Jul 4, 2024
36 min
Episode 27

TOP CEO: Smartling - 'The Translation Game' (With Bryan Murphy)

The Detective  00:00

Imagine you’re the CEO of Smartling, a cloud-based translation company using neural machine translation, and AI technology. Now picture this pivotal moment in your journey 18 months ago, with the rise of Chat GPT and generative AI on the horizon, you, the new CEO faced a significant challenge.

Bryan Murphy  00:22

As a CEO, and I've been around for a while, there's only a handful of times where I've really felt like there was like in that gut like this has the potential to be super disruptive, you

The Detective  00:31

see the immense potential and threat of AI in the translation industry. But if it works, you could provide a superior service at a lower price, but a threat, because if a competitor moved on the idea more swiftly, it would put Smartling at risk of losing its market share to more technologically advanced rivals.

Bryan Murphy  00:52

We weren't the only ones thinking about this, and we had to get out and stay out in front, you couldn't say, Well, you know, we'll see how this develops, right? Because then we'd be behind it. And it's almost impossible to catch up.

The Detective  01:03

Moving to AI requires a significant investment in new technology and new company processes. Should you take the gamble? Or stick to what you know, this is the story of Bryan Murphy. And this is the translation game.

Ben Kaplan  01:24

Bryan, take us through your thinking process on the rise of generative AI. And what that would mean for a business like yours that does translation. Obviously, there's a lot of potential to automate tasks around this take us back 18 months, and this pivotal moment in the company. And what you were thinking what you were worried about, and the actions you felt you needed to take 18

Bryan Murphy  01:50

months or so ago, maybe a little bit longer than that. Now we were sitting around and right also this thing GPT comes up on the on the horizon. And I remember it was over the weekend, I think it was I got access to it, start playing with it. So to my leadership team, and we kind of all immediately had the same epiphany. And we're like, wow, if this thing works, you could be really, really disruptive to our industry, we also got the background on sparkling is that we're a company we're a cloud native company, we're the was probably the first cloud based translation company software company in the industry. We've been a longtime a builder of cloud based products with you using neural machine translation engines and machine learning to create very cost effective high quality translation. So this was a little bit of a natural for us

Ben Kaplan  02:35

to see it. Why did you see the disruptive threat right away? Was it more of a risk of just like AI is just gonna be able to do our jobs, it's just gonna do it? Or is it more like AI is going to drive down the costs, because the more things can be automated, it could possibly be used for things like quality checks, or other things that maybe you would need a human for that you might not need a human for anymore. What was the disruptive threat, specifically,

Bryan Murphy  03:00

really all the above, we felt that Gen AI large language models in particular could have a significant impact on translation because it can do the translation, right? It's what we found is it's not as effective over time. Now, we found that it's not quite as effective the cost more and it's it doesn't deliver as good a quality as neural machine translation engines. However, what it does is it is very good. When you build it into your process at solving particular quality problems that you end up with neural machine translation, you

Ben Kaplan  03:30

saw the threat of generative AI, you felt it would be disruptive, it could change everything in your business. So what was the questions you faced at that point? You know, you had the long weekend, you're playing around with this, you're like, wow, I can do quite a lot. What actions and what is the timeline and actions that you have to take in response that are fairly dramatic actions for your business,

Bryan Murphy  03:51

you know, as a CEO, and I've been around for a while, there's only a handful of times where I've really felt like there was like in that gut like this has the potential to be super disruptive, right? It was sort of in the beginning with the internet, right? And then cloud, right? You knew just instinctually that that was going to be hugely impactful, even though at the time like people use I got thrown out of boardrooms for suggesting that their on prem solutions should be put up into the cloud, if you remember those days. And this one was one of those things where in the gut I fell in the team felt, wow, this is really got the potential to be impactful, but we didn't know. So the first thing we did was we sat down I said, Listen, kind of like pump the brakes on what we're doing. I want to reallocate resources here and people with time. And I will take a really hard look at this because we had no idea if it was going to work or not. And we quickly quickly realized that it was going to be expensive. To do this. I had to reorganize the team. I had to bring in AI and machine learning specialists and they're hard to find and they're expensive. And we're at the rejigger our processes and software to make this work. So this was not a light undertaking. This was a whole reboot to make this successful. We couldn't dabble with it. I knew I had to be that we either had to be all in on it. And if we dabbled, we lose, you're

