Prateek Keshari & Mathieu Bastian: How we work with AI

I don't remember the last time I was this excited. For the last two years, I think it has been probably the most exciting time of our lives. It's fascinating because things are changing and upskilling and learning constantly. But I think we have had a decade worth of progress in the last two years materializing and it's pretty crazy. Hi, I'm Jacamo, an engineering manager and your host for today's episode of Behind the Journey. On this show, we explored the inner workings of Get Your Guide by talking to our incredible leaders about their ideas, strategies and the winding paths that led them here to our Berlin HQ. We also talk about our culture, values and how we strive towards our company mission every day. And for those of you who don't know what that mission is, we are working on reconnect humanity with the sense of wonder by putting millions of travelers in touch with unforgettable experience all over the world. So if you're looking to embark on a new career journey and the thought of reshaping this travel experience sector in innovative companies sounds exciting to you, head over to GetYourGuys.Coreers Time to dive in. Today, I'm excited to have Pratika Shari, Product Marketing Lead and Mathew Bastian, Principal AI Engineer, Ear on the Show. They'll be sharing insight on how we work with AI, their secret to success under Journey here at Get Your Guide and lots more. So, let's start with you Pratik. Thanks for coming on. Can you introduce yourself and tell us about your role? Before I start, actually, it's great to be sitting on the other side of the table. This was literally the thing that I was trying to host this podcast on the other side. So it's good to be starting with with a view that was eyes on the other side of the table. So I'm Pratik. I lead Product Marketing at Get Your Guide in plain English. What that means is I try to tell stories of the product that our product and engineering teams build the design teams build to us suppliers which are partners in a way that's clear, usable for them that helps them get more out of a platform. And that's the core job of my role. Great. Thank you very much and thank you for being here. What about you, Mathew? Hello everyone. I'm Mathew. I'm an AI Engineer here. And my role is a little bit about what we're going to talk, I think, a lot today, which is to help drive AI adoption through various initiatives, but also play a lot with our product and part of some product development initiatives, whether it's on the supplier side or on the traveler side. So I keep myself busy working with different teams solving LLM-based problems and bringing this technology into our product. Very exciting. Thanks both. Now I want to dive straight into AI. I know you're both heavily involved in leading the way in how we use it here. Mathew, how did the AI initiative start at Get Your Guide? I think it would be incomplete if I wouldn't mention that we've been doing machine learning for the last eight, nine years, I think. I've been the company for nine years now and kind of joined that journey at the beginning to use machine learning into our product, into our search and discovery and so on. Using data at scale is not new to us, but of course, I think what we really see is a complete new chapter with generative AI. And that started really, I think, with when CharGPT came out, with CharGPT 3.5, which, no, it feels like a long time ago. A lot has happened in between. I think, Partique and I, we started to work together, I think, just on the newsletter that was probably the spark. We were naturally inclined to monitor what was happening in the industry and felt that it would be helpful to others that we summarize it and bring this state of AI newsletter that, yeah, eventually be being read by almost everybody in the company and people enjoying it. So I think that's when we realized that we actually had created a new job for us, which was to maintain and ship that newsletter on a regular basis. So that was the very early beginning of bringing that kind of new technology into the company, first through knowledge sharing, and then eventually also through tools and so on. But I'm sure we'll talk about that as well. If I can do, Matthews, I have this bit of memory. This was, I think, one of our downhalls that was going on and I walk up the match you asking for an API key because GPT-4 had just been released via API. And I was very curious to figure out what we could do. I was just curious to play with it. And from that on, this whole idea of let's find ways to scale this. It generally gets more people excited. All of that came into picture. For me, what was most interesting is that how bottom up it felt. So there was no memory or sort of, I think, curiosity just led the way. I did employ a branding like I briefly mentioned on the other side. And it was never a question of staying in your lane kind of a thing. It was just curiosity driving all the efforts. And since then, we have been writing newsletters internally and now a lot more that the team does. Yeah, it's good that you bring this. Somebody has to manage the account, deliver API keys, right? So I was, I had this very short moment when I asked myself, should that be that person? And of course, I ended up doing that. And it was interesting because this is where this, this excitement really was visible. From the moment we said, hey, you can get a GPT account. Just send that link there or subscribe. There were hundreds of people in the company who were excited to try it out. And, and yeah, that