(56) SPOTLIGHT: AI Isn’t Just Faster Translation, It’s a $40B Tug-of-War for Global Attention - YouTube https://www.youtube.com/watch?v=mVA1koQv_2s

Transcript: (00:01) [Music] Welcome to Topline, the podcast for the best founders, operators, and investors in B2B tech. Every week, AJ Bruno, Asad Zaman, and me, Sam Jacobs, will break down what's important for you to know to be the most well-informed professionals in the market. Are you ready to level up your goto market strategy for 2026? Then don't miss GTM 2025, the only B2B tech conference exclusively for go to market executives. (00:39) Join a thousand other revenue leaders this September 23rd to 25th in Washington DC for an exclusive executive onlyly experience. This 3-day event is hyperfocused on connection, strategy, and execution. Expect hands-on workshops, in-depth strategy sessions, and curated opportunities to build relationships with VPs, CXOs, and founders facing the same challenges you are. (01:02) You'll stress test your GTM approach, align your team, and leave with real actionable insights from top performing leaders. Don't miss the must attend conference designed to help you boost go to market results in 2026. Visit attendgtm.com. That is a tt n dgtm.com to secure your spot today. And don't forget to use the code topline for 10% off your GA ticket. Hi everyone, welcome to Topline Spotlight. You know Topline Spotlight. (01:36) This is the episode where we sit down with a business leader and talk about a problem that they have solved in the last 12 months. an important a difficult uh problem one that we can all learn from um by understanding how this individual thought through the problem and solved it. I am Assad Zaman, one of the co-hosts. My other co-hosts are vacationing. (01:59) So, you just have me, which is why this is the first time I'm introing an episode, but I'm very excited about it because we have Brian Murphy joining us. This has been a conversation I've been looking forward to. Brian, welcome. Thanks. Great to be here. I'm very excited to have you. (02:18) So, Brian, we're going to talk a lot about the problem you solve, but before we do that, um, let's get the audience to learn a bit about you. Tell us a little bit about yourself as well as the organization that you lead. Sure. Thanks. So, uh, I've been building, uh, SAS and e-commerce companies for the past 20 years or so, um, as a founder, um, and publicly traded companies and, uh, within, uh, growth equity or private equity, uh, uh, investors. (02:50) Um, so I spent a bunch of time building um, WHI, which was a SAS e-commerce company. We built that to two and a half billion dollars in GMV and sold to eBay. So I spent a few years at eBay running a variety of businesses. My last role there was vice president of vertical markets. So I had responsibility for global product and strategy there. (03:07) And and now I'm I'm running um a company by the name of Spartling. And we're an AI translation company. We help companies like Apple, IBM, Disney, Tesla, uh Pepsi, Verizon create multilingual experiences using AI and software that their customers love. It's one of those spaces that is huge that not I think I I don't think everybody understands how big and how important this is. (03:33) So tell us what's the size of this industry of translation and why why is Apple needing help with translation? What does it do for them? Yeah, that's I was surprised too when we started looking at the space four years ago. Uh I had no idea that it was a $40 billion industry globally. It's like massive. (03:53) Just for context, by the way, you know how everybody is talking about code generation today as this massive market that AI companies are fighting over. So you've got cursor, you've got GitHub, you've got OpenAI and Enthropic fighting over it. That market is supposed to become 27 billion by 2027. So when you think about the most competitive market in tech is about half the size of this market. That's kind of crazy. (04:21) It's wild, isn't it? So the and the reason for it is fairly straightforward. Um you know when when you look at the data 87% of people globally will not buy from a website that's not in their language. Like think about it for a second. If I'm looking at a pair of shoes um on a German website I don't speak German uh I'm probably not going to buy. (04:42) Yes. Yes. So um it it our customers the uh B TOC B2B um they do it to drive conversion rate and GMV and this this works both ways right like if you are Apple and you're looking to sell in Germany it's going to be better if your website is in German but also you know I like buying clothes and if you find a nice brand in Germany you you can't buy their clothes unless you can read the website. (05:13) No, no, you just can't because it's like the sizes, the descriptions, all that material, all that stuff. So, we we do a lot of business with the big retailers um for that reason. And they're content changes all the time. It's like, you know, it's a product information management issue and so they have millions of SKs with tens of millions of product descriptions uh flowing under that under that that are changing all the time. A lot of fast fashion customers are big customers are. So, they need they need automation. (05:39) um they need uh very rapid turnaround on translation and um and high quality. So that's that's really what what we work to to do. And what's the history