Ben Kaplan  05:01

probably thinking in your mind that if you're seeing this, if you're noticing this, other people are noticing this too. So it's not just so much of like, oh, we can keep doing what we're doing or try to embrace where technology is going. It's also the risk that other people will embrace where technology is going, maybe they could do it better or faster than you could, which means there's winners and losers in the industry, from this, my experience in these types

Bryan Murphy  05:25

of situations is it's a race, right. And if you get a head start in the race, and you maintain that Head Start, you stay out in front. And I felt like if we were right, and this was going to have the impact on the industry that it was, we weren't the only ones thinking about this. And we had to get out and stay out in front. So the risk was, it couldn't be status quo. You couldn't say, Well, you know, we'll see how this develops, right? Because then we'd be behind it. And it's almost impossible to catch up. When you're behind like that, because

Ben Kaplan  05:51

of the nature of your industry. Because other industries, maybe what's less clear, like, oh, in a general sense, generative AI, seems like it can impact us. But for you, it's like no general AI can really do potentially the core function that we do. So we've got to figure out how to use it better than other people.

Bryan Murphy  06:08

That's exactly right. And I think we took that lead, we've got that lead. And now our job is to stay out in front, I think about translation, right? So I come from the clients, I was an executive at eBay, and many other large, you know, b2b and b2c type of companies. So I come from the client side, and I know that I need to have my global experiences multilingual, right, that's critically important. I have to have digital footprint, I have to have high conversion rates, all of those things. So it's, it's really important. The problem with traditional translation is that it's slow. And it's expensive. I mean, traditional rates around 20 cents of words. So I can I'd ration what we translate, maybe, maybe just this much. So I very much do translation as being a service like cloud service, or data storage, where I want to make this a utility. That's cost effective, high quality and instantaneous usage for our customers. And that's why that's why our business is growing as fast as it is, because we're able to do that better than most of our competitors, I still think there's there is we're still running hard, I think there's room for improvement. But that, to me is the name of the game, at

Ben Kaplan  07:15

this point in time, when you're going to sort of make this shift and resources in personnel and more. How big is the company at this time? It's like, how many people? How much revenue are we talking about? In a broad sense,

Bryan Murphy  07:27

we've got a couple of 100 employees around the world. So it's, you know, it's not small, you know, it's not small, it's not huge. But you know, it's like with anything, when you start moving, you know, moving the cheese around organization

Ben Kaplan  07:37

and a couple 100 employees at the time, you know, 18 or so months ago, how many of those, what percent of that needs to shift? What needs to happen of that, let's say a couple 100 employees at this point,

Bryan Murphy  07:49

I would say that virtually 100% of the employees have been affected by it. I would say that we've probably taken, you know, from an engineering product resource,


a third. Okay, so third is going to dramatically shift like just different sort of people, everyone might be affected, but a third of just like a totally different type of person to embrace it. So you make the decision, it sounds like pretty quickly. And you've got a couple at that point issues, you need to figure out one, can you actually sort of embrace generative AI? Just, you know, technically speaking, right? Can you use it to become more efficient? You've got to sort of see, can we prove out the concept of this, but then to it sounds like you've got to do some hiring that has to take place fast. You alluded to these are maybe difficult people to get particularly at that time, when it's suddenly in high demand, everyone's sort of seeing this. So how do you tackle some of those challenges?