was clear that from the start, we were not the only one who were curious. I think since then, we want to keep that spirit when there's something new. We do want to the next day. We want to give it a try. We want to get people to try it as well and see what it can do for them. That's fantastic. Thanks for sharing. Pretty, I think it's safe to say that you use AI every day. How impactful as AI being to get your guide. And do you think it will continue to make a difference to us in the future? I think AI as a practice is now deeply embedded into how we even think. So, for example, personally, I use AI every day. And one of my core objectives from using AI is using it to challenge my thinking, using it to refine my thought process. And not just, for example, right, let's say product launch campaign or whatnot, but actually getting that different perspective. And I'm just one of the many examples of people that get to guide who now have AI deeply embedded into how they how they use it to shaping their thinking into shaping what they end up achieving with it as well. And to second part of the question, whether this will continue, I definitely think this is, I think we just crashed the surface. There's so much more, there's so many developments happening across the industry, whether it's through whether it's in text, whether it's in video, audio, imagery, for example, all of that is super, super exciting. And I think we're just getting started on the journey. And we've come far on driving that understanding in the company. And I think we can do a lot more. And there is there's just a lot of a lot of curiosity and a lot of momentum in the in people all across that is also just driving that that energy across all the departments, which is very exciting to see. So I definitely believe that this is just a starting point for what to come. And to that end from what we end up doing, I think continue to deliver impact for our partners, we'll continue to deliver impact for our travelers. And there is this so much more opportunity across. Fantastic. That's great. Your enthusiasm and excitement is definitely contagious on the topic. For me, one of the most incredible things about AI is access to BDT. Or useful, it can be whether you come from tech word or not. But you as principal AI engineer, I'd love to hear your take on this. Can you give us some insight into how non technical individuals at Get Your Guide are encouraged to use AI and audit technology? Do you even have any stat about how much it's used already? And what are your goals to increase this? Yeah, sure. Happy to, that's kind of your long answer, so bear with me. I think the way I'm thinking about this accessibility and it's I totally agree. It is impactful in so many different parts of what we do. There are however some structure that we want to put into this. Okay, what is this AI adoption journey? If you share any data and metrics, what do they mean? What does it mean for us? Because there are different ways, different levels, so to speak, where someone can use these tools. And I think all goal is that and that has been from the beginning that folks who haven't tried it out, they for them is very easy to trial. I think that I don't think there are many people who haven't tried it out at this point, to be honest. But that's moving from zero to the first step. Then for those who have this entry level, what maybe that entry level is like replacing Google Serf. That's already interesting. Like you can get better results by prompting an alarm that to searching on Google. That's good, but it's you're not getting a ton of value by doing that quite yet compared to what you can do, especially on this getting feedback on your ideas and having a kind of a sidekick, what you were saying, using the lm to bounce back to think deeply to strategize that next level. Then for other techniques like deep research or there are other tools that can tell me that you've crossed that beginner to intermediate, maybe even to expert boundary. What we look at today at Get Your Guide is daily usage. And hypothesis behind this is that if it's a habit, if you're using it daily, and again, we don't look into what you use it for because that will also be technically difficult, right? We don't want to inspect your prompt. But basically, we assume, okay, you use it daily, so it's a habit for you, so you must be getting great value out of it. It doesn't answer this, how do you level up? But at least it gives us for sure a good understanding of our people taking advantage of these tools in a way that really is meaningful for them, because if they're not using it everyday, most likely, it hasn't been transformative. So daily, instead of looking at weekly or monthly, that was one of the things we said, we want to measure daily usage. How can we get to measuring that? And today at Get Your Guide, we have about 50% of the staff using those tools daily. So every single workday, we have half of the company who are at least doing one prompt or one chat session. And we track that across a number of tools. So it's pretty comprehensive data, and we have history for that data to see that back in September, last year, in 2024, we were around 30%. So there's constant improvement. However, we see differences between teams, right? There are folks who are early adopters, others that need more kind of inspiration. And our target is 100%, right? Pretty. I think you are obviously using it every single day, me as well. And I think we'll eventually get there, but right now, I think two-third is where we put the objective for this