Sub Club by RevenueCat
00:47:50 · Jack McDermott
[00:47:50] S1: Well, yeah. I was just gonna say we we did a quick analysis on on the consumer side, and and we can look at just what's the frequency of voices that people choose. Right? And are there clusters or do they tend to kind of congregate? Because most apps, right, would allow you either no selection or two or three options. We have over 600 voices on our consumer app, and it allows you to pick your own voice narrator to read read great audiobook con... Great audiobook content. And no single voice has more than 10 or 12% of use. And so the the tail is extremely long. Right? Like, the median amount of use is probably close to one or 2%. Right? When when you think of just how, you know, people perceive voice selection. And my my preferred voice is not the same as David's. It's not the same as anybody else. And so I think that's one thing that we're seeing.
00:48:49 · David Barnard
[00:48:49] S0: That actually kinda leads into the next question I was gonna ask. I'd be remiss to not kind of voice what everybody in the industry is kind of concerned about. And I think ElevenLabs, I'm sure, is thinking about this as well. But do the big three LLM companies, OpenAI and Anthropic and Google, do they just subsume everything? You know? Do all apps or all apps just become a little interface inside of ChatGPT? Do... Is all voice subsumed? And I think that's a good example of your 600 quality voices. And I I mean, same thing with Jacob. Like, I'm not a huge fan of the voices of some of the other LLMs and... But ElevenLabs is like, that's where you lead and that's where you stand out. And it seems like that's one of the differentiation points. But, yeah, how how do you think about competing with these other labs?
00:49:38 · Jack McDermott
[00:49:38] S1: I think there's a few things that come to mind. One is I think that we have continued to increase speed over time. Right? And so we have continued to focus on voice, speech, and audio being our our expertise and have continued... I continued to double down there. And so there's lots of great models in video and image and, yeah, you know, of course, in text. Right? But but we we have continued to double down each turn to be able to to move fast and keep our edge and stay ahead when it comes to voice. I think that we're all aware though that that, you know, to to continue to add and gain value, you you need to do more than just one thing. And so I think that's part of why we... We've tried to build... Take our voice agents platform as a good example. Right? It... It's not just pick your voice and and talk. We we've built in custom evals and workflows and integrations that allow you to begin to concentrate a lot of your your work around that that specific voice experience. I think the same holds true a little bit of of what we're doing with the apps. Right? Like, if you get greater lock in and greater relationship when it comes to the brand that you have, if you've got a creator suite that extends to the mobile, if you've got a set of enterprises that that build a great text content that can then reach consumers. Part of the approach is that you gotta begin to build a bit of an ecosystem and have a lot of interlocking parts if you wanna be able to continue to advance. Again, like, one of the many ways I've been wrong in the last two years, I think a lot of people have, was that, like, I think I would have guessed, like, oh, you'll
00:51:20 · Jacob Eiting
[00:51:20] S3: just be the best voice model. And, like, that will be your edge. But like, you can kinda extrapolate far enough that, you know, Amazon. Yeah. I think of Amazon is like the sort of like Costco of like computer services. Like, at some point, they will have. Like, Amazon Web Services will have a a model that's probably pretty good. Maybe not as good as like the very, like, top edge whatever, but it's probably good enough. And so you might lose that edge. And then it feels like, at least from where I'm standing two years into this, like, crazy hype cycle, it does feel like what you're saying. It's like the the advantages accrue into the application layer. Right? It's like, what and how? And that's where Amazon Web Services is never gonna be able to clone a voice model. You know, they may be able to clone a passable voice model, but they're not gonna have all the little like, all the little application layer tricks. And, you know, was saying before the, you know, call, I played with the the audiobook stuff last year, and you were saying that there there might be, you know, more to that coming. Like, who's gonna build like a a plat... The best platform for using a voice model to do audiobook, like, generation. Right? Who's gonna do that? Maybe some people, but now you've compounded an edge that that that that honestly, it's like stuff that can be like tokens in tokens out gradient descent and like can can be taught that way is gonna become commoditized. The stuff that's not is like interacting with the people actually using these tools, understanding their human needs. I don't know. I don't know how you guys do. I'm constantly going back between like my job is over forever and like humans will actually be the dominant like software creators. Like I still don't know. Every day I go back and forth. But that's kind of... I mean, following what what you were just saying, Jack, like that that kind of promotes a like human centric version of the future of software development, which is like talking to the people and figuring out how the models actually fit into their needs might become the edge. Models. Right? Like everybody will have the... And then that's on the building side too, cogeneration and all those things.
00:53:08 · David Barnard, Jacob Eiting, Jack McDermott, Tanmay Jain
[00:53:08] S0: And also the brand side. I mean, the fact that Jacob and I both recognize ElevenLabs as being like the best voices in the industry by far. [00:53:16] S3: Certainly, I'm not gonna go to AWS first. [00:53:18] S1: You know I mean? [00:53:18] S3: Like like, I've heard of the good one. I'm gonna try the good one, and then maybe we'll look at the cost or whatever. But, like, yeah, certainly, like, that that certainly is gonna give you some sort of advantage as long as it's Yeah. [00:53:28] S2: I mean, it goes back. Like, just just animator is compounding. Right? Right. Every time someone thinks about voice or speech and audio, ElevenLabs is the name that floats to mind immediately. Right? And sure, the the model providers will go and make their best effort to release something, but our mix of focus of relentlessly beating the drum and launching often and, you know, pushing the edges faster than anyone else in the market and compounding that on the earned media and compounding that with, like, you know, a smart growth model that really enables that and really empower engineering teams. All of that just stacks up so that you have this moat. And I I don't know what you call that moat. Maybe it's old school. We call that brand or... You know? [00:54:09] S3: Brand almost seems too simplistic. You know what I mean, in 2026. Yeah.
00:54:09 · Jacob Eiting, David Barnard
[00:54:09] S3: Brand almost seems too simplistic. You know what I mean, in 2026. Yeah. [00:54:13] S0: Well, I I think that's a great place to wrap up. Well, I've got three questions that I've been asking every guest now, and either both of you can answer each one or one of you can take it. But what's the most impactful experiment or change you've implemented this year? What's the biggest win you've seen this year that you think people could take a lesson from?