s working in every language on our traditional UI copy stuff, but this was definitely something new to sit side by side with an engineer and get into their code. And that was actually a really important piece, because that meant the writing role was really critical to this phase of development. Just by applying the writing craft into the prompts, we were able to make these really big improvements.
Larry:
It sounds like some kind of extreme co-writing. You’re doing it, because you still have the same concerns about governance of voice and tone and style and accuracy of the information and all that, but you’re going way deeper in the tech stack to implement the ways of getting that kind of content that you ultimately need. Well, you mentioned when you started that you’re on the writing systems team, which is sort of parallel to the design systems team. And so I’m assuming it’s almost like do you have an AI Ops now or something like that that content design and UX writing are a part of? Or tell me, I guess, organizationally how is this kind settling in?
Chris:
Yeah. No, there’s not really an AI operations team, for example, but now that the MVP product that we built in those two months in the sprint is launched, it’s like a traditional product that’s like any other at Booking. Now, there’s a more regular sized team that handles this product. So there’s a product manager and there’s a designer that I work with, and so there’s a couple of researchers that are maybe split between this team and another, and a handful of engineers. Mostly on the product team, there’s the app engineers, like iOS and Android.
Chris:
And then machine learning has always been a thing at Booking, so we have a separate department of teams for machine learning, and a lot of them are actually based in Tel Aviv, and we work with them virtually. So there’s the machine learning team that works on the models, and then there’s the front-end team that’s making sure the experience is good, and we have to work virtually with each other to make sure those two teams are coming together in the end, but there’s not an AI operations quite yet.
Larry:
Got it, yeah. And one thing, as you mentioned that too, another one of your new collaborators you mentioned earlier on is data folks, like data scientists, data engineers, folks like that. Again, it is kind of about how deep in the stack are you going? I know there’s a lot of concerns in that world around data pipelines, provenance of the data and its cleanliness and tidiness and suitability for this stuff. Have you got that deep into this?
Chris:
Me personally? No. It’s mostly how the researchers are interfacing with data scientists on how to pull learnings from the feedback we’re getting because we have a feedback system built into the Trip Planner, so they can analyze that feedback or they can analyze the different ways people are using the tool, and how many messages are in a conversation or what point they’re leaving the conversation and how we can reengage them, that sort of thing. So I haven’t had the need to sit down with a data scientist yet, but we’re using a lot of their findings, and they’re reporting regularly, and we’re using a lot of that information to guide us and inform some of the decisions and directions we take.
Larry:
Yeah. I was thinking just of the, I think, content pipeline looks differently these days, and I’m just trying to follow it as far upstream as I can. But I’m just thinking about the day-to-day work of this, I think one of the things, I feel like you’re a little bit in the future for a lot of the folks listening, I think, that you’ve actually built a product and shipped it and it’s live and it’s out there. I’m just wondering, can you help other folks accelerate their learning curve? What would you really advise to somebody embarking on this now? Is it important to understand the tech stack or the language around this or the new roles that you’re dealing with?
Chris:
So when it comes to learning a new technology to work with AI, like I said, I was already playing around with AI, so I kind of knew the lay of the land a bit, and I knew some of the lingo, so I didn’t have to pick up too much, but at the moment, AI is still pretty new to everybody. So everybody’s sort of picking up what the lingo is and how to talk about this stuff and what it means and how to stay up with it, so it’s not too hard to start reading or pick up on it. But that will change with time, obviously, as more people do this. But I think anyone that’s going to work on an AI product should have familiarity with some of the GenAI tools out there, like ChatGPT or even the visual stuff like DALL-E and know what the strengths and weaknesses are. But those are also evolving. Every time there’s a new release, they’ve gotten rid of all their most popular weaknesses, and now they have all these new strengths.
Chris:
But one of the things I think for me that I learned or that I thought about for our writing community and writers in general is that if you’re going to be working on an AI product like this, and you’re going to be working more closely with engineers on prompts, which I think is super important, being familiar with the tools that your engineers are using or already knowing how to get around something like GitLab so that you can submit a merge request to your engineer if you see an improvement to the prompts, knowing that is going to be important. But I can imagine that not all writers and content designers have that sort of experience. So it definitely helps to speak the language of your engineer and be able to navigate their world a bit so that you can talk to them and have influence on the prompts, because if we’re not getting in there as writers into the prompts, then they’re not going to provide the best output.
Larry:
And one thing, it’s been observed a lot that the prompt engineering looks a lot like content design, that you’re designing content to elicit an appropriate response from these AI agents. How does the prompt engineering talent feel at Booking? Is it kind of equally divided between you and the ML engineers, or tell me a little bit about that, how the prompts get written, I guess?
Chris:
Yeah, I think it’s during the MVP phase when we were building it, it was sort of like the UX stream and the tech team, the ML team were sort of running side by side, but not really aligning aright. And then at the end, once we built something, we said, “Okay, let’s bring this design together with all this engineering you built and see how it works.” And that’s where we had to do a lot of the work was making the things work together. So they were building a lot of the infrastructure of the prompts, and there’s a lot of “engineering” that goes into these prompts, because you have to connect all these different pieces and different ML models and different backend systems. And that’s all the stuff that, as a writer, we don’t need to think about.