Bryan Murphy  08:42

So that's exactly right, we get when we first started, like, you know, GPT was that we access the API wasn't even reliable, you know, your, it would timeout, and, you know, it wasn't ready for prime time. So this was early days. And I think that we did a good job of evangelizing what we were trying to do. You know, if you think about AI, I think a lot of people felt this excitement, they felt this transformative nature of this product. So we approached it as a vertical, you know, a vertical software application company, leveraging AI or having an AI application that's attractive to a lot of people, right. So if you think about the AI space, there's three places you can either, you know, work for, you know, you can either do work for one of the LLM type of companies, the fang companies, right, where you can go into the pick and shovel business, you know, the data centers, etc, chips, etc. Or the next and probably the biggest bucket, our, you know, vertical AI software application companies, and we're one of those. So I think we actually were pretty attractive to a lot of people to come, you know, as a place to work and to grow, and then that helped us.

Ben Kaplan  09:45

But did you have to sort of do this new recruiting initiative? Did you have to dramatically change anything about how you recruited or reprioritize or get more recruiters or anything else or is it No, we're just going to change the job. descriptions on our site. And you know, on Monday, it says these 10 jobs and on Tuesday, it's the separate tech job.

Bryan Murphy  10:05

Now, it's tough because you're, you're trying to hire people that, you know, they have expertise, right, or, and what we did a lot of is retraining our team or existing, we've got really, you know, top notch engineers and product people. Even then, you know, even if you bring in someone, they don't understand your business. So we did a combination, we obviously we hired some really great people, but we also probably what we did more than anything was getting our existing like the people that had the the ability to pivot and to think this way to train them and give them the expert choice expertise that they needed to develop this.

Ben Kaplan  10:44

Okay, so you're sort of trading them saying this is a new direction? What did you have to do, though, from a point of view of sort of CEOs vision perspective? What were you giving up by doing this? What I mean by that is, presumably, there was some other kind of visionary said, Hey, this is our, you know, this is what we're doing. This is our mandate, this is our 123, you can't probably just add four or five, six on top of 123, you probably had to cut some stuff. What what changed? What did you shift to say, we're all in on generative AI?

Bryan Murphy  11:13

It's hard for me to remember exactly what got cut from the product roadmap, but a significant bucket of stuff did get cut, right. So it's one of those things where, you know, people watching this in the company will probably laugh, because like, I look back at this, and, you know, it's kind of like, you know, I jumped up on my soapbox, and they say, pull the cord, break the glass, whatever stopped the train, we're gonna go do this. And as a CEO, it's a very, that's a very risky maneuver, right? Because it's like you get, you get accused of chasing the shiny object, right. And if you're wrong, it's a big credibility of that. So. And I would also say that to change the course of a train or a ship or whatever analogy you want to use, you can't just say something once, and then you know, you dark to the left or your dark to the right. It's like constant like, every, every day, having this conversation with people grinding

Ben Kaplan  11:59

them, do you have to keep reminding people of what this is? It's not like you say it once, and everyone's on board. But you have to keep saying, and it's also the applied division. Because maybe there's some other decision someone's makes, like, should I do? Should I spend my time on this? Or should I spend my time on that? And door? A sounds pretty good. And you got to kind of remind people that like, no, no, we have this. We're all in on generative AI. So even though door A is probably needed, we're still gonna take door B, right? It's a lot of like, smaller decisions, that kind of cascade down from the overall direction.

Bryan Murphy  12:32

Yeah, that's right. It's, it's not rocket science, but it is discipline. And I, once again, I feel like people watching this will laugh inside the company, but like, we have an operating plan, we review it. Every year, we do a scrub to kind of do a stand up for days of the week, there's five things on that operating plan. And that's the only thing we talk about. So you know, it's got to be one of those items, AI became at least one, maybe two of those items. And so just by force of that focus, like that's, and frankly, adjusting people's, you know, KPIs and their compensation and all that kind of stuff, you got to like, you can't just say something, you have to actually get into the guts of the organization and re re plumb it is probably the best way to describe

Ben Kaplan  13:15

it. And it was this at all an added challenge. So I think you started in the CEO role at Smartling. Like kind of the end of 2021. So this wasn't that far after you weren't like long in a tenure at this point. Right? Where you had like, a lot of trust reserves built up like oh, you know, yeah, Bryan see some stuff? We don't see. Just go with him. Go have a Bryan does is that correct? Was there a little bit of an added challenge for that? Cuz you're relatively new on the job?

