IIT to Silicon Valley: He Built the World's Fastest Voice AI | Karan Goel, Cartesia

What's our biggest differentiators? A, we have the fastest models in the world. In speech, you need to have low latency in the interaction. If there's too much lag, you'll be like, is anybody else on the line or is it just me? And I think Delhi is a pretty vibrant place. People find ways to build in the chaos, and I think that's what companies are about. It gives you a little bit of a natural advantage when you can sort of deal with chaos well. One of the themes in my life is I didn't know what I wanted to do next. Grad school is a very common way to delay the decision of making a decision.

You were able to convince a bunch of people who were doing PhD with you as co-founders, and you ended up raising about $27 million before you had a product.

And yeah, you're right that when we raised that seed round, we didn't have a product, we didn't have revenue, we didn't have a clear sense of what it is we're trying to productize. But what we had was a really good technology and we had some really interesting demos. That was where we started.

What does Cartesia do?

Imagine you're a small business and you can't afford a receptionist. So you actually, you know, go out, you're on the job all day, right? You're a plumber, maybe, or you're an electrician. If you're not there to take the call because you're on the job, you're basically losing revenue. AI can answer the phone for you. It's all hands-free. What we basically are doing today is helping businesses automate conversations. So it's about access to information and making that easy. From an ambition perspective, people should try to believe that we are going to not just build the best technology, we're going to export it.

That should be the ambition, not just build it for India, build it for the world from India. What if What if AI could answer your calls, qualify your leads, and speak fluently in any language, all in under 45 milliseconds? My guest today is Karan Goel, co-founder and CEO of Cartesia, the company building the world's fastest, most localized voice AI. From IIT Delhi to a Stanford PhD to raising $27 million before having a product, Karan's story is one of following curiosity, Not a Plan. Let's get into it. Karan, excited to have you on the show.

Great to be here. Thanks so much. And by the way, do you say your name Shiva or Shiv?

Shiva.

Okay, great. Great to be here, Shiva.

Awesome, Karan. You know, we first interacted at our common friend's place, Akrit Vaish, and it was a very interesting conversation that you and Akrit had. And I've been wanting to have you on the show even before that. So I'm glad, you know, I made it to that event and convinced you, Karan. Look, Karan, you're building a very, very interesting company. You know, it's certainly one of the top AI voice companies in the world called Cartesia. Before we get into that, Karan, you know, we are really byproduct of environment. Uh, you know, what really shaped you the most while growing up in India?

Uh, yeah, that's a pretty, pretty, uh, big question, I would say. Um, so I, I grew up in, uh, Delhi, and, uh, I, uh, I would say a few things definitely that I was pretty, uh, impacted by. One was, of course, uh the school I went to, which I think has been in the news a lot, which is DPS Agyapuram. Pretty good environment to learn and sort of be with people who are excited to build. I think you see a lot of founders come out of that high school and middle school for some reason. There's something about the environment. But also, like, you know, I was always into math and science and you know, just liked computer science, engineering. And so obviously that lends itself well to the career path I'm on right now. And I think, you know, Delhi is a pretty vibrant place. You know, people find ways to build in the chaos, and I think that's what companies are about. So I think it gives you a little bit of a natural advantage when you can sort of deal with chaos well, and I think that's what maybe a lot of Indian founders have. Yeah.

So look, you decided to do electronic engineering at IIT. You always wanted to do that, and what was the experience like at IIT?