been of this industry in the sense that I imagine that the importance of this and also how this is fulfilled as a service has evolved a lot as the technology industry has evolved. (06:06) And so 20 years ago, was it just a bunch of humans reading things and sending, you know, a written up translation that somebody would manually do and then it slowly became more and more what h how did the industry evolve? Well, that's still being done a lot today. That's why it's, you know, it's pretty remarkable. (06:23) Uh we run into these massive enterprises and we're like, "Wow, I can't believe you're doing it that way." Okay. So, um it's it's really interesting and that's one of the reasons why it's 40 billion. I mean when when we I I've always looked for big um verticals that have the potential to be disrupted right so we did it in auto parts selling auto parts online um at eBay we built a $4 billion a year business selling h used cars online uh etc. (06:48) So, um, when we started looking at the space, we're like, "Wow, it's 40 billion." Uh, one of the reasons for that is that rack rate is like 20 cents a word and it's fairly slow and, um, and not automated, right? So, it's literally like extract the content, put it into a spreadsheet, email it to the translator, bring it back, revise it, back and forth it went. (07:11) And I've done this my entire career, and it's and it's always expensive, slow, and takes a lot of resources. So I try and a ration my content b uh do it as infrequently as possible which is really to the detriment you know of of of the company as we just described if 87% of people whether they're we're trying to convert leads for a B2B or we're trying to sell B TOC that's that's a problem right so when I when we started looking closely we're looking at tech in the translation space because we recognized it was a big space with potential for disruption we ran into a few companies and we really like Smart Link because we're Wow, these guys have um a great (07:44) SAS platform. They're leaders in machine learning. They're leaders in neural machine translation. So, at the time, we could do things faster and we could do them cheaper, but we couldn't do them better uh three and a half years ago. And that's really where AI came in and sort of it was like, aha, that's how we solved the problem. (08:08) Interesting. Uh when you said we were looking at uh companies, was that when you were in uh in a previous organization looking for options or were you looking at companies from an acquisition lens? Uh give us some color there. Yeah. So I' I've had a relation. So So Battery Ventures is is the uh is the owner the investor of of Smartling and I've had a relationship. Good friends with the battery folks. (08:31) Okay. They're they're the best. Uh we've uh I've had a relationship for them for with like 10 12 years. And from time to time we'd, you know, we kind of look at things and uh decide, you know, whether we want to do something or not. And they called me um about this space. (08:48) And the more I looked, the more I got excited about, in fact, I liked it so much, you know, my original thought was like, well, come on, I'll be a board member, yada yada. And then I'm like, no, I want to I want to run this business. This is this is huge. So you you find the organization and then you can do it cheaper, you can do it faster, but you can't do it better. (09:06) What was the delta there? like if if I if one was using smart lane pre-AII or pre this version of AI versus one of the traditional methods of what was the delta in quality that AI is now helping bridge that that's a great question so um we use a quality metric called uh MQM or multi-dimensional quality metric okay it's a it's a super complicated uh sort of qualitative quantitative thing but it it it puts out a score right ranging from zero to 100 So, um, at the time at the time we could do machine translation that would give you a score of like 86 to put things in context. And, um, let's I'll get yelled at this, but you know, (09:45) you can't say it's like 86% accurate, but let's just call it that just for the sake of argument. So, 86% good, right? Um, human translation is around 98. So, that kind of gives you an example. Okay. Uh, it's never perfect because people have preferences, right? I one translator might say it this way, I might say it that way, but it's 98. (10:05) So that was where we were. We were kind of like at this 86. And you could use that for lots of content that you could think about. User use it for like medicine, let's say. No, you wouldn't want to use it for medicine. You wouldn't want to use it for like a landing page or a homepage or um a marketing email, let's say. But there is lots and lots of content that you couldn't. We did use that for. (10:24) And we had, by the way, we also provided other forms of translation as well, human translation. and we still do. um uh and and it is in fairness I should say I'm going to talk about when we talk about AI there's two there's two forms of AI there's what we call AI human translation which is uh AI with a human in the loop right so this is about making our translators vastly more productive so we're just talking about code generation right no one like uses AI to write code and then release it to production right that just that happens very very rarely it's similar with with (10:57) AI human translation senses uh human