Chris:
But there’s a lot of the prompts that involve just setting a tone of voice and giving it examples of conversations that you want it to learn from. And if you’re relying on the engineer to write that information, then they might not write it in the way that you want to as a writer. And so if you go in there and you give it better instructions on how to speak or how to have a conversation with your user, you’re going to have a better output.
Chris:
And some of the instructions, it’s about giving a complex instruction to the bot so they can do something with your instructions. And developers, they always want to talk about writing clean code that’s efficient, but maybe when it comes to writing a prompt in English, they’re not the best at explaining a complex topic in a really simple amount of language. And that’s something UX writers really excel at. So if we get in there, we can actually help them to improve the outputs of some of just their technical prompts, because we can maybe shorten some of their sentences. And with all this AI stuff, it’s all about not using too many tokens so that the AI can read all of your instructions and not forget anything. So the shorter and simpler you can make those instructions while still having the same impact is really important. So that’s one of the ways that the writers can get in there and improve the code, as it were, and really help the product get better and produce better outputs.
Larry:
I love that. Yet another argument for plain language and simplified and terse and concise, because you’re paying for those tokens and you better tighten it up. Hey, one thing is I’m kind of a content management nerd, and I’ve done a few of these kinds of projects and am already hitting the challenges of how to manage these prompts. They’re built into custom GPTs that you build yourself or prompt chains that you build in different ways. And you’re a systems guy, so I am going to guess that you have as good a chance of figuring this out as anyone, do you have ways of managing and also maybe learning from developing prompt styles or prompting best practices and things like that?
Chris:
Well, we’ve been sharing our learnings internally, just sort of like I said when we were talking beforehand, that I spoke at a internal UX conference, and I gave a bit of feedback about the things we learned from building this product. So we’re sort of sharing those prompt learnings internally, but we don’t have anything set up in place to create guidelines just yet. We’re still at the early stages of figuring out how AI is going to impact our workflows and things like that, because this is one product that is a generative AI customer-facing thing, but there’s going to be a lot of AI impact in workflow that’s just productivity and things like that.
Chris:
One of the reasons we were able to so quickly roll out this Trip Planner was because we have a really flexible design system in place, and I wouldn’t say that we had to build any new components for that. We didn’t have to build an AI component for our design system. Because done chat interfaces before, we’ve done even chatbots before, and so we could pull in existing pieces of our design system, and that was one of the reasons why it was so easy to or quickly able to design something was because of that design system.
Chris:
But I can imagine in the future that as we reimagine the way a user interacts with a generative AI chatbot, we might want to rethink what some of these components look like or the way that they behave, especially when you see every day there’s new advancements coming out, whether it’s from Google or OpenAI, and just how the AI can respond to you and thinking of new interfaces and new ways. And so I can imagine that these components will need to evolve soon, but we haven’t had to make those changes just yet.
Larry:
You’re reminding me that the profession of conversational design has been around for a while. I know 30 years ago they were doing IVR stuff for phone systems, and there’s a well-established discipline there, and I can see that showing up in your chat agents that you already had in Booking. So that kind of goes back to what I was thinking earlier about the UI to the end user, it doesn’t look or feel that much different, but boy, the stuff under the hood has changed a lot. That’s interesting, but so far so good. The design system is holding up and giving you what you need to express this stuff.
Chris:
Hey, I want to circle back, Chris, because I loved what you said earlier, and I don’t want to let this hang, because I think it might be also a way for folks who are not reticent, but just kind of trying to find a way into this, your “story robot” with your toddler. I just love that idea of, “Hey, let’s ask this story robot what kind of story we’re going to tell tonight.” Tell me a little bit about what you learned in that, and sort of you’re prototyping and testing with your kid.
Chris:
Yeah, yeah, it was pretty simple. At bedtime, he’d gone through the same books we have at home several times, and he would like to create his own stories just based on maybe what he did that day. It was sort of like, “Tell me what Spider-Man did today.” It’s like, “Oh, he went to the zoo, and he saw an alligator,” things that maybe Spider-Man never does in a traditional Spider-Man story. But then if you tell ChatGPT, “Hey, I’m writing a bedtime story. Make it about Spider-Man going to the zoo,” it might write a story about Spider-Man going to the zoo. But I found out pretty quickly that you had to give it better instructions for it to be maybe age appropriate or maybe to be more of a bedtime story rather than just a story. So the more instructions I started inserting into this thing, and then I would just add the details that my son wanted at the end, it got a lot better.
Chris:
So at some point it was like, “I have a three-year-old, and it’s bedtime. I need a short five-minute story, and here’s what needs to be in it.” It’s like what I learned with the Trip Planners, the better instructions you give it, the better output you’re going to get. And so in the end, now I’m thinking I need to go back to that and see how I can integrate the illustrations, because I think that was the one thing my son was always bummed out. He would watch the text appear on my phone by ChatGPT, word by word, and, “Wow, it’s so fast.” And then there’d be no images. You’d be like, “Where are the pictures?” And so I think that’s one thing I’ll be interested to do now that ChatGPT has DALL-E built into it is I can actually ask it to create some illustrations, and that’ll probably really blow his mind.