Bryan Murphy  13:41

Yeah, totally. Of course. I mean, people are like this guy, you know, people have been, you know, working in translation for decades. And here comes this guy who, who's never been in translation before. So what's his credibility? I was aided by the fact that I've got, you know, a very, very talented leadership team and technology team that also grokked you know, this, this whole

Ben Kaplan  13:58

concept at the top everyone bought in, so that kind of gave it a sense of momentum, even though you were still new at this point, we should mention, I mean, you're not new to important roles and companies, but you're new to this particular industry, when you accepted the C E O position.

Bryan Murphy  14:13

Yeah, that's correct. So, you know, that being said, you know, of course, when you're making changes like this, and you're reallocating resources, etc, there's always a lot of friction and challenge and people things, people dynamics that you have to manage. And that's always people sometimes people overlook that. But that's probably one of the biggest challenges in making these types of

Ben Kaplan  14:31

changes. Well, what about just on the technical side, because as you're starting this process, and you're seeing it through, sometimes with these kinds of major changes, you're making some bets or some assumptions that have profound implications on everything that comes downstream from it, meaning you're like, ah, we can approach this two or three different ways. And if we pick the one we you know, we might go down six months or nine months, the rest up, we pick the wrong approach. Now we got to go back, right? It didn't pan out. So is there any risk of that? Or how are you just sort of managing approaches to incorporate generative? Yeah, I

Bryan Murphy  15:07

think we've got a process that lended itself to this, you know, a constant, you know, continuous development process, we do about 3300 releases per year. So, you know, we've got this constant, it's not these big monolithic releases, we're able to be pretty nimble. The other thing that we that we did is, we established a research and development department, and they were really focused on AI. And so what they would do is they would very quickly experiment and iterate with the large language models and how we use them, and get them to the point of competence, not 100% confidence, like because we're, we're inventing all this stuff. So we say like, we're like, 80% confident that this will work. And we would say, okay, 80% is our threshold sounds good enough, let's get it into production into product and engineering and test it, right, and then then have a clear set of KPIs or outcomes was really important business outcomes. So we were measured monitoring that in real time. And we would see, okay, did we move the needle or not? Did it go the right way, and then constantly, you know, without continuous development, okay,

Ben Kaplan  16:06

so it's just you have this kind of experimental lab that kind of could work outside of normal things and just try to do stuff quickly and not have to worry about as much business considerations or other things like that, they could just try stuff out just to validate stuff before you brought it into a normal kind of like business process to bring it to market.

Bryan Murphy  16:26

Right. And that's, that's not a cheap process, right? That's an expensive process. And you know, you leave a lot more on the cutting room floor, then you get into, you know, into production. But, but that's how, you know, that's, that's what's worked for us. And that's part of what's working for us that collaboration between research and development, and the engineering and product teams, is really how you get get this stuff done.

Ben Kaplan  16:47

What about just the business risk? What I mean by that is, let's say you're successful at everything we've talked about thus far, meaning you kind of uplevel your existing team to do generative AI, you bring in some new people who have some specific expertise, you don't have create this research lab, they start experimenting with stuff, they find some things that are impactful, you bring it over to your core products, you operationally sort of bring it in, you're starting to do that. But there's still a risk that you could bring prices down. And it impacts your bottom line, like, Hey, we're used to getting $20, a word for this kind of client, we've done everything we said, we're going to do using this type of technology. And now we can charge 10. And it's half. And that's a risk to right at the end of this whole thing. It economically hurts you. So even though changes, your pricing changes coming, we have to do it. But how do you manage that as well? Because you're saying like, hey, we might be able to last a little bit longer and charge more? If we don't do it, where do you or do you keep prices up high? You're just like, hey, we can add profit margin to this. And when competitors force us to lower prices, we'll do it, but not until they force us to how do you think about that? Yeah, I'm