translator pre AI could do like 2,000 words a day. Now we're with the productivity tools that we're g giving them through AI, it's like 8 10,000 words a day and climbing, right? So we're making our human translators way more productive. Um and then not to complicate things, but then there's this version of AI translation which is a click in quality below. (11:24) So if that's 98, AI translation is like 96 97. uh and that's purely automated with AI and that's onbrand uh that is matching style guide so grammat um terminology um tone of voice all of uh localized all these things uh at a fraction of the cost and turnaround time. So that's that's the big unlock and that business now is growing you know over 50%. (11:50) And is that the future of the business? Like do you think this is where because that seems like if it's 87 if it's 97% right like that that would be something that gets a little bit better as the models get better as well um and these tools that people are using. (12:10) So do you see that as the future of the business or do you think the future of the business is still going to be a combination of the human in the loop one for maybe a pharma client that wants some extra due diligence um or for a very particular marketing leader who wants every word really thought about um whereas for most people the other one would be the right service. (12:28) Yeah, I think it's going to be a combination and I I'll step back and give you an example like um the amount of content that's being produced is looks like this, right? It's like exponential, right? So there's not less content and most companies ration content to translate because of that cost 20 cents a word. It's too expensive, right? And time involved all that kind of stuff. (12:45) So I taking a step back especially, you know, when I think about my days at eBay, we thought a lot about digital footprint and we thought about conversion rate, right? What is digital footprint? So if I'm a US company, my digital footprint, the size of my website, all the content, the blogs, da da da, is probably pretty large. (13:04) But in the markets that I'm growing, let's say it's uh Germany, France, Japan, those websites are teenytiny. And so how does Google look at them? Google's like, "Oh, that's a teeny tiny website. Doesn't have great engagement, so I'm going to knock it down in terms of uh SEO." Right? So what the problem that we're helping our customers solve is not only we helping them create localized websites, applications, etc. (13:34) that boost conversion, but also boost engagement, right? So I'm landing on this page. Oh, great. It's in German. I'm going to read this. Right? Um, but we're also helping them increase the digital footprint so they can be more competitive in those countries because now it's a large website. It's got great engagement, great conversion rate. (13:50) It's sticky, less bounce, all that. So Google looks and say, "Oh, great. I'm going to move you up." Interesting. Interesting. See what I mean? So it's so my point on that going back to what we're talking about is that I believe for a long time our customers are going to be they need a combination. They need I need this volume, right, which we're going to solve for them. (14:08) We also need very very high quality for certain types of assets. So it's really going to be less about hey, we're going to move everything to AI. It's going to be more about we're going to do a lot more with the right type of translation. That is fascinating. Got me thinking about the future of our SEO as well. So that was I hadn't thought of this. This is very interesting. Thank you. Um okay. (14:26) So we're here to talk about a problem that you have wrestled with in the last 12 24 36 months. Um tell us about the problem that you decided uh was the one you wanted to talk about today. So it really revolves around this conversation around this AI. (14:45) So right so remember like you know a couple years ago we were doing things faster, cheaper, but we weren't necessarily doing them better, right? So I I remember very distinctly when GPT or OpenI you know announced you know hey you can now use GPT. So that Saturday morning I you know hopped on and logged in and start using it and immediately the light bulb went off in my head and we were the leadership team all had to have the same kind of epiphany. We were all slacking each other. (15:12) One of the things about Smartling is it's a SAS it's a cloud first company over a decade of machine learning neural machine translation which is a form of AI. So we immediately saw, wow, okay, this is a gamecher and talked about how we're going to apply it to boost the quality and solve a bunch of linguistic problems that we were stuck on. So um that week I pulled the LT in and immediately reset the operating plan. (15:36) Uh put in place a uh a project to go out and make sure we had the right talent. So hiring in the right talent and reorganized our tech team structure and process to enable rapid innovation that delivered outcomes that matter to our customers. Right. I'm making this sound like oh it was so easy. What you're saying is super interesting. (16:02) Um how this was that moment for a lot of us where we're like okay the world is can you can especially if you were in technology once you saw the first version of it you could extrapolate out and say this is just going to get better and better and if this gets