The $11B Bet That Voice Will Replace Everything | Mati Staniszewski x Nikhil Kamath | WTF Online

Nikhil Kamath 0:00 [upbeat music]

Mati Staniszewski 0:07 How many times have you-- this is your f- uh, how many times-

Nikhil Kamath 0:10 Fifth. Fifth time.

Mati Staniszewski 0:11 Fifth time.

Nikhil Kamath 0:11 Yeah. But never for podcast. Always for my fintech company.

Mati Staniszewski 0:15 Of course, yeah.

Nikhil Kamath 0:16 Are you in India?

Mati Staniszewski 0:17 We are.

Nikhil Kamath 0:18 Yeah?

Mati Staniszewski 0:18 We have, uh, I think it's between ten and fifteen. I think it's fourteen people now.

Nikhil Kamath 0:23 Nice.

Mati Staniszewski 0:23 If, uh, if we count the new people that are joining or joined.

Nikhil Kamath 0:25 And, and which city?

Mati Staniszewski 0:26 Mostly Bengaluru and few, uh, close to Mumbai.

Nikhil Kamath 0:29 I'm from Bengaluru.

Mati Staniszewski 0:31 Oh, are you, do you-- are you based there or you are part-

Nikhil Kamath 0:33 Partly based there.

Mati Staniszewski 0:34 Where is the other part?

Nikhil Kamath 0:35 Uh, Mumbai and Goa.

Mati Staniszewski 0:37 Okay. [laughs]

Nikhil Kamath 0:40 [laughs] What do you think the opportunity is? What can one build in voice that can be a big profitable business tomorrow? You're based out of which city now, Mati?

Mati Staniszewski 1:18 From London.

Nikhil Kamath 1:19 Right, from London. Do you know Carl, the Nothing guy?

Mati Staniszewski 1:22 Yeah. Yeah, yeah, yeah. Of course.

Nikhil Kamath 1:24 Yeah, yeah, yeah. We're investors of his.

Mati Staniszewski 1:25 Oh, no way.

Nikhil Kamath 1:26 Yeah.

Mati Staniszewski 1:26 I'm a tiny investor too.

Nikhil Kamath 1:27 Really?

Mati Staniszewski 1:28 Yeah. What-

Nikhil Kamath 1:28 What do you think?

Mati Staniszewski 1:30 Well, I l- you know, it's so hard to innovate in hardware.

Nikhil Kamath 1:32 Yeah.

Mati Staniszewski 1:32 And Carl is one of the very few that is actually doing both innovating-

Nikhil Kamath 1:37 Mm-hmm

Mati Staniszewski 1:37 ... and got to a good scale-

Nikhil Kamath 1:39 Yeah

Mati Staniszewski 1:39 ... um, which is so hard.

Nikhil Kamath 1:40 Yeah.

Mati Staniszewski 1:41 Um, and I think he, he's, like, thinking around design and how there will be a combination of AI native devices and maybe they all take slightly different forms.

Nikhil Kamath 1:49 Yeah.

Mati Staniszewski 1:49 I think there's a, a great chance for, for Nothing. What do you think?

Nikhil Kamath 1:52 Uh, so we engaged recently actually. We got on their cap table. I like Carl. I like Carl a lot. I feel like it's a new guy who's trying something new. Uh, the sales look great. Uh, I think they went from five hundred mil last year to about nine hundred mil this year. Uh, but it's such a tough space, right? Hardware.

Mati Staniszewski 2:13 So tough. But it's, you know, I, I'm also excited.

Nikhil Kamath 2:17 Mm.

Mati Staniszewski 2:17 You know, maybe that's a little bit biased.

Nikhil Kamath 2:18 Mm.

Mati Staniszewski 2:19 But he, he can lead with voice in many ways.

Nikhil Kamath 2:22 Yeah.

Mati Staniszewski 2:22 You know, the, the Nothing headphones I think are, are great.

Nikhil Kamath 2:25 Yeah, yeah, yeah.

Mati Staniszewski 2:26 And they are actually one of the first ones-

Nikhil Kamath 2:27 Yeah

Mati Staniszewski 2:27 ... to start integrating, like, AI-assisted part of the-

Nikhil Kamath 2:31 Yeah

Mati Staniszewski 2:31 ... of the experience, so you could speak with the headphones, you could personalize music-

Nikhil Kamath 2:34 That's not out yet, right?

Mati Staniszewski 2:36 They're testing a few things in alpha.

Nikhil Kamath 2:38 Mm. Yeah.

Mati Staniszewski 2:38 But if they, if they, um, you know, if they move quickly, I think this could be, like, the first truly AI native device that's kind of with you, around you.

Nikhil Kamath 2:47 Yeah.

Mati Staniszewski 2:47 Um, and I mean, that opens so many opportunities. I'm frequently speaking with Carl of, uh-

Nikhil Kamath 2:52 Yeah

Mati Staniszewski 2:53 ... um, how, like, in the ideal future, you could potentially speak any language-

Nikhil Kamath 2:58 Mm-hmm

Mati Staniszewski 2:58 ... and it automatically gets real-time translated to any other language around you, so we can communicate and, and travel, travel around.

Nikhil Kamath 3:05 Yeah.

Mati Staniszewski 3:06 Um, break down the language barriers. But the main part that's going to be blocking it is going to be the hardware or, like, what's the, what's the device to make this possible, even if the technology exists.

Nikhil Kamath 3:16 Mm-hmm.

Mati Staniszewski 3:16 And if Carl does well with Nothing, Nothing could be the device that actually makes and breaks that-

Nikhil Kamath 3:22 Yeah

Mati Staniszewski 3:22 ... where everybody can, can have that device.

Nikhil Kamath 3:24 I hope it is. Like, after this, I'm actually going to London. I'm gonna see Carl. Their design studio is there, right? Seeing the, uh, the earbuds, the previous version. But if it's able to translate without a latency issue, and if it's able to... You, you know that device you wear like a pendant? What is it called? Uh-

Mati Staniszewski 3:43 The, the one that, uh, the-they fold it? Or the... There's, there's, like, a few, but, uh, there was the Humane, I think.

Nikhil Kamath 3:51 Yeah. Yeah.

Mati Staniszewski 3:51 But that didn't doesn't-- I think this was-

Nikhil Kamath 3:53 Yep

Mati Staniszewski 3:53 ... acquired. But there's, yeah, there's like Re-Re-Retranscribe or something like that. The one that, like, takes notes and summarizes afterwards.

Nikhil Kamath 4:02 Yeah. If the earpod could do all of that, like his earbuds-

Mati Staniszewski 4:06 Mm

Nikhil Kamath 4:06 ... if they do that, they translate, they do a bunch of other things. The thing Nothing also has going for it, I think, is it's cool, uh, because a young person runs the company. So hopefully they do well.

Mati Staniszewski 4:17 And they are trying some different things.

Nikhil Kamath 4:19 Yeah.

Mati Staniszewski 4:19 I also, I would like that, you know, pendant-- this is maybe slightly, slightly off topic, but, um, frequently, I mean, as a company now, we will attend a lot of events.

Nikhil Kamath 4:29 Mm-hmm.

Mati Staniszewski 4:29 We'll have a lot of people in those events.

Nikhil Kamath 4:31 Mm-hmm.

Mati Staniszewski 4:32 And of course, after the events, we are now trying to figure out how to take a lot of the follow-ups, action points. So we created a little voice agent that you can speak with and, um, and capture a lot of the feedback. Um, so that already kind of helped shorten the loop so you don't have to manually input everything after the meeting.

Nikhil Kamath 4:48 Mm-hmm.

Mati Staniszewski 4:48 You can speak with the voice agent. But the ideal version, assuming people opt in, and this is the, the kind of safe environment, would be to have a pendant. You go to a meeting, everybody knows that the notes will be transcribed, and then you can engage in those events, in the conversations, and then straight after at the end of the day, you get, like, what happened, what are the follow-ups. Um, we are, we are working with now one startup that I hope will, um, will do it in a good way, where we could deploy it in our, in our next summit, where everybody across, across the team could, of course, with the knowledge of the participants, take the notes, not need to worry about all the notes afterwards. I think how amazing would that be?

Nikhil Kamath 5:28 Yeah. What is it called, the startup?

Mati Staniszewski 5:30 I don't know if I can certainly name yet because they are like-

Nikhil Kamath 5:32 [laughs]

Mati Staniszewski 5:32 ... they are in a stealth mode, and they want to announce it.