Bryan Murphy  17:53

a big believer, and you know, the innovators dilemma, you know, you know, faster, cheaper, better, I just think that, once again, I view this as like cloud services, or data storage, right, the price per unit, whatever that unit it is, is coming down and it's got to come down the technology exists the genies out of the bottle. So we make it up in volume, right? So and I, I know that like if you pick pick a statistic, like 1%, or half a percent, or 10th, or a 100th of a percent of the world's content is is translated today is multilingual. Right? So it's a very, very small fraction of total content that's actually translated. And you think about the amount of content that's being generated daily, I know that our customers have to ration how much they translate because of the cost and time and complexity involved. So the more I can, the more friction I can take out of that process, the more they will translate, that's been our experience, as we brought prices down, budgets haven't come down, revenue hasn't come down. They're just getting a lot more value. Bang for the Buck,

Ben Kaplan  18:57

if I understand correctly, I'm paraphrasing. It's not even just that technology's gonna make things more efficient, it's gonna bring prices down. So we've got to go acquire new customers, we don't have, you're saying, even with your existing customer, that there's untapped demand there that if we bring this down, they're going to do more. And heck, if they have their whatever their annual budget that is, you know, you know, 2x And suddenly, that amount of work they could do now cost 1x, there might just do twice as much, and you'll be fine. Because they're able, they're able to do that. And it will support the revenue and maybe even grown over the long term because then they'll say, hey, I can afford to do more, I have to do more. I don't need to ration this as much. They're

Bryan Murphy  19:37

getting significant value by creating multilingual experiences they've got they've got a bigger digital footprint drives more traffic, right? It's better for SEO, they're improving. We know that improves conversion rates, we see that all the time customer engagement CSAT all that. So the real strong benefit to them for creating, you know, like in market content. So for us, the trick is it's just been expensive. The expensive and slow. So the faster the cheaper, the higher quality, we can make it for them, the more seamless the easier we can make it for them, the more the more they're going to use take us through


then up to this point, all of these decisions like what period of time is of this? It was 18 or so months ago? How much time does it take you to get all this in motion? is six months? Is this a year is just a couple of months? What sort of take us through just getting all the wheels turning?

Bryan Murphy  20:24

Yeah, I would say it's actually probably been, it's probably been longer than 18 months, I have to go back and see. But we launched our AI solution in January of 23. So last year, you know, so basically, January, February is like the first month we started creating that content. Now, about 65% of our translations are AI translation. So grew that fast. Between then that and now.

Ben Kaplan  20:53

But at what point in that timeline, but even before getting that did you start to realize like it starts to feel like this is going to be successful. Did you start to sort of say, you know, hey, I made this bet, I want to embrace this new technology. And I start to feel like we made the right choice.

Bryan Murphy  21:09

I feel like probably about six months in, you know, we started to feel the, you know, the needles, the needles started to move from our ability to to improve the speed at which translations were getting done. Right, we could see the quality, we could test the quality. And we could see this happening at scale, right. So that that it started with maybe we did like a hand like we did very, very little and in January, right of 23.

Ben Kaplan  21:34

What does scale mean? How many translations? Is that? Or how many words? Is that? Or what is full implementation mean? Yeah,

Bryan Murphy  21:41

I mean, we processed you know, 7 billion words last year. So that gives you a sense of the scale, massive,

Ben Kaplan  21:46

massive scale and you you start to see it working after six months, what were the next six months like? Or what was everything else? Like? Was it just fine tuning and tweaking? Or did you start seeing like other players in the industry doing this more competitive? Other things? What did you sort of experience after just like you've launched the product, January of last year? And now what else do you have to do? It's not just the launch? Yeah.