better and better what are some of the implications and how can the world evolve right like let's just assume it gets 20 times better what's that world and so I think a lot of us went through this period of like thinking through that and saying Okay, now it's time to start making adjustments because you're still early (16:32) in the curve, right? And so you can start making adjustments right now and ride the curve. Um, but a lot of organizations struggled with acting on their insights. And so tell me a little bit about you guys had this moment, you saw the technology, like oh this is this is this can change everything. (16:51) What was the time period of pontificating about the future versus acting on like here's a plan now and this is what we're doing now let's go and implement this plan. I I think this is where you know I've been doing this for a while and I've been a founder for a lot of my life and I think this is where the founder in me uh really became a a benefit sometimes it's like Dr. Jackekal and Mr. Hi. (17:14) Right. So, I remember getting yelled at at eBay for for my founderness. Right. So, anyways, uh but I think this is where I immediately saw the potential for disruption and what I and so I immediately went into founder mode, which is like, okay guys, time out, break glass, plans out the window, right? And in a big company, everyone would be like, what are you crazy? Right? You know, kind of thing. (17:40) But in our company, and I had a bunch and I have very like-minded um I have a very like-minded LT uh leadership team. So it it wasn't that hard to to con and people had this conviction. But you know when I looked at you know I just like to me there's been kind of like I'm going to really simplify it three big moments in tech in my career. One the first was the commercialization of the internet. (17:57) The next was the move to cloud and this I think is going to dwarf both of those or at least be on par with um uh with the commercialization of the internet. So when you see that as a founder you you know you have to like immediately break glass and change the plan. (18:17) It's that's the one thing that I've noticed is very different in founder organizations versus not is like the speed at which they can act the with the and the conviction with which they can act. You see this in the market today right you see Apple that has really struggled to come to terms with this platform shift in so many ways. And you see examples on the other side and Google is also an example non-founder business that has the best one of the best models in the world. (18:42) Gemini is one of the best models in the world and then try using Google Home. Google Home is the one voice product that is worse than Siri. But these guys own the best model. And so it's kind of crazy. Whereas you see companies like Meta that you might not love their strategy, but there is conviction and strong action, right? And there's a speed at which they're moving. (19:06) And I I find that really fascinating between like the founder mindset versus non. So you you got the troops aligned. Everybody saw it. Everybody instinctively figured it out themselves as well. You guys decided we need to move quickly and started moving in a particular direction. What were the biggest hurdles as you were going forward? What were the ones? What were the problems that you knew you would want face as you solved this? And what were some unexpected things that came about that you had to wrestle with? Yeah, that that was the that was a big part of the problem, right? So, everyone (19:37) was really excited about this and so everyone wanted the cheese, right? You know, the who moved my cheese thing, right? So, everyone wanted to to do it and so one of the really hard things I had to do was recognize, okay, this can't be jump ball. I've got to have some structure here. (19:54) So we we always start with um I at least in my opinion we always I always like to start with and we like to start with focusing on delivering outcomes that matter to our customers right so what are those pillars right so in our particular case it's quality speed cost and ease of translation those are the pillars right so that's the northstar that drives everything in our business including the product roadmap so once we aligned done. (20:20) We we're we aligned on those obviously and then how do we think AI is going to affect that now how do we actually integrate it and build it um and when we say build it something that's never been built before right so um I had to make the decision to sort of create almost like a production line within our tech organization that started with R&D right so like this R&D department also once again like our engineering team is like one of the best in the world I we 3,300 production releases a year, right? We're very very efficient team. Um 99.99% uptime, 98% on time (20:57) delivery. This is one of the best run uh engineering teams I've I've ever had the privilege of being involved with. I didn't want to disrupt that. So I we hired in some really great AI talent. We structured a a process that we would do R&D, but like once again the R&D has to have guardrails around. It can't just like spin around in a loop. (21:22) So we set up basically um confidence levels and time boxing. So we would come up with an idea and we would say, okay, we think if we do this, it's going to solve this particular