Larry:
No kidding, because the multimodal stuff is definitely coming in here. This is so weird. We’re recording this on December 8th. I’ll probably release it in two or three weeks, and half the stuff we say will be out of date by then, probably the way things are moving here. But I think it was just a day or two ago, Google released the Gemini stuff, and I think that had some multimodal stuff in it, but I know that the new ChatGPT does as well.
Larry:
Hey, another thing I wanted to ask you about, there’s a lot of concern about job security among content folks these days. How are you feeling about that, the prospects of the UX writing world in this new world? Do you feel like the robots are actually coming for our jobs, or is there still going to be stuff for us to do?
Chris:
Yeah. I know this is, I think, your favorite question to ask on the podcast is GenAI coming for our jobs? I’m not too worried about it coming for the job of UX writers, at least not yet. With GenAI, I guess we can never say never. A few years from now, who knows what it’ll be able to do. But I sort of see this in two different parts. First, there’s the productivity piece that I mentioned. Most people are probably more likely to experience AI these days through enhancements to their productivity or through their company trying to improve productivity. So in that sense, AI is having really big impacts by reducing redundant busy work, and that frees up more time for us to be more strategic or think more about content strategy, that sort of stuff, and instead of spending all that time on completing tasks.
Chris:
And so someone at our company said that this way, and I thought it was really nice, is that the first thing that the AI machines are going to replace are the old bad machines, and a lot of our systems and tools are going to get smarter and more efficient. And so for writers, AI is just another tool in our toolbox, and we should be getting familiar with it. And I think I heard someone else say this on your podcast, and I was going to say the same thing, is that UX writers won’t necessarily be replaced by AI. They’ll be replaced by other UX writers that know how to use AI. So you should be getting familiar with this stuff so that you can continue to grow and be relevant in the business.
Chris:
The second part is if you’re actually working on an AI product, a strong UX writer is really critical to the success of those projects. As I experienced with the Trip Planner at Booking, using my UX writing skills in the prompts, in the code was just as impactful in the end as using our skills for UI copy that we traditionally work on. So the AI models, they sometimes, like I said, they have a complex set of instructions, and we’re really good at expressing complex things in a short way.
Chris:
So I think writers will be indispensable to AI, because the code is just writing, and it’s kind of the developers have to learn a new language. Maybe if English isn’t their first language, it’s easier for us as writers to improve the code in that sense. If we’re not in there influencing the way that the AI is generating the content, then who else in the product team is going to be standing up for brand, tone of voice, consistency of terminology, inclusive language, things like this. It’s all the stuff that we worry about all the time, but now it’s just a new way to have that influence.
Larry:
Yeah, we’re still human-centered designers. And as you were talking, I was just thinking about that even as these agents appear more and more human, they’re still machines, and it’s still HCI, it’s still human computer interaction, and we’re still involved in that. It’s probably going to be more important as we go along to just remind ourselves of that, and this should all be in service to humanity. I personally don’t want to live in a world of robot overlords.
Larry:
But hey, Chris, I can’t believe we’re coming up close to time. Is there anything last, anything that’s come up in the conversation that you’d like to revisit or that just want to make sure we share before we wrap up?
Chris:
Yeah. I guess the main thrust I’m trying to promote at the moment is that thing we just talked about, which is that I don’t think writers should be worried about AI replacing them. I guess it’s easy for me to say when I’m at a company with 80+ UX writers. We’re probably not going to get rid of 80 people to replace them with AI. But there might be a startup out there that’s thinking, “Oh, maybe they were thinking they were going to hire a UX writer, and maybe now they’ll think about if AI can help them.” But I think in general, the skills out there that we have as writers can have a huge impact when products involve AI. Otherwise, we’re just going to have our productivity improved by using AI as a tool, as people have said on your podcast And I’ve repeated again. It’s a tool that we need to learn to use and it’ll improve everything in that sense.
Chris:
And the only other thing I’ll say is that I am trying to write a blog post about this very same thing. So hopefully by the time this episode comes out, it’ll be published and people can check out blog.booking.com, and it should be available there. Yeah.
Larry:
Yep. I was going to ask if you had documented this in a blog post, but I always feel guilty about that. As a former editor, I always feel like I’m making an assignment when I’m just asking the question, so I’m glad to hear that you’re already on that. Hey, one very last thing, Chris, if folks want to stay in touch, what’s the best way to connect with you online or follow you?
Chris:
Yeah. I’m not super online professionally, but I guess if people want to find me, LinkedIn is the best way. You just search for my name in Booking, and you’ll probably find me.
Larry:
Excellent, and I’ll put that in the show notes as well. Well, thank you so much, Chris. Really enjoyed the conversation.
Chris:
Yeah. Thanks a lot, Larry.