Nikhil Kamath 5:35 Right. Right.

Mati Staniszewski 5:35 And, you know, we are trying to figure out a partnership. But-

Nikhil Kamath 5:38 Yeah

Mati Staniszewski 5:38 ... but there are, like, a few companies. The, you know, the, the, the form factor does matter. Like, you don't want the pendant to be too big. You want it, of course, to have a, a pretty, pretty good support, um, of course, battery and-

Nikhil Kamath 5:50 Mm-hmm

Mati Staniszewski 5:51 ... um, um, and, and, and, and, and I, I just, like, feels like it should exist.

Nikhil Kamath 5:56 Yeah. Everyone I speak to in AI, uh, and I've interviewed a bunch of them, largely in the Valley, it's refreshing that you're in a different regionThey all say voice could be next, voice is likely next, voice is a natural way to communicate that we are accustomed to, but nobody seems to have cracked it yet. Uh, do you think the medium of interacting with tech will move from hardware, which does feel unnatural to hold a thing in your hand and, like, type into it with your fingers.

Nikhil Kamath 6:35 How does that transition from that kind of hardware to voice happen?

Mati Staniszewski 6:41 Yeah. The-- You know, like as, as we think about the future, hundred percent voice will be a big interface and big way of how we interact with the technology around us.

Nikhil Kamath 6:51 Mm-hmm.

Mati Staniszewski 6:52 That's, that's we think about ElevenLabs' mission. It's to transform how we interact with technology around, uh, holistically, and whether that's, you know, calling in to customer support, whether that's how you will learn, uh, at school and how you'll deliver education, all the way through to how you interact with the robots or devices around you in the future. Um, and k- to your last part, like, I think the most exciting part of that is like h- could you have the technology kind of fold into the background, the phone goes back into the pocket, and you kind of immerse yourself in the, in the, in the, in the world around you? Um, so I think there's like, to make this possible, there's at least three things that need to happen. One is, of course, the foundational technology and research. It needs to get to the level where, where you feel like you're interacting with a voice at the level of, of, of a human. It's, it's-- You can interact-- you can interrupt it, it has the right emotions, intonation, and it's quick. Um, it has the high intelligence. And I think the current model is, is good for set of, um, set of use cases. It's not yet like a full human level. Um, so that's the kind of the first component that needs to change on the, on like, that needs to be true, otherwise people will not want to have voice agents arou- everywhere around you. Um, like you have Jarvis in Iron Man-

Nikhil Kamath 8:09 Mm-hmm

Mati Staniszewski 8:09 ...and this kind of level of quality. That's ideally where you kind of get to, uh, where it's both great voice and, and, and kind of gets all the knowledge and gets all that combination. So that's the first thing. So the voice, voice quality and interaction needs to be great. Um, second, and that goes a little bit to, to the challenges of deploying that in different, in different settings, is it's of course important that the voice is good, but it's also important that it has the knowledge access. Um, and as like we think about the, the, the, the development in the future, as we think about our role in the development, increasingly, there's a, an important combination of, of allowing customers, allowing users to bring that knowledge in an easy way, whether that's the integrations with their, with their systems, whether that's figuring out how you can deploy, um, the voice agent in, in WhatsApp and different devices that customers or people will interact with. Um, but ultimately like how do you bring all the knowledge ahead? How do you have the memory of things that, that happened in the past, whether that's what you, what you are interested in, what you know, um, if you are a, a business customer taking the CRM data and pulling that in so you have all that information. So that'll be my second one. So-

Nikhil Kamath 9:16 Mm-hmm

Mati Staniszewski 9:17 ...that's the, the second. And third one, uh, which we kind of started the conversation from is the, is the form factor-

Nikhil Kamath 9:22 Mm-hmm

Mati Staniszewski 9:22 ...which I don't think is, it has been figured out. Like, now we have all the components for a voice agent to be there and understand you. How do you deploy it? My take is, uh, it's like, you know, frequently the conversation is, is it glasses, is it phone, uh, is it headphones? And it's probably the, the, the kind of the, uh, the least interesting one, but I think it's all will be a combination of, of, of diff- form factors. Some people will definitely prefer glasses, some people will prefer the headphones. In the future, maybe Neuralink is version of that part-

Nikhil Kamath 9:53 Mm-hmm

Mati Staniszewski 9:53 ...where you kind of start and you always have voice agent available to you. Um, I'm excited for the headphones, but not the headphones that, um, that are in ear, but effectively the ones that are behind the ear.

Nikhil Kamath 10:05 Right.

Mati Staniszewski 10:05 Um, where you still get the same voice experience.

Nikhil Kamath 10:08 Mm-hmm.

Mati Staniszewski 10:09 Um, and now there's set of innovations to try to effectively detect your-- without you actually saying the spoken word, detecting your mouth movements. So you could theoretically imagine speaking with a voice agent without vocalizing it out loud and having that support. Um, so like if I were like as a user, so from personal perspective, I would be most likely to adopt the headphones in the future.

Nikhil Kamath 10:31 Mm-hmm.

Mati Staniszewski 10:31 Um, and then I do think it most likely will be the most natural form factor, given it's already one of the, one of the key devices. Um, and that kind of shifts that, that interface, you know, in a different way.

Nikhil Kamath 10:44 So you would say headphones with the phone in your pocket?

Mati Staniszewski 10:46 Yes.

Nikhil Kamath 10:47 A combination of the two.

Mati Staniszewski 10:48 Yes. And probably as some, you know, because you also need the capture of the information.

Nikhil Kamath 10:52 Mm-hmm.

Mati Staniszewski 10:52 That's where that kind of the knowledge part will come, start coming in. I, I do think there will be some version of a, of a pendant or a wristband that you will, that you will carry to just capture more-

Nikhil Kamath 11:02 Can the headphones do that?

Mati Staniszewski 11:03 They could, but then you have, you, you have limited set of information. Like you don't have the same, um, like say over time you will want something about the movement-

Nikhil Kamath 11:13 Mm-hmm

Mati Staniszewski 11:13 ...or something about how you, um, how you capture a different perspective, and I think that's where that will be, that will be valuable.

Nikhil Kamath 11:21 Do you have-- Can you venture a guess on what, uh, Jony Ive and Sam are building in terms of device?

Mati Staniszewski 11:29 Uh, the, um... I heard few, few guesses.

Nikhil Kamath 11:32 Mm.

Mati Staniszewski 11:32 I would love to ask you the same question.

Nikhil Kamath 11:34 Yeah.

Mati Staniszewski 11:34 What's your guess?

Nikhil Kamath 11:34 I asked Sam, actually.

Mati Staniszewski 11:36 What'd he say?

Nikhil Kamath 11:37 He didn't say. [laughs]

Mati Staniszewski 11:37 Okay. [laughs] Yeah. My guess, my guess would be... Huh, I don't know. If you were-- Maybe, maybe, maybe if you were guessing-

Nikhil Kamath 11:47 Right

Mati Staniszewski 11:47 ...from what you know, what's, what's your guess?

Nikhil Kamath 11:49 On Sam and Jony Ive?

Mati Staniszewski 11:50 Yes.

Nikhil Kamath 11:51 I feel like for revenue and to really scale, because human behavior is so accustomed to using a device in your hand, uh, I think that's where scale, scale lies initially. So I wouldn't be surprised if it's a phone-looking device with a UI which is not nativelyAndroid or anything that already exists, but something which is a lot more different with a partner, with a buddy device maybe, which is like a pendant or something like that.

Mati Staniszewski 12:25 Yeah.

Nikhil Kamath 12:25 Because they don't have a hardware play. I feel like for AI, at, uh, some point it'll become relevant that hardware comes pre-installed and with the software, which is AI first, and to be in hardware for all of these AI companies will become increasingly relevant. Hence actually the nothing bet, because I always thought a AI company will come in and, uh, buy them out at some point.

Mati Staniszewski 12:51 They can still... I, it's, it's still possible.

Nikhil Kamath 12:54 Yeah.

Mati Staniszewski 12:54 Uh, uh, it's, it feels... But, but i- in general, like, I think OpenAI has a, has like a good path to building like a personal assistant that's next to you.

Nikhil Kamath 13:03 Mm-hmm.

Mati Staniszewski 13:04 So there will need to be some device that you carry, of course, every, every, every day.

Nikhil Kamath 13:08 Yeah.