Bryan Murphy  22:10

Right. So I mean, we were we just came back from one of our big industry events in Dublin a couple of weeks ago, and literally, like joke was, we should have a drinking game, you know, use the word AI, and everyone's using the word AI. So everyone's talking about it, there's, I think a handful of companies are actually doing it and delivering real business outcomes. The reality matters, though, is like it is like it is people are really paying attention to and out figure out how to make it work. So I think for us, as we got confidence in our ability to do this, it just pushed us to go faster and harder. Right. So we are pushing more resources into it, we're getting a lot smarter about how to use it. You know, it's just like every month, you get better and better at this thing. And it is a cumulative process. There isn't any one particular breakthrough moment, right? It's just literally grinding every day, getting better and better about what you're doing and learning in the process.

Ben Kaplan  23:05

And what was the response from customers? What did they did they understand it? Or was there a sort of hesitancy to say, like, oh, we thought we're, you know, we're paying for humans or something that isn't, you know, that can't hallucinate and make stuff up that can't do things. Do we have confidence in this? What was what was there anything you had to overcome there?

Bryan Murphy  23:26

Yeah, there's no confidence. Like, that's the big thing. Our industry is very much about quality. And so one of the things we also had to do at the same point, I, you know, I recognize it, okay, if we launched this thing, people are not going to believe that it's going to work, right? There's just this human disbelief and fair, fair, right. You know, that's, it's brand new, no one, like no one knew if it's gonna work or not. So we ended up standardizing on a, we built a whole quality process into our business that we had, but really, it wasn't as robust as it needed to be. So we built this whole quality process that's called L QE, or language quality estimation, which is sort of in process. And then l QA at the end was sort of an audit at the end, we standard on standardize on a process Correct. Terminology or methodology called MQL. Okay, so that's just what we just we built that. And we had to build that, that. So doing the as one thing, building up the standing up to this very rigorous quality process was another thing, and we needed to do that. And so it came down to we'd go to our customers, hey, we've got this thing. It's gonna be faster, better, cheaper, would you like to use it? We recommend it. And they're like, Well, I don't know, how do I know it's gonna work? Well, let's test it. So give us some content, we'll run a test. And we'll run it through this quality process. And part of that process is actually a third party audit. So it's not us telling you it's good. It's actually a trained third party, and the results came back good. And once they saw that data, then they got a lot more comfortable and they kind of stepped into it. So many of our customers started out with tests and then they started loading more and then eventually loading more and more of their content in Intuit as they began to trust the process,

Ben Kaplan  25:02

I see so because of the nature of generative AI, you almost have to innovate this whole other trust process, also, and it sounds like embrace some standards that were outside of your company that you maybe didn't embrace before. But you felt like this was now warranted. You needed this, including third party validation on some of this as well. Yeah.

Bryan Murphy  25:23

100%. I mean, listen, AI hallucination is real, right? In case you've never experienced, it doesn't get everything right. And so anyone

Ben Kaplan  25:29

listening, a lot of people know this at this point. But it's like AI can make stuff up in a very simplistic way just based on the probability of certain words or concepts or phrases, and it can easily get something wrong just by the nature of things and chance. And so that's what we're talking about, and on a important business thing, translate into another language, you just don't You don't want things made up. Exactly. Right. So

Bryan Murphy  25:51

what so if you're creating content, let's say you're creating, like a homepage or a marketing email, right? And even if you want to create a used AI to create the email, let's say, and then you say, Great, I'm gonna create this email. Now I want it in 15 languages. Okay, great. So now it's, you know, English to Japanese. Take that. Okay, great. What did you just write in Japanese? You have no idea. You don't speak Japanese. Right. So that's where that caught that LTV and LGA process is so critically important. So now we can, we can read that. And then we can say, Yep, good to go. We've checked it, it's good. Or you know what, hang on a second, it needs to get reviewed, or edit it because we caught something. And that's where we bring the human into the loop. And that's what we'll bring in an expert linguist, right translator who's English, Japanese language, bear and a certain maybe he's automotive content, or finance, whatever it is, reviews and says, good to go. Or I've got to make edits. You said,

Ben Kaplan  26:47

there's certain things that generative AI can do. That sort of traditional more like machine learning algorithms and systems cannot do as well as is that related to understanding concepts and how things are together? We're like, okay, it's a literal translation of the word. But this concept, there's context for this, that really requires nuanced human type understanding of what you're saying, it's not a literal translation. Is that the type of things you mean that that sort of improves the quality?