customer problem. Okay, great. (21:39) you have x amount of weeks to to prove that out and you're going to be able to measure it with a confidence level and you have to achieve a certain confidence level within that time period or it's you're out. Interesting. It gets bru and we move on to the next thing. That way we don't get into it. It's really key. Uh unless you just work through a bunch of ideas, right? So you're like, "Okay, we don't know like let's try a bunch of shit. (22:04) " And so then you're working through it, but you have a mechanism to not get stuck in a loop where it's just not working. It's not good enough, but you keep at it and then you're too distract like you don't have the resources to do all the other ideas. So it's like that's really interesting. Yeah. I mean that that was really key to it because you could literally people you know like human nature you get really excited about your idea, right? And you think no, no, I could I just I just need another work. I can make this I can I need another week. I can make this work, right? we give a little bit of latitude on that, but at some point (22:28) we got to uh let's move on. We got something else on the list that we think will work. And so that way we're able to, you know, continuously move forward. And then once it gets to a certain confidence level, then then uh it it rolls into the product and engineering world, right? Like okay, roadmap, we've been talking about this, great news. We're at 80% confidence level. (22:51) We're very confident that we can get done what we need to do. now let's start moving it into the machine and then it and it gets developed then it gets released. So as you look back, you know, I I find that usually some of the best learnings come from I'd come at the end of having wrestled with some really interesting problem that wasn't an easy problem to solve. (23:18) And then you kind of reflect back and you're like, okay, this is how I would do things. This is how this is what I have learned that informs my view on how I would solve other sorts of problems moving forward. when you look through this experience and you're probably still in it, right? Like you're still because the technology is still rapidly evolving and so what you can do with it for your customers is probably rapidly evolving as well. (23:40) So you're probably in some way still in the midst of it, but you have shipped some hu really interesting innovations into the market. And so at this point when you reflect anything you know the way that you thought about creating this unit within the organization that was your R&D and experiment unit and they had a framework with which how they were looking at these things and once it got really good they would ship it out. (24:05) I imagine you learned that somewhere else and have perfected that over time. Any other things that you've learned through this experience? Yeah, I think the key for me is always having that northstar and I learned this at um I think I learned this at eBay really. They were probably I had uh um I was really fortunate to have a great um mentor there, Christopher Payne, who like used to he was he would always terrify me. He just beat the heck out of me. (24:31) It was always like, okay, the two things you have to keep in mind is what is this going to do for the customer and is it material? Like is it going to move the needle, right? You know, is it big enough? And so that's the challenge that's the gauntlet I I always lay down for my team which is we talked about what is that northstar so everything that we do has to have a material impact on quality speed cost or ease of translation right and if they can't demonstrate that if they can't um you know communicate that as part of like the BRD process (25:03) right it gets it gets it doesn't even get off the ground so having that that customer outcome uh those pillars of a customer outcome are super helpful and create a great discipline. Um and then on the softer side, right? So when you have to go through this type of change, like I'm I'm joking like this was incredibly disruptive and painful, right? I mean, uh I remember Yeah, it was it was painful because it's somewhat of a traditional industry as well, right? And so it it's different when you're trying to evolve a traditional industry versus, you know, something that's at the cutting edge of innovation. (25:41) Yeah. Yeah. Uh and for us, um there's there's two elements of that. Number one is, you know, like making this change across an organization that was used to doing things for a long time and had a lot invested in that, right? So um I probably I'm sure I went tops down too hard uh at first, right? um the founder and me I guess uh and you know that doesn't really you can't go tops down you it doesn't really work it creates a lot of auda so I had to back off a little bit and spend time talking with the team listening listening to the team um and making those adjustments you have to be a good (26:18) listener a good communicator but you also have to like have like the steely resolve on what your vision is and that's really that's a tough balance because otherwise you just get stuck um so those are those were some of the big challenges then with customers, they don't believe you. They're like, "Oh, no way. (26:36) I've been doing this for too good to be true. That sounds too big to too be good to true." So, we kind of