Mati Staniszewski 13:08 Um, I do think voice will be a big part for that strategy as well.

Nikhil Kamath 13:11 Mm-hmm.

Mati Staniszewski 13:12 Um, uh, uh, uh, from what I see and hear-

Nikhil Kamath 13:15 Mm-hmm

Mati Staniszewski 13:15 ... they will be ramping out the voice efforts, so-

Nikhil Kamath 13:17 Mm-hmm

Mati Staniszewski 13:17 ... that'll be good, good competition for us to see whether our quality keeps-

Nikhil Kamath 13:20 Yeah

Mati Staniszewski 13:20 ... keeps ahead as it does today.

Nikhil Kamath 13:21 Yeah.

Mati Staniszewski 13:22 So that'll be a good-

Nikhil Kamath 13:22 Yeah

Mati Staniszewski 13:22 ... good test.

Nikhil Kamath 13:23 It must

Nikhil Kamath 13:24 worry you, right? Like, I always wonder about OpenAI. All these people use the backbone of OpenAI, and they build something. And then given enough time, OpenAI sees it, and then they want to build the same thing themselves. They seem to have done that in a couple of places. Um, when I talk to friends of mine in the agentic layer of AI or people who have built great applications on top of a model, uh, they all seem to be scared of what happens to these large language models when they come after the end product. Are you worried about that?

Mati Staniszewski 14:00 You know, like in a, in, in our case, we are of course continuously looking what are, like OpenAI, Google, incredible teams, and probably like as we think about the space, the, the, the biggest potentials of, uh, of, of, of effectively competing with ElevenLabs. As you think about the company, and I only started it in 2022, like the approach from the beginning was doing both the research and the product. So doing the foundational model work for, for audio and then building product around it. And we think both are important, um, and like would define the company where we today of course don't depend on OpenAI for, for, for voice and, and we can stay ahead and build models that are both great in English, great internationally, great across India languages, great across European languages. And, um, and that's both for generating speech with text to speech, understanding speech with speech to text, and, um, and that's important. I think like over the next two, three years, there's still so much you can do in voice, and like we spoke about how you make it more fluid and interactive. I think that still can be cracked in a better way. So of course worries us whether we continue being ahead-

Nikhil Kamath 15:10 Mm

Mati Staniszewski 15:10 ... but it's also good motivation. We have-- my, my co-founder is one of the smartest people.

Nikhil Kamath 15:14 Mm-hmm.

Mati Staniszewski 15:14 He's been able to assemble a, an incredible research team, and, um, and they keep crushing it, so-

Nikhil Kamath 15:19 Mm-hmm

Mati Staniszewski 15:20 ... I think they will, they will, they will continue proving that we can be ahead. But then 100%, you know, we cannot just rely on research. We need to have more, um, both defense points, but also how we deliver value effectively to the customers, so it's kind of independent of the research work. And here, from our perspective, we have kind of two mia- main things. One is, um, we have a creative platform offering and then an agents platform offering. Creative platform is helping with narrations, voiceovers-

Nikhil Kamath 15:48 Mm-hmm

Mati Staniszewski 15:49 ... localization, um, from individuals, so from creators, uh, podcasters, um-

Nikhil Kamath 15:56 Mm-hmm

Mati Staniszewski 15:56 ... news articles, uh, through to some media entertainment companies like Star.

Nikhil Kamath 16:00 Mm-hmm.

Mati Staniszewski 16:00 And then on agents, we help automate voice agents and, and, and, and customer experience, training, education. So, um, Amisha would be a great example.

Nikhil Kamath 16:09 Can you, can you describe to us what ElevenLabs does overall, like in a, in a simplified manner that anybody could understand?

Mati Staniszewski 16:18 We built foundational audio AI technology. So we built models to make speech sound incredible and speech be understood. Um, and then we built products on top of that research that help you create voice agents that can elevate your customer experience, that can, um, elevate, um, a- and, and train people internally or externally. And then we help on the creative side with customers, uh, um, um, um, um, and how they can communicate with, with their user base.

Nikhil Kamath 16:48 So 50% of your business is creator and 50% is enterprise in a way?

Mati Staniszewski 16:52 That's right.

Nikhil Kamath 16:53 And say I'm a creator. How do I use ElevenLabs?

Mati Staniszewski 16:58 The, mm, let's take this podcast as an example.

Nikhil Kamath 17:01 Yeah.

Mati Staniszewski 17:02 The, you know, after, after we, we, we, we, we've finalized our conversation, something might have not been recorded.

Nikhil Kamath 17:09 Mm.

Mati Staniszewski 17:09 You want to add a line.

Nikhil Kamath 17:11 Mm-hmm.

Mati Staniszewski 17:11 You can post-process that with ElevenLabs, make sure that we effectively recreate my voice. We smoothen that line out.

Nikhil Kamath 17:19 Mm-hmm.

Mati Staniszewski 17:20 That's the first simple part. Then let's take a second step. You wanna take this podcast, and you want to localize it outside of English. You wanna bring it to Spanish or French or Hindi.

Nikhil Kamath 17:29 Mm.

Mati Staniszewski 17:29 Um, and here you go to our interface, and you can both automatically dub the entire podcast and then go through any corrections that you need. So when we did a podcast, um, with, between Lex Fridman and Prime Minister Narendra Modi, that was effectively going through that workflow.

Nikhil Kamath 17:47 Did you dub that?

Mati Staniszewski 17:48 Yeah.

Nikhil Kamath 17:48 Mm.

Mati Staniszewski 17:49 And that was like going through then human-in-the-loop, making sure the corrections are there, and that was like a, both like an, a, a, like an incred- very long conversation, incredible test of the, of the software, but then you can enjoy that in, in a number of other languages.

Nikhil Kamath 18:01 How much does it cost? Say, for example, how much did Lex and, uh, our prime minister, how much did it cost to dub?

Mati Staniszewski 18:09 You know, this one was harder. In, in, in general, like if you just wanted to automatically dub, then like, um, an hour podcast will be few dollars, up to a few dollars.

Nikhil Kamath 18:17 Mm-hmm.

Mati Staniszewski 18:18 Um, but with, with any conversation like, like the one with the prime minister, we did employ a wider set of, of people that would actually check every translation. Um-Uh, mostly with, with working with the Lex team and the other side of the team to, to make sure that the content is preserved. Like we, we, we know-- it's like the cale-caliber of the conversation where every word matters. So that was like a, a, maybe a more unique on that side.

Nikhil Kamath 18:42 Like we have this use case. We want to dub all the content we make into like regional languages, but we've been struggling with solutions really. Uh, more than the cost of it, I feel like the emotion behind what is being said is always lost and everything sounds a bit robotic. How do you change that? I've not tried ElevenLabs.

Mati Staniszewski 19:07 Then perfect.

Nikhil Kamath 19:08 Yeah. [laughs]

Mati Staniszewski 19:08 It's like almost a cue to, [laughs] to, to, to, to show you. No, I think, I think that's like one of the big parts that I think we, we now can bring to the, to the equation that was lost. Like, how do you both preserve the voice, preserve intonation, preserve em-emotion. Um, and the way, you know, we do it today is we, we of course detect your speech. We then, uh, detect the speakers. We, we translate it, but then the moment we generate a speech, it actually captures the contextual information from the previous response and, um, and tries to recreate that in a, in a new target language. Like effectively our model, the way we've-- we created that for the localization use case, has been trying to match pairs of languages that, that, that we have and how would you deliver that in a, in a professional setting and how you can reapply that in the, in the automatic dub. Um, so I think it works really well. Um, the, the, the hard things is like, you know, of course, when you have like amplitude of emotions like-

Nikhil Kamath 20:04 Mm-hmm

Mati Staniszewski 20:04 ... laughter, scream, and you then shift the, uh, ordering of the sentence, then it's hard-

Nikhil Kamath 20:08 Like say-

Mati Staniszewski 20:09 But I think it works

Nikhil Kamath 20:09 ... say for example Mati is smiling while you're saying a

Nikhil Kamath 20:12 certain thing. Mati is the right way to say it?

Mati Staniszewski 20:14 Yeah, yeah.

Nikhil Kamath 20:14 Yeah.

Mati Staniszewski 20:14 Yeah, yeah.

Nikhil Kamath 20:15 Uh, when you smile when you say a certain thing in English, isn't it hard to take that emotion into Hindi, for example?