year. And I think it's really not, there's no silver bullet, right? So in my view, something that even if you were doing nothing, because we love what we're doing, right? So we love being talking about AI. We love trying out these new tools. We love working with folks to help them grow in their LLM and AI journey. But even if you were not here, I do think our culture of curiosity and also growth over comfort. These are some of our guiding principles. They will, they are making you experimental, they foster this experimental culture. So you are encouraged to try out the tools. So even if you were not here, I do think this will grow organically. We are only, I would say, accelerator providing some guidance and maybe making this journey a bit faster. But I'm confident that even if you were not here, this would work. Given the kind of culture we have and how well adopting a new technology really honestly fits our culture, this is something that we never look back and say, hey, we shouldn't have pushed on that. This is something that you know, we see no resistance across the board from the executive team to focus on individual contributor roles. They are there are no like oppositions to change and that's why I think the culture eventually take care of itself. There is obviously a differences between how AI is adopted between our departments. For example, you'll see naturally a much higher usage in engineering followed by let's say content and other teams. I think what's most interesting to me is how other teams are using AI for things they would have not thought about using before. I'll give you an example. Let's say people in sales building their own small tooling or building scripts to automate boring stuff. I think we're slowly getting into this phase where the fear of start is vanishing and this is like very permission less kind of a more dry and I think that's pretty exciting. Imagine someone in for example communication, so marketing in general previously you'd spend 30 days, 60 days IDA day and then building an MVP and whatnot. Now that time to go to production time to go to time to go to market has significantly reduced. It also is more exciting because you have full control and you can play on yourself or with your teams to build on stuff and I think that's pretty exciting. And I think that's the whole idea of that you take a lot of effort to get to something meaningful is slowly reducing and I think that's pretty exciting across data. This is something that we are seeing in different kind of teams whether it's like content or sales or marketing. So it's not just it's just not bucketed in it's not just a bucket for engineering. I think that's that to me is very it's a good signal that adoption is indeed but adoption is is also different in this case. It's indeed a fantastic signal. Coming from engineering for me is this tool, keep the momentum with your idea. All of a sudden realizing what you had in your mind becomes so fast and so approachable that it can really close the gap between your tools and the result you want to have at the end. And for me is also incredible what you mature saying. We look at the numbers of course but we look at them to see how this adoption is going. We're not forcing anybody to use AI. We just provide these tools and just through our culture of curiosity and trust let people use them and see how high this rocket can fly. And one of the thing to follow with what you just said I think one of the things that I keep asking Matthew and when we think of like workshops or when we think about education in general about AI is that what is that one wow moment or high moment that we can deliver when we deliver let's say a workshop. So we did a workshop when we did our employees summit week and that is exactly what we are aiming for. When they try out something I was in a workshop where we were using deep research with Gemini 2.5 Pro and on that particular day itself they had released a way for for a user to get the output from deep research and then convert that into a web page or infographic. Even I did not know about that. So it was also a wow moment for me that you can go from idea like literally a prompt to something that you can share with your team is much more accessible, much more visual and I think that's pretty fascinating. So I think a lot of the times when you're thinking about driving adoption within the get a guide ecosystem as well we also try to think about what is that one wow moment that we can deliver to a new user especially who are just starting out about fear of starting out. Really making it that easy for them. That's fantastic. I'm coming back to you, Prateek. You introduce a bit your journey and your career but you had really an interesting career to get your guides. You really shifted between different roles and sector. Can you tell us where you started and how you got where you are today? Yeah so I spent my formative years at least education wise learning all the intricacies of computer science. So I was an engineer at least as I studied but I was really bad at it and spent countless hours did not perform really well and naturally always had the inclination to writing stories instead of writing software. So it was pretty fascinating jump and I became a marketer. I joined get a guide. I started doing an employer brand which was helping individuals all across the world find get your guide and make a career decision to move and be with us and spend their time with us and I spent good two years in the role and absolutely love it and naturally then found an extension also