have to put our money where our mouth is. And the way that we've solved that is generally with a proof of concept. We'll say to them, "Fine, great." Um, I remember having a funny having a funny conversation with a customer like where I bet the customer 100 bucks that uh that we could deliver what we said we could deliver. (26:58) Uh, and um, we did a proof of concept on our nickel and we delivered and now they're a multi-million dollar customer. Wow, that is fascinating. This is this has been great. Um, Brian, thank you for sharing all of that. I think I learned a ton along the way. A couple of questions for you as we wrap this up. (27:18) Um, I'm going to do different questions than than the ones we thought we were going to ask. I think these are better. What is some who is a person or a book or something you have experienced or consumed earlier on in your career that has had a very lasting impact on the professional that you are? Uh well that's that's funny. I actually have them actually have them right here on my desk. So, uh, Jim Collins, uh, obviously Good to Great. (27:46) Um, Execution by Larry Bosidity, uh, Lessons from the Top by Jim Citroen, and, uh, and Jack Straight from the gut, Jack Welsh. So, like those are all great books on how to, um, you know, really to, you know, when I was young in my career, like you gota, you look at these icons and how did they execute, how did they do so well? And I think that these were the these are the ones that, uh, formed me quite a bit. (28:08) That's brilliant. Um what about as you you are in a world you're a technology executive you've got to keep on top of things. How do you do that? You're really busy and then there's all of this change that's happening. What is your mechanism or method to stay on top of things so that you can make decisions maybe a little bit before your competitors do? Do you like to read a lot? Do you carve out time for it? How do you do this? I I do. That's a that's a great question. (28:42) So, I I always I always I always tell new hires it's really important uh that you be intellectually curious and I always have um and I think that that's a that's a a that's a really important strength. So, I carve out time every morning. I think I think it was um um Jamie Diamond talked about this. (29:03) He gets up at like five and he reads he reads like seven different newspapers, something like that. And I don't get up five, but but I do get up early and I make sure that I read uh a number of different uh periodicals. I've got newspapers, etc. I have a number of different newsletters that I get. (29:21) And then I try and spend a lot of time talking to customers, right? So that's sort of like those are the two big um uh the two big things that I like to morning time to just take a second like a slow morning. Bezos uh talks a lot about slow mornings and I think finding a way to have like a slow start to the day is really helpful where you can just like take stuff in think and then you go into the thick of things, right? Last question. You're in a stressful job. (29:45) You've been in stressful jobs for a long period of time. What are your what's your favorite or what are your favorite ways to manage your nervous system to just connect recharge to not burn out [Laughter] just I just cram it all just cram it down deep I just compartmentalize it and fingers crossed yeah I think you know like I learned this early on when I was as an entrepreneur like you have to learn how to compartmentalize I remember uh when we were first when I first was launching companies I I would I was so stressed. I couldn't eat. I couldn't sleep. I would (30:20) not be able to sleep. And then finally, I'm like, "This is not going to work, Brian. You either have to find a new career or you have to learn how to compartmentalize." And I And I think that that's probably the number one thing that I do. And um and then, you know, other things I I like to um you know, I like to exercise and you know, do some yoga or do something. (30:37) What's your favorite form of exercise? Um probably um you know what I'm really liking right now is is yoga. terrible. I'm terrible at it, but it's like it just makes me feel good. There's a I think there's a massochist in most founders where they like doing things that they suck at because it's like a humbling but also very rewarding experience cuz you start and you're like at it, but you're so good at like being an entrepreneur and you're really bad at this thing. And there's an interesting dynamic there. (31:05) It's so embarrassing. I'm so unflexible. Brian, this was fantastic. Thank you so much for your time. Thank you for sharing this with us. we'd love to have you back um and talk about how this industry that I'm sure most of our listeners do not really know much about is evolving with AI. (31:22) So, thank you for your time. Thanks. It was great to talk with you. [Music] Thanks for listening to Topline. New episodes come out every Sunday. If you like what you heard today, subscribe so you never miss an episode. Leave a fivestar review and share it with your friends. (31:43) Don't forget to join our topline Slack channel to connect with us and discuss the topics we cover with other listeners. Click the link in the show notes. Lastly, if you're interested in joining Pavilion, you can learn more about membership at joinpavilion.com/membership. Thanks again for listening. [Music]