Mati Staniszewski 20:24 The, um, uh, no, it's, um, you know, it's, it's not easy 100%.

Nikhil Kamath 20:30 Mm-hmm.

Mati Staniszewski 20:31 Um, and then there's like kind of two of-- like there's the present and the future. In the current way, you would still translate it. You would of course dub it. I think you can preserve the emotion, but of course my movement of my lips-

Nikhil Kamath 20:43 Yeah

Mati Staniszewski 20:43 ... when I smile will... You-- That, depending on the language structure, um, it might be in a different part of the sentence. So like maybe an easy example is German.

Nikhil Kamath 20:52 Yeah.

Mati Staniszewski 20:53 From English to German, German usually will have noun at the end. So like yes, that will exactly happen. You will like smile at a different part than-

Nikhil Kamath 20:59 Mm-hmm

Mati Staniszewski 20:59 ... than the conversation. You can do some tricks to, to try to, to, to, to, to, um, to still smoothen it out. Um, but yes, there will be some discrepancy. So that's, that's in present can still work. In the future, you know, and, and there are some limitations of the current delivery softwares. You'll likely do a lip animation too or lip reanimation to move the lips a tiny bit to apply to where, where you would smile in that part of the sentence. So I think that will happen 100%. I hope this happens this year.

Nikhil Kamath 21:30 Is that something you would do?

Mati Staniszewski 21:32 We would want to do it. I think like as we think about anything that can amplify the audio experience, we want to bring that into our, our, our, our flow. So on that creative side, like today if you were to localize, hopefully in the next few months you will be able to, um, animate lips straight away as well.

Nikhil Kamath 21:48 Could a tiny company be built just around this? Just around changing how the lip moves when something is being dubbed?

Mati Staniszewski 21:58 It could. There's, there's actually good set of companies that are like crushing it and going very quickly.

Nikhil Kamath 22:03 Right.

Mati Staniszewski 22:03 You know, of course, I think this is an entry point, not the end point.

Nikhil Kamath 22:06 Mm-hmm.

Mati Staniszewski 22:06 They will likely try to, to add more of that functionality but, but, uh, but 100%. You know, it's like... And the, the, the-- of course, now we speak about the static part.

Nikhil Kamath 22:16 Mm-hmm.

Mati Staniszewski 22:16 But the moment you start moving into the real-time part, those tiny companies now can be huge companies. And that's kinda like, you know, we spoke about the creative side. If you think about the, the, the agent side, and I think there are some use cases here for, for, for, for both the lip animation and even for, for, for your-- for you-

Nikhil Kamath 22:35 Mm-hmm

Mati Staniszewski 22:35 ... um, uh, uh, I have a, I have one idea for you. Um, so you know, we spoke about the creative side. On the agent side, effectively what we do is, is, is allow you to create a voice experience, bring any knowledge that, that, that you want inside of that experience, um, and, um, and then deploy that in, in, in different settings, um, while monitoring, evaluating how it performs. I don't know if you're familiar with Masterclass.

Nikhil Kamath 23:01 Yeah.

Mati Staniszewski 23:01 So Masterclass works with us where they would recreate, um, and work with people like Gordon Ramsay-

Nikhil Kamath 23:07 Mm-hmm

Mati Staniszewski 23:08 ... to create AI Gordon Ramsay to teaches you how to cook while you cook.

Nikhil Kamath 23:11 Mm-hmm.

Mati Staniszewski 23:11 Uh, the top FBI negotiat-negotiator, Chris Voss.

Nikhil Kamath 23:15 Mm-hmm.

Mati Staniszewski 23:15 Is part of the ecosystem. You can negotiate and learn negotiating by negotiating with Chris Voss. So you could do your masterclass based on a lot of your existing content, ex-existing work that you already did and, and, and, and spoken about, worried about, and actually create a, um, a Nikhil that you can speak with-

Nikhil Kamath 23:33 Yeah

Mati Staniszewski 23:33 ... and learn about bootstrapping a three billion dollar company-

Nikhil Kamath 23:36 [laughs]

Mati Staniszewski 23:36 ... uh, and, uh, and, and how to achieve that kind of impact and scale and break boundaries.

Nikhil Kamath 23:41 And where would that agent sit? How would somebody find him?

Mati Staniszewski 23:45 You know, we work with all types of customers, um, but if you are an individual, you probably would start your own company.

Nikhil Kamath 23:52 Mm-hmm.

Mati Staniszewski 23:52 Like we actually have five hundred companies that build voice, uh, voice agent variations just in India.

Nikhil Kamath 23:57 Mm.

Mati Staniszewski 23:58 And then, uh, now I think over ten thousand across, across the world that, that usually they would create their own website-

Nikhil Kamath 24:04 Right

Mati Staniszewski 24:04 ... or web delivery. But the, the, the most common now that's kind of shifting is, um, is you, you have a phone number that you call and then the voice agent answers. So, um-

Nikhil Kamath 24:14 Do you think there's a market for that? If there's an agentic version of myself which has learned from everything I have ever said, all the stupid things, and it speaks like me, do you think somebody will seek it out to interact with it?

Mati Staniszewski 24:31 I think it, so quick answer is yes.

Nikhil Kamath 24:33 Mm-hmm.

Mati Staniszewski 24:34 Um, I don't think it's, you know, like it replaces the experience, but there's, um, there's like how do you go deeply into the subject that interests you, and that's what your agent would allow. Like, I have a few questions. This is what I know. This is what I don't know. You've, you have that kind of wider set of knowledge that you del- delivered over the years. How do you deliver that efficiently? You can, you can, you can grab to the specific parts that I'm interested in, and, um, and I think that would be too would be valuable. But then, of course, at some point, there's like, okay, now I actually want to speak with a real person to go deeper-

Nikhil Kamath 25:08 Mm-hmm

Mati Staniszewski 25:08 ... because that wasn't in a knowledge base or wasn't, um, it wasn't something that, that I could ask about. So I think it kind of shifts the part of, um, of the conversation, but not the entirety of the conversation. You know, it's... I don't know if it's exactly the same, but, um, on the business side, we, we, one of the, the, uh, uh, like many startups mention this use case. I don't think it's, like, fully AI automatable, but AI SDR version of, like, learning about the product, learning about the use case, learning about the, the company. Um, and we've seen this, um, play out in, in a few cases where here too, the best way to deploy AI SDRs is, is more as a augmentation of the flow-

Nikhil Kamath 25:50 Mm-hmm

Mati Staniszewski 25:50 ... where you can, let's say you, you have a website. Um, you want to capture the sales leads. Um, the sales leads come through. They usually will leave their details in the form, and now instead of replacing the whole experience, you can add effectively a voice agent saying, "Hey, do you want to accelerate your application, um, uh, by, by speaking with a voice agent?" And then you can, you can go ahead and, and, and click through. Uh, and then you are more willing and, and, and, and, so, so we've done exactly that flow for ElevenLabs.

Nikhil Kamath 26:25 Mm-hmm.

Mati Staniszewski 26:26 And then what you know is that both it improves the number of people that are keen to leave things in a form, but second, they actually leave a lot more details, and then the moment they go to their next conversation, they know much more about the product. They are more ready. We can go after, um, and, and with those customers much quicker. Um, so similarly as you think about your kind of delivery of knowledge, I do think this would both make people more excited about, about, about learning and going deeper. But then the second time you have the conversation, I think the, the quality of questions will be better. We actually did it with, um, a company in India called TVS Motor.

Nikhil Kamath 27:00 Yeah.

Mati Staniszewski 27:00 Um, and they, they've done exactly that flow.

Nikhil Kamath 27:03 Mm-hmm.

Mati Staniszewski 27:03 And it improved both sides. It improved the customer satisfaction and improved the sales, uh, uh, lead qualification. So it's like both sides were happy.

Nikhil Kamath 27:11 Right. We were doing a bit of research on the market size of voice overall. The dubbing is a small market, right? It said less than a billion going up to three billion in the next decade.

Mati Staniszewski 27:24 You know, dubbing is, is like as, uh, mmm, the interactive use case is the biggest market.

Nikhil Kamath 27:31 Right.

Mati Staniszewski 27:31 I mean, that's everywhere from, you know, from media education through, through the enterprise.

Nikhil Kamath 27:35 So that, that could be a really big market in the next five years.

Mati Staniszewski 27:37 That's... Yes. That-- I think that will

Mati Staniszewski 27:39 shift everything.