Bryan Murphy  27:18

Yeah, one of the things that the team has got really good is building engineering problems, right. So you could use LLM to solve specific problems. And I'll give you an example of one. And so morphology issues like neural machine translations have trouble with gender, which anything because like, if you'd look at 50 different languages, they all handle gender very differently, right. And so you

Ben Kaplan  27:38

mean, this coffee cup, whether this is masculine or feminine, can vary in a lot of languages, and it changes the ending and what's correct and all of that, right.

Bryan Murphy  27:47

So if you ever like I used to, I used to live in Tokyo, and I tried to speak Spanish, like, you know, if you ever tried it, like, it's hard, right? So what So we've created to give you an example, we have patents, we have patents on some of these products, we have 30, different brown 30 different sorts of things that we do, but one of them is morphology, right? So we'll we'll review the content, we'll insert glossary, and we'll make sure that the the gender is correct, right. So that's a problem that we saw, that's a very typical sort of problem. So now, the, the translator doesn't have to deal with it. That is a quality translation that runs through, which is great. And by the way, one of the other questions a lot of people ask is, Well, gosh, have you tried to put translators out of the business out of business? And I say, Absolutely not. I want them to be more productive, actually, we're working hard to make sure that we pay them a little bit more the same or a little bit more than they were getting paid before. I just need them to be more productive. And also, I think that I also want them focused more on bigger challenges, right. So like, I don't know, that's particularly rewarding to fix a morphology problem, right? Like, that's kind of like a, that's a daily grind. But I think there's, you get now you get it down to the point where you're solving really sort of interesting translation problems. And I think that, that

Ben Kaplan  29:00

I see some maybe like the cultural context around this where this notion of a concept of a bucket list, it's not like a bucket list is a maybe a term that's a Western like, Oh, this is the thing that you're going to do you know, the things you need to check off that you want to do. It's not literally a list that's in a bucket that maybe I don't know translates to every culture, maybe some maybe not all, you can solve for that problem. That might be a very nuanced problem. How do you explain that concept of a bucket list?

Bryan Murphy  29:27

For instance? Yeah, I think our translators are, are very talented. And they're very good at localizing, which is what you're sort of describing there, which is like, I've got a term for that in one language that doesn't transfer correctly, or translate correctly. Let's make sure that's the right thing. They take little great pride and doing that kind of work. And that's what we and we value them doing.

Ben Kaplan  29:46

Let's go to sort of present day now. You said 65% now runs through this kind of like this new generative AI system. It sounds like you give your customers the choice of whether they want to do this or not. So that means two thirds, I guess in Total volume have opted in 1/3 is not Is that correct? Are there certain languages or things you don't touch yet, or certain types of content that you don't bring into the system?

Bryan Murphy  30:08

Yeah, we give them the choice. We offer everything from traditional translation, we still do a fair amount of that. And some people, some people prefer that, and that's absolutely fine. And then some people have gone the other direction, and they're, you know, 100%, machine translation. And, and so it varies, it varies by the customer, it varies by the content type. And one of the things we do a lot of is helping them understand, sort of consult with them to say, Okay, this type of content, the most effective way to do this is just straight machine translation. This type of content really is trans creational. It's very sensitive marketing content, where you really want to have a human touch on that. And the rest of this can kind of sit in the middle, and it's straight up AI translation. So that's a really important aspect of what we do is consult with our clients to help them select the right translation mix,

Ben Kaplan  30:56

I say to sort of see the level of human attention particular content needs based on importance or priority. What is the challenges to the business now? Like moving forward? What do you think about what keeps you up at night? Sounds like you've made a lot of progress on this, but more changes coming. The models are getting better. Things are changing all the time. There's, there's more players than just open a eyes Chat GPT there's other players coming in. So there's a lot of change going on, what do you think about now, that will affect your success moving forward?