because I was developing and I found an extension where skills around engineering. I wanted to blend that with product and that's how I found this tool and I moved across across a few floors now into a new team which is a double communications team. So that's how things ended up happening and was very natural as well. I saw that as a natural progression and when I started doing product marketing this is about one and a half years back. I actually found that it is the blend of what I always envisioned which is a blend of telling stories with an engineering mind for example and that was that has been a super super valuable. I can offer a perspective that I feel can be valuable to teams and that this has been this has been what I've been doing so far and that's that has been my journey. As I was saying an incredible career. It's also been like one a lot of I would imagine bumps. It's never straight forward journey. I'd never imagined that I did employ a brand that we would end up doing product marketing or Matthew and I will bump into each other and we'd think of something that we'd want to do with the AI and we still don't know sometimes that what we are doing but we figured it out. Yeah it's a side job for me it became the main job which is something that is very exciting right so my my responsibility is focused on this AI journey for you particularly you're doing it on the side and you're loving it. It's very impressive. Yeah but I think it also speaks volume of how your career can evolve through the corridor of our office here in Berlin and take a weird unexpected and fantastic shift. I think that's true for a lot of people I've seen a lot of it in the moments. It's always been cultured it's always been surprising for me that when we think as individuals about changing careers, a natural thing about it, can I find something within Get Your Guide? I think that's awesome. That as a culture, we're so good at this that there is an extension of you that can find its place within the company, sometimes like I said, within a few floors. And I think that's beautiful. I think that also gives people like a launch pad, not to scale what they're doing and not just absolutely start from a blank canvas. And I think that's great. When I think Get Your Guide does a very good job in enabling some of it in general. By the way, AI can help you a lot when you're learning your skills. There's something that somehow goes very well together. Actually, some of the stuff that we've been doing as well as we literally just think about it and find a way to learn it and then ship it. Yeah, I think it's from zero to one. It's never been easier to get to something that you can demo or show to others, which is amazing, even for people outside of tech. These days, my go-to hack is that I see a message from Matthew. I take a screenshot of it. I paste that screenshot in the get a repo in cursor and have it to wipe-- so basically wipe code the feature. I've done it now a bunch of times. And it's pretty fast because we can go-- So I write the prompt for you. So he writes the prompt for me. And I go from idea to coding in cursor and then actually shipping to production in about 10 minutes. It's for internal tooling, obviously. It's not for external stuff, but for internal stuff, it's pretty fast. It's just amazing. We have a good collaboration. Yeah. Hi, agree. Matthew, we're just speaking about non-tech role. And individuals with limited exposure to tech adapt very well to tech-driven roles sometimes. Matthew, how has the use of AI helped to create more seamless transition for those in tech departments without a tech background? Yeah, this is something that I really care deeply about. And this is also why we-- it's not just about building a newsletter or sharing content, but what's also the platform behind it. So is this set of tools? Are those LLMs accessible? Can you easily log into it? Do you need anybody's permission? What about your manager? There are so many small things that could create friction for maybe someone who is not familiar with this tool. So how can we eliminate those friction? How can we make it super clear that you can use these tools? This is how you get started. So there's a platform behind it with some handholding a little bit, which I think is really great to get people on the first step. And then they usually get to understand the value of it, and then they continue their journey by themselves. But there needs to be some self-service sources. So this only works at scale, right? So we can do workshops, I think, in particular. Of course, I mentioned that. We love to do workshops, but we can do workshops every one all the time. So in order to make these, I think, accessible, you need to think a bit about it in systems, right? So what's that onboarding journey to AI at Get Your Guide, right? So it's clear enough and how do you get to access to the next levels, how to learn these skills? And this is also one of our key insight, which is that there are some of these self-service sources that are pretty helpful, but again, they are great for really the early journey. What really works very well afterwards is the peer learning. So it's all about learning how your peers, somebody in your department, or in your group, or in your team, you learn from them on how they're using AI, right? So what's then the system so that these folks are encouraged to share and demonstrate? I know you have an engineering background, so you know what peer programming is, right? So I think this is a similar spirit. How could you encourage somebody to share their prompt