Nikhil Kamath 27:39 So a lot of the audience that we speak to are young wannabe entrepreneurs. What do you think the opportunity is? What can one build in voice that can be a big profitable business tomorrow?

Mati Staniszewski 27:54 From my perspective, the, the most clear entry point is combining a domain expertise in one of the, one of the domains that you, you know well or you want to know well and learn quickly. Um, and it can be a domain where, where there's like no innovation, and these are like almost the best ones. And people-

Nikhil Kamath 28:11 Can you give me an example?

Mati Staniszewski 28:13 Automotive industry, healthcare. I mean, it's like truly, I think it's like mmm, the more traditional domain, the likely more lagging it is w-with the innovation, um, where we are seeing already entrepreneurs building for automotive industry of use cases. Um, but effectively here, you know, whether it's, um, uh, how... It's like f-f-from the simplest case of how you innovate on customer experience to like what is the in-car experience look like in the future? So much opportunity to figure out how do you deliver the right knowledge, how do you write the right interaction, and how you deploy it.

Nikhil Kamath 28:48 But can I build a company just around that?

Mati Staniszewski 28:50 Yes. Yes.

Nikhil Kamath 28:52 So-

Mati Staniszewski 28:53 And it's like, you know, like-

Nikhil Kamath 28:53 Let's take the example of a car company, say BMW. What, what do I build that they will pay me enough for that I can have a company in voice?

Mati Staniszewski 29:04 You would start probably with something that's top of mind and, and something that is easy to deploy. So here you would start with customer experience side. So just being able to help as people call in, how do they get the best service, understanding everything about BMW in this case.

Nikhil Kamath 29:21 Mm-hmm.

Mati Staniszewski 29:21 How they understand about how the car works. You, you spend a lot of time on, um, bringing that into the equation. So that's kind of the first layer. Then in-car experience is like pretty open-ended, but how do you create an interactive radio inside of the car that people can, can interact with-

Nikhil Kamath 29:39 Mm-hmm

Mati Staniszewski 29:39 ... um, across all the time? And here too, you need to figure out what type of languages might BMW might be interested in. How do you create the right voices to make sure the users in those regions are excited and happy? Um, what type of experience will BMW actually need? So there's a lot of actually deployment work that would come in. And then as you build all those steps, you need to figure out, okay, if you are to build like a, a big company, not only how I build it once for BMW, but how I abstract it away to, to be able to replicate that into the other car companies. Um, so you need to capture some of those knowledge elements back into, into, into other, uh, into other car, uh, parts. Automotive industry is a, is a good example where I like haven't seen as much yet because it's, it's hard with some of the on-device requirements. Like, some that already are getting disrupted are e-commerce, healthcare, financial services. Like, healthcare is a easy example where you just need so much specific domain knowledge to be able to, to cater to that. Both the data is very different-So capturing all the, all the medicine names, uh, all, all the flows that might need to happen for the patients that you interact with-

Nikhil Kamath 30:49 Mm-hmm

Mati Staniszewski 30:49 ... to make sure that they get the right experience, that's like, it's hard.

Nikhil Kamath 30:54 Mm-hmm.

Mati Staniszewski 30:54 And then you need to go to the hospitals, speak with the hospitals on how the, the to deploy that, whether it's in the operating space or whether it's in the, in the, in the, in the, in the support space outside of the, uh, in-hospital operation. So that's, that's the one that's act- already pretty, pretty heavily going across. Um, e-commerce, I mean, recently you had IPO of Meesho that-

Nikhil Kamath 31:15 Yeah

Mati Staniszewski 31:15 ... that you work with, incredible company.

Nikhil Kamath 31:17 What do you do for Meesho?

Mati Staniszewski 31:19 So we help automate their customer, customer experience, customer support.

Nikhil Kamath 31:22 Mm.

Mati Staniszewski 31:22 So they have now 60,000 calls coming in, um, to, to, to ask about the products-

Nikhil Kamath 31:29 Mm

Mati Staniszewski 31:29 ... ask about shipping, refund, and in Hindi and in English, and, and we help, we create a voice agent that you can interact with. But a very cool thing with Meesho is that-

Nikhil Kamath 31:38 Mm

Mati Staniszewski 31:38 ... now, you know, so the support is there, like, what's the future of e-commerce? Could you have an AI concierge the moment you go into the website that you open a website, you have a voice agent that tells you about the products available, you tell it what you're looking for, then shows you different products, you decide which ones you are interested in, and then you order that. Like, having an assistant as part of experience would be, would be incredible. Um, so here too, like I think in e-commerce, going slightly deeper on like s- s- what type of e-commerce, similar. I think, I think companies combining the domain knowledge and, and, and building quickly the, the voice agents will, will, will be huge.

Nikhil Kamath 32:16 Actually, you should help us build a agent of our own. Could be fun as a project.

Mati Staniszewski 32:22 What would be the agent?

Nikhil Kamath 32:24 Like we'll feed in all the things we have said up until now and have somebody talk to people who are lonely and just see how the project goes. [laughs]

Mati Staniszewski 32:33 [laughs]

Nikhil Kamath 32:35 Just as a fun project. I speak to a AI agent all the time. I think the, a big use case in the world tomorrow will be to solve for loneliness, you know, for intimacy.

Mati Staniszewski 32:46 It's, I like, I think voice a- I agree. I think that'll be a big-

Nikhil Kamath 32:50 Yeah

Mati Staniszewski 32:50 ... part there. It's going to be a big balance of figuring out how it's like not, uh, you know, you, you, you, you kind of help with the virtual side, but like kind of try to bring people into the real world too, which is, which will be at like an important balance. Like the one, the most exciting one for me is education, like how will the education... Like could you have a personal tutor across subjects for everyone? Like I love Richard Feynman, could I have Richard Feynman teach me physics? And, uh, and, um, uh, and like kind of both the style of delivery is so cool, like he's kind of combining the knowledge-

Nikhil Kamath 33:26 Mm-hmm

Mati Staniszewski 33:26 ... and the, and the laughter, and then I have it up- I'm, I'm learning maths and I'm, I'm, I can ask any question to, to, to Richard. That'd be, that'd be amazing.

Nikhil Kamath 33:35 What's even more interesting is if I have a device in my ear which negates the need for me to have to regurgitate any kind of information, what happens to education? Then I don't need to learn or remember much of what traditional education needs for me to remember.

Mati Staniszewski 33:57 Yeah, you're almost like... I don't know how, um, how, how you would see this, but like the way I think back at, at, at, at kind of my education, like I kind of, I, I rarely use any of the formal backgrounds in mathematics. I almost never use that knowledge. But like the, the most valuable part was kind of learning how to learn, so I hope there's like still that component that education brings. And then of course the second part of education, it's going to be like almost the opposite of tech, which is how do you learn how to, how to be with other people, how you like d- be social. So like what you have, like I feel in the future of education you'll have like percentage of the time which is like fully dedicated to learning with AI and tools, and then percentage of the time where there's like no tech and no, no, um, no, um, no technology, just like speaking with other people, being close to them, figuring out that part. The percentage of both will be interesting. Like with AI, can I now be more efficient, so I just need like 20, 30% of the, on the former and 70% on the latter?

Nikhil Kamath 34:57 Mm.

Mati Staniszewski 34:57 Uh, or, or, or, or no, versus, yeah, versus like a fully augmented version throughout.

Nikhil Kamath 35:02 As somebody not in the Valley, uh, my day job is a investor. What do you think of...

Nikhil Kamath 35:10 You've raised a lot of money yourself, I think like $300, $400 million. Your company's worth a few billions today. What do you think of AI valuations in the Valley today? These huge, uh, companies with a revenue number that doesn't historically warrant the market cap numbers that they have. Do you feel like there's a bubble brewing there?

Mati Staniszewski 35:40 And did you say you, you invest in the companies not in Silicon Valley, or you said, did you invest in companies in Silicon Valley?

Nikhil Kamath 35:46 Some in Silicon Valley, but largely companies out of, uh, India.

Mati Staniszewski 35:51 That's what I thought, and, and first of all, I think that's great. I-

Nikhil Kamath 35:53 Mm

Mati Staniszewski 35:53 ... it's like, uh, similarly as we think, uh, like it's so cool to see more companies in, in, in India. Like we care so much about companies brewing, and I'm from Poland, in Poland-

Nikhil Kamath 36:08 Mm

Mati Staniszewski 36:08 ... in, in, in UK of course. Um, so s- and, and like as we think about our team today, most of the recent engineering is from Europe, and they are like some of the most motivated, passionate people about the space. So like hopefully a lot more of that innovation comes across f- from, from Asia, from Europe, internationally for the global companies. Um, I think the, the valuations in Silicon Valley, you know, like I don't think the, the, the bubble is brewing.