Bryan Murphy  31:28

You know, probably the simplest way to describe it is, you know, you know, I was just I would say to the team, listen, you know, there's a bear chasing us in the woods, you know, we've got to run faster than that bear period paragraph. So constantly innovating, staying out in front, there's this this, we're still super early days into this whole thing. Like if I had to do an inning thing I might say anything to I don't even know what it's like, there, there's probably 50 MLMs out there at least that last count. We're paying attention to many of them, not all of them were large language model agnostic. So we're working on integrating all of the the large models. So because our sometimes our customers prefer one or the other, or sometimes they're better than the other. And each new enhancement, each each new model comes out. It's a step change with evaluate that. So there's a lot of work there. But I think that the team, we're in a good rhythm right now. We've got super talented people that are really focused on the things that our customers care about, which are quality cost, and speed, user experience. Right. So those are those are our, you know, those are kind of the 10 poles that we focus on and delivering value on those.

Ben Kaplan  32:34

What about the risks from other companies? I mean, other companies are trying to do things. How do you think about that? How do you think about your place in the market place? There's obviously a lot of players in this space. There are some big players publicly traded companies, how do you think about all of that, as CEO moving forward? Do you think there's going to be some major market ShakeOut?

Bryan Murphy  32:55

Yeah, I do. I do think there will be, this is a $55 billion industry, depending on how you measure it. I think there's gonna be some winners and losers in this. I, you know, I, you know, I kind of think that there'll be a number of specialty firms that will probably do fine, because they've got a very special specialized thing that they do. I think that the bigger companies probably have access to technology, maybe they invest in technology, they'll probably do. Okay, I think the middle market where, you know, they don't have technology, they don't have the ability to innovate or invest in innovation are going to are are going to struggle, is the answer. So, you know, this is an industry like, like any other that is, and I think I shouldn't say any other I think that our industry, like some others, like maybe content, production, etc, are going to be and are being significantly affected by AI particularly affected by AI. So I don't think I think if we had this conversation a year from now, I can't I couldn't predict what our industry would look like, but I guarantee it would be quite different. Well,

Ben Kaplan  34:00

and final question is, what do you what is your advice to other CEOs, either specifically on embracing generative AI, if you think it's going to be very disruptive to your business, or just more generally speaking, technological change, that involves changing processes? What is your advice on how to embrace that and how to get maybe a more successful result if you're going to, you know, take a dramatic kind of redirection in how your business works?

Bryan Murphy  34:31

Yeah, I think this went to me, as I talked about earlier, I think there's in my career, there's been maybe three, like major inflection points, right. This is one of them. And it's at times like that, that you stop, you break the class, you pull the pull the chain on the car, however you want to describe it, and you have to stop and you have to grab your leadership team and have a really serious conversation about how you're going to address this. And that means you're going to interrupt your daily processes that are probably for be finely tuned and you've got budgets set and expectations set with people and goals and OKRs and all that kind of stuff. You're gonna have to, I think this is one of the things where you have to, you have to, you're gonna have to juggle that whole thing. I think if you dabble in this, you're probably not going to get the out you're probably going to get if you have a business that is exposed to this kind of thing, you're probably going to get passed by

The Detective  35:22

the rise of generative AI threatened to disrupt their traditional translation models posing risks of obsolescence, market pressure and the daunting task of integrating cutting edge technology. Bryan's leadership saw Smartling invest heavily in AI reorganize their processes and hire specialized talent. The stakes were high, but through strategic innovation and relentless focus on quality, they turned potential threats into opportunities revolutionising their business model. So, what can we take from the story? embrace change, fearlessly invest in innovation, and never lose sight of quality and customer trust. These principles can guide any business through the challenges of a rapidly evolving landscape. And with that, it's CASE CLOSED.

New episodes will always updated regularly

Waste of resources our competitors are jumping the shark.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.