and with others to demonstrate how they've been, how they've been using AI? So if you take these components together, I think you create the environment for anyone, I think, to pick this up. And then there needs to be also some decent data integration, right? So where is the data? Can I bring it into the LLAM and so on? I'm not going to get too technical, but there is-- the LLAMs are only useful in the company if you can somehow access the company data and knowledge. Otherwise, you quickly limit it if you can use your own documents and so on. What I want to also add is that we hear a lot of interesting use cases from those teams, and we are also learning a lot from them on what would make their journey easier. It's been very humbling to do some of these workshops, actually, for us to understand what is it that really is confusing when you're using AI and still a lot. So even though I'm very optimistic, I think there's still a lot of progress that I expect in the level of proficiencies that people have. I think that's also-- imagine a lot to do with the ways of how things are developing as well. I have, for example, changed my way of working with AI. And I probably imagine my way of working with AI every other month, for example. There are new models. There's new way of prompting them. There's new tools. There's new workflows without matter of fact. So I think it's a-- so at the center of all of this, I think is a constant desire to learn, right? I think that we-- to the beginning, it's more about curiosity. And I think when you have that at the center of what you wanted to-- what at the center of the-- basically your philosophy being that you want to learn, you will figure it out and you will find out, OK, this is not just new, but this could be beneficial. And you can evaluate it for your workflows, your own use cases. But I think it's a constant journey. It's never-- I don't imagine AI as a skill is something, at least, at this point in time. It's something that you can attain, get a certificate, and then call it a day. It's not going to be that case for-- at least for the foreseeable future, because it's still in a phase that there's a lot of delupments happening. Sometimes, even, for example, when we-- when Matthew and I start drafting a newsletter, it's just so much of noise as well that we have different model of these different APIs that are now being exposed and whatnot. And we have to filter all of that out to make it really snivel for the team's identity guide. But most importantly, I think there is just so much momentum that you have to always-- you have to constantly learn. One interesting angle that we discover is also how the leaders in the organization play an important role. I think you want to have the-- of course, your executive team. And that's the case. I'll get your guide be supportive. And you want the curiosity to be very facilitated. But then you also want the managers to be encouraging these types of exploration. And sometimes, it can feel disruptive, meaning you need to learn and learning if you're busy. That's next time. So I think there is-- yeah, it's not easy to get into this habit of learning, necessarily for everyone. Especially, again, I'm a pretty simple engineer, but honestly, I know nothing about LLMs. There's still so many things to learn. And I think that's true for every expert. So having the curiosity is great, actually implementing it so that you really do get on the new updates when you need to is another-- is something else. It's harder to achieve consistently. I think you're right. I've seen that within some of the LLMs can be actually push for an adoption as well. And it's great to see. I think it drives the momentum, gives people a space to experiment and find new workflows, find new ways of working. I think over time, like Matthew said, obviously, when you don't have something very set, that you're not playing the same cards, then it's obviously a little bit disruptive as well. I think if you have that understanding within your team, within departments to experiment, I think you can come out stronger. And I think there's a lot of momentum that you can build as a team when you do that. Especially if your team heads in leaders to that. I think that's beautiful. Yeah, absolutely. And having been in this industry for a while, I see around AI that level of excitement that maybe in the technology world, we were missing for a while. And it's fantastic to finally have it here. Yes, it's going to be disruptive and it is, but it's also incredible to be able to work with these tools and with the promises that they carry. And obviously, it's fantastic to be able to work with two like you that are on the very front line in this. And I've been adopting this technology from the very beginning and are guiding us through this. It is super exciting. I don't remember the last time. I was this excited for the last two years. I think it's been probably the most exciting time of our lives. It's fascinating because things are changing. And it's also like you're upskilling, you're learning constantly. But I think we have had a decade worth of progress in the last two years and it's pretty crazy. Definitely. Thank you both so much for your insight today. It's been a great chat and thank you all for listening, of course. You can find more behind the journey episodes on all major streaming platform. Catch us next time to learn more about how Get Your Guide is revolutionizing the travel experience industry. Don't forget to head to getyourguys.careers and check out our open roles to start your journey.