Nikhil Kamath 36:35 Mm.

Mati Staniszewski 36:35 I think there's like... What, what I think is a very different part as I compare it to, to, to other cycles, I think there's a clear value, clear revenue numbers for a lot of the, the companies.

Nikhil Kamath 36:44 Mm.

Mati Staniszewski 36:45 I do-- You know, there are some pockets where I'm, like, a little bit skeptic of, like, does it make sense, you know, not to just... Like to maybe give a concrete example is all the, like, GPU providers, um, that are built on top of Nvidia that, like, help you get, uh, e-effectively inference, um, uh, um, capacity.

Nikhil Kamath 37:05 Mm-hmm.

Mati Staniszewski 37:06 Like, I'm a little bit skeptic on some evaluations there, but, you know, like, for a lot of others, it's very clear it's, like, high risk, high reward type thing. Um, so I understand the bets-

Nikhil Kamath 37:15 Mm-hmm.

Mati Staniszewski 37:15 But they are probably higher risk than higher, higher reward. I do think AI is, is in this, in this, in this trickier space where, um, where if you win, you can win in a, in a significant way, in a significant impactful way. Like, you will work with a lot of the customers around the world if you are a good company. Um, so I kind of understand why, like, in the early stages of the companies, companies get such a high valuation. I mean, y- I, uh, y- I'm sure you are referring also to a lot of the research companies being formed. Here, here it's like almost the way I kind of s-see it is there's like, um, like, research is still so incredibly important and will be that, um, that, that-- like, that multiple still makes sense. Like, you, you-- those people, like, if they invent one incredible thing, it, it's already worth it. Um, well, but I hope, I hope companies around the world will get the same valuations, uh, soon. And if we-- It's like if they build globally, they should.

Nikhil Kamath 38:09 Yeah. The way I'm looking at it with all the geopolitical stuff going on, if I were to wager a guess, I would think at some point in the future, we will all live in a more multipolar world where we-- individually countries go down the path of Russia and China, where most large countries have their versions, their own little version of these products which have global reach today. So different countries will have their own social media. They'll have their own voice companies. They'll have their own, uh, uh, most interactions with technology, even their own data centers. A country's data, I think, will sit in that country. So I feel like when that happens in the world, these companies which have large scale today, like, you know, take a WhatsApp, which is a chatting service. I don't think it is a unique enough of a product that can't be disrupted, or I will make the switch from a WhatsApp to another chatting service given a very small incentive. I think the, the reasons for Indians to use WhatsApp might go away, uh, when the world becomes more geopolitically confused, and I feel like it's going in that direction.

Mati Staniszewski 39:29 Don't you think, like, let's take WhatsApp, there's, like, a network effect of just so many users already using WhatsApp where it just, it's just convenience of being on the same platform?

Nikhil Kamath 39:39 I would have thought that, but when I look at what's happening in China and Russia, and they have their own versions, it seems to be just fine. Uh, and also in a way, when there is evidence of these global tools being weaponized by governments in one way or another, uh, I think the world will be a lot more skeptical in using a product just for the network benefit. You know what I mean?

Mati Staniszewski 40:06 I, I do. I know what you mean, although I do, I do disagree. I think it's all-- partly disagree.

Nikhil Kamath 40:12 Mm-hmm.

Mati Staniszewski 40:12 I, like, I agree. Data residency, yes, hundred percent should be in a, in a specific countries. Also agree that, like, the, the, the, the, sorry, the product shape, uh, will, will, like, kind of evolve. So, like, every country will use their own, own propriety data to create their variations of the product.

Nikhil Kamath 40:34 Mm-hmm.

Mati Staniszewski 40:34 But my, like, you know... I don't know if that's, uh, op-optimist or, or, like, the, the, the kind of the value part, but I do think there's, like, a global component of, of a lot of infrastructure or software still being international. Like, let's take building foundational AI models. I think the, the, kind of the, the base will be pre-trained by-

Nikhil Kamath 40:55 Mm-hmm

Mati Staniszewski 40:55 ... few companies in the world, just given the resources required. There'll be open source, um, um, um, um, of course, um, offerings, uh, in that, in that bucket. But then probably must have been like a government space or companies in a lot of the sectors, they will use those to fine-tune it based on their data, and it's still huge advantages.

Nikhil Kamath 41:13 Mm-hmm.

Mati Staniszewski 41:13 Uh, but just, uh, just not, like, entirely built into independently. Um, on the product, there's a very, like, I mean, interesting, interesting that you think... I, I can see, like, how creating software is becoming so easy that you could, like, imagine creating that on the, on the fly everywhere. I just, I just think the network effect of having people on the same place-

Nikhil Kamath 41:33 Mm-hmm

Mati Staniszewski 41:34 ... is compounding in many ways.

Nikhil Kamath 41:35 Mm-hmm.

Mati Staniszewski 41:36 Um, like, almost like, as, you know, like, as I think about, um... It's not the same, but in, in our case, we have a voice marketplace in ElevenLabs-

Nikhil Kamath 41:44 Mm-hmm

Mati Staniszewski 41:44 ... where people create a voice. They can sh- We authenticate it, then we can share it, and then the moment the, the AI voice is used, you earn money. Now we have ten thousand voices created this way, um, and we paid back over eleven million dollars, coincidentally, back to that community. Um, and it definitely helps. They have, like, so much styles, accents, and it's, uh, you know, of course, you could replicate at scale, but you won't have that same community mechanism of paying back and making sure people can trust that software. So I do think there's some element of trust, um, uh, uh, uh, that, that is embedded. So maybe network and trust. But you know, I guess what you are saying partly is with what happened is-- with the geopolitics side is, like, the trust element may be broken, and you kind of need the alternative for-

Nikhil Kamath 42:28 It seems to be breaking,

Nikhil Kamath 42:29 right?

Mati Staniszewski 42:30 I hope, I hope it's, you know, it's a temporary glitch-

Nikhil Kamath 42:32 Mm-hmm

Mati Staniszewski 42:32 ... but it's, like, more, like, it's not, um, it's not going down. It's more like sine X times X curve-

Nikhil Kamath 42:37 [laughs]

Mati Staniszewski 42:37 ... when it's continuing going up and-

Nikhil Kamath 42:39 Yeah

Mati Staniszewski 42:39 ... and, and, and it's, yeah, it's just-

Nikhil Kamath 42:41 Yeah

Mati Staniszewski 42:41 ... like a little bit of a correction.

Nikhil Kamath 42:43 Yeah. I mean, I watched a bunch of these people speak over the last week, and, you know, even when they walk out of the room, it could be the European leadership or the American leadershipPeople don't look happy. People look like they're angrier than they were last year, for example-

Mati Staniszewski 43:02 [chuckles]

Nikhil Kamath 43:02 ... or the year before. Every year people are progressively angrier, and it seems like they're voicing it for the first time. You know, people walk out and speak in their own languages. Most of it you can't understand, but you can see the expression on their face in a way, I think.

Mati Staniszewski 43:18 I do think people became more authentic in the like, like last year or two-

Nikhil Kamath 43:21 Yeah

Mati Staniszewski 43:21 ... last year or two. Maybe it's like in a version of like, there's just so much AI content-

Nikhil Kamath 43:26 Yeah

Mati Staniszewski 43:26 ... and, and things that like-

Nikhil Kamath 43:27 Yeah

Mati Staniszewski 43:27 ... you need to be authentic to, to, to, to like break through that noise of AI.

Nikhil Kamath 43:32 Yeah.

Mati Staniszewski 43:32 But, um, I feel, you know, in, in less on the politics side or geopolitics side, because I agree-

Nikhil Kamath 43:38 Mm-hmm

Mati Staniszewski 43:38 ... that's like that has definitely w- influenced the emotions. Um, and I'm probably a little bit too much in that bubble of, of AI and tech-

Nikhil Kamath 43:45 Mm-hmm

Mati Staniszewski 43:45 ... but huge, huge positivity compared to last year or two of like shift from how bad, how much of the risk it pose, po-po, like f- um, how much of the risk AI can mean, but increasingly into also how much opportunity AI can, can give. Um, but of course, it's such a, such a tricky week of geopolitics this week, so.

Nikhil Kamath 44:06 I think ev- no one's contesting that productivity will go up in societies. I think everybody's aligned to that. Uh, the only question is should you be... People with most risk capital will have the best models by virtue of how much money they pump into those models. But if these large model companies are going to cannibalize the end product if you build on top of them, I think for a young entrepreneur, it's important not to be short-sighted and kinda like diversify and hedge what they build on top of. Uh, not, it's not a, it's not the right analogy, but not end up in a world where an Android or an Apple can charge you twenty percent fee to live on that platform, but use more open-sourced versions and diversify that risk, I think is important today.

Mati Staniszewski 45:06 I agree that like, you know, you need to take a like a longer term view, and f- it's like the, the, the starting point can prevent you from being in that, in that, in that long term. So, you know, if you, if you, if you can accelerate your value delivery by building on any of the platforms, it's fine. But if, if there, if you think, you know, as an entrepreneur, if you think that any of those platforms might take over your space in next few years, um, then yeah, then you need to be, you need to be cautious.

Nikhil Kamath 45:35 Because these platforms have to make money somewhere tomorrow. I saw OpenAI is experimenting with ads for the first time, but to justify those valuations and to stop burning money, they have to make money.

Mati Staniszewski 45:47 They do. It's like almost the, you know, the, um, like Google will have so much advantage by-

Nikhil Kamath 45:54 Mm-hmm

Mati Staniszewski 45:54 ... potentially running longer with all the infrastructure and not doing ads in the same way.

Nikhil Kamath 45:59 Yeah.

Mati Staniszewski 45:59 Um, so we'll be, we'll be, yeah.

Nikhil Kamath 46:02 [chuckles]

Mati Staniszewski 46:02 I'd say it'll be an interesting one. But I, I like, you know, like, uh, yes, your company's built on top of LLMs voice visual. I think that the, the domain, domain combination, like if you know the domain extremely well-

Nikhil Kamath 46:13 Mm-hmm

Mati Staniszewski 46:13 ... then the AI model is, you know, gives you a good starting point, but then integrations with the system, the legacy systems of how you deliver the telephony, um, the, uh, how you bring the knowledge, that will take so much time. Then like building the trust with evaluations, monitoring, that will take a lot of time. Um, so I think there's like so much opportunity for entrepreneurs to just take any of those models-

Nikhil Kamath 46:35 Mm-hmm

Mati Staniszewski 46:35 ... go deeply into the domain and combine both, best of both.

Nikhil Kamath 46:39 We've been flirting with the idea of building a social media platform in India. How do you think-

Mati Staniszewski 46:44 Under Zerodha or separately?

Nikhil Kamath 46:45 Separately. Uh, Zerodha is more like fintech-

Mati Staniszewski 46:49 Exactly

Nikhil Kamath 46:49 ... asset management, broking, uh-

Mati Staniszewski 46:51 But you could do like a super app effectively as you know. It's, uh...

Nikhil Kamath 46:55 You know, super apps have never worked in India. A lot of people have tried it. Actually, outside of China, they've not worked anywhere, right? Really.

Mati Staniszewski 47:03 Yeah. I, I'd say I'm wondering whether in, uh, yeah, I don't-

Nikhil Kamath 47:06 Do you think that has something to do with-

Mati Staniszewski 47:08 You know, I have an example where it did work, which is like-

Nikhil Kamath 47:10 Oh

Mati Staniszewski 47:10 ... uh, an interesting one. Uh, but yeah, the, the, um, in Ukraine-

Nikhil Kamath 47:15 Mm-hmm

Mati Staniszewski 47:15 ... they have a super app, which I think is brilliant. Um, uh, uh, also we have a pleasure of working with, with their government, uh, and where they are trying to create the first agent in government, but the, the, the they have a Diia app.

Nikhil Kamath 47:27 Mm-hmm.

Mati Staniszewski 47:27 It's effectively a citizen support app where you can ask about the, the benefits programs, employments program, the, the, the different, different, the, um, things that are happening in India. Um, and that's how it started. So like effectively an app that every citizen can have. Uh, of course, the passport information, um, um, visa information, et cetera, that you can travel and ID yourself with. But then they have multiple, um, ministries and the way they operate. Um, so they have multiple ministries where you have tech resources embedded in those ministries. So Foreign Affairs will have a few engineers working there.

Nikhil Kamath 48:07 Mm-hmm.

Mati Staniszewski 48:07 Internal Affairs will have that. Education Ministry will have that. And now those people try to figure out how to automate part of, of, of, of the government work and then, then bring it back to the digital transformation ministry-

Nikhil Kamath 48:21 Mm-hmm

Mati Staniszewski 48:22 ... um, which is responsible for the citizen support app. And now what has started happening is they've started effectively adding modules-

Nikhil Kamath 48:28 Mm-hmm

Mati Staniszewski 48:28 ... across the entire government, so now you can have healthcare information. Now you can have education where you have personalized content for you based on things that you know. And, and I thought it was brilliant. It's-- and what we do with them is now help bring the voice into the equation so you can communicate and call a lot of that, or you could learn by speaking. Uh, but you literally have the all the modules deployed across the government services. And I, I thought it was brilliant for two reasons. One is like that you embed those resources, like in a AI world, it feels so right. That's, that's, that's like as I think about young entrepreneurs building their companies, like you want, of course, engineering talent, but you want the engineering talent in, in sales and go to market and, and, and, and, and, and, and finance and, and like all divisions because it can help get smarter. Um-So I liked it for that. And two, despite everything that's happening there, they are actually con- truly from everybody worked of leading the way with, like, the one of the best apps that, that I've seen.

Nikhil Kamath 49:24 In places like India, uh, there's the David and Goliath problem. When you're David, everybody loves you when you're starting something. When you become a Goliath of a company, inherently people dislike large corporations in a, in a strange way.

Mati Staniszewski 49:41 But that goes everywhere a little bit.

Nikhil Kamath 49:43 Yeah, yeah. Even in America, I think. There's a little bit of anti-incumbency and a rebel inside everyone, and hence I don't think a super app works in

Nikhil Kamath 49:51 places like that.

Mati Staniszewski 49:52 Why did you want to start a social media?

Nikhil Kamath 49:54 Uh, I believe social media is broken today. Uh, we were looking at data from a bunch of people that the really young people, sixteen to twenty, uh, take Snapchat, for example. They might be spending-- They might be logging into Snapchat the same number of times, but the amount of time they're spending on it is significantly lower. If you take Instagram, it's become almost no organic content. Like Mati will not post a picture of his, you know, I don't know, you having a drink on a weekend, nor will I or nor will anybody.

Mati Staniszewski 50:29 Yeah.

Nikhil Kamath 50:29 None of my friends. It seems to be a place where brands are showcasing different things and influencers are trying to sell one thing or the other, either consciously or subconsciously. Uh, to have an app which is not governed by an algorithm which kinda insinuates and incentivizes a reaction in you by triggering, uh, hate or, uh, uh, emotion that I don't think is conducive in the long term, I think is useful. Also, I don't like the fact that a foreign nation controls the algorithm that defines the mood, nature, and evolution of the youth of my country.

Mati Staniszewski 51:19 And in the incentive side, do you have a take of how, like, how you change the incentives to get people to post the organic side?

Nikhil Kamath 51:26 Uh, I think we are thinking of a very, like, very tight funnel on the top where we control who enters the app. Like incumbent social media probably has bots engaging with you and trolls engaging with you, and I don't know how much of it is organic and how much of it is not. But the idea is to have a place where real people can connect with real people, speak about real things without-- I don't think anybody can take out bias altogether, but with significantly lower b-bias, and a place which is safe to have discourse. You know, I, I might agree with you on something, I might disagree with you on something, but I don't think there is a place or ecosystem today where you can have that conversation, so we hope to kinda create it. But I'd love, uh, you know, advice from you. Broad thoughts on what should a new social media have. Also, how voice plays a part in whatever the newer version of social media is.

Mati Staniszewski 52:33 Uh, first of all, I think you're, you're not, like, the aspiration or the idea, like, I mean, as I, I agree with. Uh, like, I, I hope even the current social media platforms will get there too. Like you, you know, like, I think there's a clear aspiration for them to have, like, a little bit more of the positive sentiment rather than kind of the, the, the, the, the fear driving the, the, the, the algorithm. And I, I, I don't know if, you know, like, I don't know how this will get figured out yet, but there's definitely this element of, like, authenticity being now the, the way to go. Like, nobody wants to read, like, a corporate, you know, like manicured press release statements, and I think that's, that's great. So if I were designing social media, it's like how do I get more of, of, of that, uh, part would be, uh, like one of the key elements. Um, I'm thinking through, through both the, the kind of the voice aspect and, um, um... Do you know AirChat, by the way? Do you see-- Have you seen AirChat?

Nikhil Kamath 53:28 I have seen AirChat, yes.

Mati Staniszewski 53:30 They're trying-

Nikhil Kamath 53:30 Right

Mati Staniszewski 53:30 ... to build like a social media based on voice and podcasts.

Nikhil Kamath 53:33 Mm-hmm. We do have that moat that we already have. Like, I think our podcast is very, like, entrepreneur heavy. We try and tell people how to, A, start a company, and then we try and help them on the path of starting a company. We have a grant, we give them some money, we try and build a network for them to go out and build. Uh, so it can be pivoted around a podcast not just of mine, but of a bunch of people who have the same outcome in mind.

Mati Staniszewski 54:05 Mm-hmm. So maybe, you know, like kind of starting the kind of opposite way on the, on the voice side first, like I could imagine the whole interface very, being very different where you, you kind of f- the same way you would imagine Alexa or Siri delivering you the knowledge or some version of, of Jarvis where, like, you wake up, you want to learn what happens on the social media and, and it kind of summarizes things that are interesting to you, and you can ask questions, go deeper. Like for me, that would be like part one, where social media comes with like embedded assistant that kind of k-keeps you up to date on what's happening, and you can engage with that. Where, where I could like, "Oh, leave a comment here," or, "Hey, engage with this part," or, "Respond to this person." And you could, you could, like, imagine both responses both being text or, or voice directly. So that, that would be like a first, like a very shift into like the static timeline view into more of like interactive companion view. Um, so that would be one component of this. Two, yeah, I'm, uh, I'm kind of, I'm hesitating because I'm thinking like how would I incentivize the people? And that's the tricky part. But like I do think there, you know, if you, uh, if you record a video message from you, like that's, that's great. Second to that, you record audio message, image, then text. So if you could haveA, a, a voice as part of that would be great. Simple version of that on social media could be like, you know, you write a, you write a post or you write a, an essay, and I can listen to that in your voice and, and, and, and enjoy it. But that's, that's kind of part of how can I leave the messages and comment a-across, uh, would be another, another big one. Three, which is both on language level but on the voice level, making it straight international, translated and, and then, and then enjoyed across languages would be, would be another one. Um, but look, I think it's a very tricky, tricky one, um, in terms of like the incentive side. Like how do you shift that? Uh, you, you know, you're probably more familiar given, given, given your work of like, uh, tokenize based or, or like incentive, monetary incentive based approaches. I'm not... I-It's like I, I can see this working, but I'm not a big fan. I don't like this like kind of transactional approach. Like it should not be that. Ideally, you like engage because you're interested rather than because y-you think it can give you like the monetary award.

Nikhil Kamath 56:20 Mm-hmm.

Mati Staniszewski 56:20 Like it should be a secondary versus primary effect. Um, I think, you know, like the, the... what, what Sean to be working... I, I think the verification side, I would keep the verification side.

Nikhil Kamath 56:30 Mm-hmm.

Mati Staniszewski 56:30 I think it's valuable for people to know that the, the other side is, is a, a, a real person.

Nikhil Kamath 56:35 Mm-hmm.

Mati Staniszewski 56:36 And then over time, there'll be a, a question of how you verify for, for human

Mati Staniszewski 56:40 part. I think that's going to be another big topic, but I would, um, I would have that. If I was creating social media, I would bring that as a big part of the equation. Um, uh, the... I'm still thinking about the incentive part. Like how, how would I incentivize the positivity versus the negativity? And that's-

Nikhil Kamath 56:57 Mm-hmm

Mati Staniszewski 56:57 ... uh, that's, that's, that's, you know, like almost the world got at, at some places calibrated this way where negativity gets more attention.

Nikhil Kamath 57:06 Yeah.

Mati Staniszewski 57:06 And like how do you shift that? Is it just giving more-

Nikhil Kamath 57:09 It doesn't even have to be positivity. It could be neutrality.

Mati Staniszewski 57:13 Positivity, like, you know, like, like... But then, um, uh-

Nikhil Kamath 57:17 Like if you're gonna trigger a emotion, like say the emotion could be greed, the emotion could be to pontificate. By being here, you sound smarter. This is a bunch of people who are trying to feed curiosity. Uh, there's a bunch of thing, other emotions that you can play with which are more on the right if the curve is left being negativity and right being positivity. You're more towards the right.

Mati Staniszewski 57:45 No, I agree. I... It's, it's, you know, like I think the, the, the key thing is you want to avoid the, the kind of the algorithm just rewarding what gives the, the highest, um, uh, knee-jerk reaction.

Nikhil Kamath 57:56 Correct.

Mati Staniszewski 57:57 Um, and, and it's like, it's like if you could have a platform that like inspires curiosity and, and like, and learning or like different points of view in a way that doesn't lead to just like two sides screaming at each other, incredible.

Nikhil Kamath 58:10 Yeah.

Mati Staniszewski 58:10 I mean, I-

Nikhil Kamath 58:11 Maybe we, maybe we can work on this together. You can take care of the voice part of it.

Mati Staniszewski 58:15 Let's do it. Let's do it.

Nikhil Kamath 58:15 Yeah.

Mati Staniszewski 58:15 It's... I feel like the... It's like the... It's, you know, there, there's no, yet there's no product in the space-

Nikhil Kamath 58:21 Yeah

Mati Staniszewski 58:21 ... on the consumer space that has like AI enabled in the right way. So in the-

Nikhil Kamath 58:25 Yeah. Yeah

Mati Staniszewski 58:25 ... it's like, uh, there's no AI social product today.

Nikhil Kamath 58:28 Yeah. Maybe we'll catch up in London or we'll do a con call. I'll add my team and we can ideate. They have some ideas as well.

Mati Staniszewski 58:36 Or in India.

Nikhil Kamath 58:37 Yeah, yeah. You're coming in Feb, right?

Mati Staniszewski 58:38 Exactly. And just for you.

Nikhil Kamath 58:39 Yeah. Hundred percent. Yeah. Okay. Thank you so much, Mati, for doing this. I think we're out of time.

Mati Staniszewski 58:44 Thank you. Thank you so much.

Nikhil Kamath 58:45 And we'll catch up soon.

Mati Staniszewski 58:46 Let's catch up soon, and let's, uh... Thanks. Uh, it's amazing to see how you support the ecosystem, so hopefully, hopefully we can do that too. And, and, and thanks so much.

Nikhil Kamath 58:53 Thank you. [upbeat music] How long are you there for?

Mati Staniszewski 59:04 For a week.

Nikhil Kamath 59:05 Yeah. Do you have time-

Mati Staniszewski 59:06 Yeah

Nikhil Kamath 59:06 ... to travel?

Mati Staniszewski 59:06 Yeah. I, I'm, I'm gonna be between Bangalore and Mumbai mostly, but happy to go where-

Nikhil Kamath 59:10 I'll take you to Goa for a day or two if you have time.

Mati Staniszewski 59:13 What's in... What's, what is it? Do you have offices in Goa?

Nikhil Kamath 59:16 No, I have a house in Goa, which is a lot of fun.

Mati Staniszewski 59:18 Oh. [laughs]

Nikhil Kamath 59:18 And I have dogs in Goa. [laughs]

Mati Staniszewski 59:21 [laughs]

Nikhil Kamath 59:21 It's, uh, it's like a beach town, which is very relaxing.

Mati Staniszewski 59:27 Okay. Amazing.

Nikhil Kamath 59:28 Yeah. We'll plan it.

Mati Staniszewski 59:29 Thank you.

Nikhil Kamath 59:30 Yeah. I think we'll-

Mati Staniszewski 59:30 If you're in London, let me know. I'll take you too. I don't have dogs, but, uh-

Nikhil Kamath 59:32 [laughs]

Mati Staniszewski 59:32 [laughs] I can... I'll take you to cool places.

Nikhil Kamath 59:35 Cool. [upbeat music]