Yeah, I mean, I think, you know, like when I was— I like tinkering when I was growing up, and, you know, I When I was in high school, I actually wanted to be a doctor for a while, like a biology doctor, not a PhD, which is what I am right now. So, you know, irony. But I think the, you know, I ended up just sort of as kids do in 11th, 12th, we're getting into the whole JEE prep and deciding that I wanted to maybe just study for IIT. and I was fortunate to get in. And then, you know, you get your rank and you figure out what you want to do. I wanted to stay in Delhi, uh, and, um, I got electrical engineering, and I, I was, you know, I was happy with either computer science or electrical engineering, so I ended up doing that. Um, I think, um, it's interesting because it's called electrical engineering, but obviously in the US they call it ECS or EECS, uh, and it's pretty similar, right? It's just like a combination of like you learn about how computers work and you do some amount of like computer science work, and then you also learn how circuits work. That's kind of the curriculum. And I think IIT Delhi is a pretty interesting environment, right? Like, it's— there's a lot of really smart, high IQ people there, uh, cloistered together, uh, in a pretty, uh, nice campus in the middle of the city. I think in terms of like location, it's up there, right? You're literally in the middle of the city, you know, horse cars and SDA and Wasan Bihar and, you know, all these places. So, um, I think it was a really amazing kind of experience to be there. Uh, I think obviously I learned a lot about, uh, not just, uh, electrical engineering, but, you know, one of the things that happened was one of my undergrad advisors, Professor Mawsam, who still is at IIT Delhi, and he joined, um, from the University of Washington as an assistant professor So he started teaching AI and it was super fascinating. And, and, you know, he was doing some really— I think he's just very passionate as a teacher and a researcher. So I, I got really excited about that, uh, by, by kind of connecting with him on that. And then I started working with him and that's what led to—

yeah, that's what led to what?

That's what led to me getting interested in pursuing, let's say, grad school in AI and things like that. But, but, you know, I think it was just the way that he created an environment where he got people excited. Many people from my batch and even junior batches, I mean, they, they really got interested in AI because of him. And even like literally in December, we met up. I mean, we meet at the machine learning conference, you know, and a lot of the ex-students get together and hang out. So, so I think that's like a there's a community he created in terms of like people who really want to be at the frontier of AI and doing the research and thinking about these problems. So that was really the place where it started for me. Yeah, got it.

Uh, and Karan, post that, you know, you moved to U.S., you went to Carnegie, you did machine learning, uh, and then you decided to, uh, do your PhD at Stanford. So kind of like just talk us through that journey.

Yeah, hi. So I, uh, I, in undergrad, you know, I didn't really know what I wanted to do with my life, so I did research and I, I decided, uh, to apply to grad school. And, uh, mostly I think I was like interested in doing more in AI, and, and, you know, it seemed like the obvious way to do that. And, you know, keep in mind, this is, uh, I graduated in 2016. This is sometime in 2015 when all this is happening. You know, this is very much before, you know, all the stuff that's happening now in AI, right? So at the time, it wasn't like you go work at, you know, at some like great AI company or something like that, right? That's not a career path that existed then, at least not an obvious one. So I think for me, like, I was excited about these problems enough to decide to go to grad school. I then, you know, ended up at CMU. CMU was interesting as an environment. You know, I, I think they have, um, they have an amazing like, um, machine learning program. They have a robotics, um, program that's amazing. And they have a school of computer science, so there's like 7 departments or something like that. And machine learning, actually, I think they were one of the first schools to actually have a dedicated department for machine learning. So I was— I did a master's in machine learning that came from that department. So it's a pretty small program. I think there were like 17 master's students, 8 PhDs, and pretty tight-knit group. So I think the environment was super amazing for getting super immersed in AI, which is rare, right? Because for most schools, it was still one of the things they taught. So it would be like, you learn operating systems, you learn distributed systems, you go to, you know, algorithms, so on. But like, here it was just like everybody in that department just cares about AI and machine learning. So the concentration of talent and sort of like the amount of work they put into building a curriculum that covers sort of the breadth of what you can learn there, it's, it's pretty amazing. I think it still continues to be one of the most amazing programs. So I learned a lot. I mean, I would say in undergrad I developed an appreciation for research, and then in a master's program I got super into like just more deeply into machine learning and learning more. About what's important in the field and what are the problems people care about and so on. So that kind of led to me, you know, kind of getting excited enough to go apply for PhD. And by the way, once again, one of the themes in my life is I didn't know what I wanted to do next, which is often a very common theme. Like, I was just like, okay, like, I did this master's, what do I really want now? You know, I think grad school is a very common way to delay the decision of making a decision. So, um, so I ended up just deciding, hey, like, let's just try doing a PhD now, like, you know, because maybe I don't want to work yet, right? So that's kind of what I— and I ended up applying to grad school. There was also some interesting set of circumstances in my master's that happened. I was fully funded, my professor moved to Stanford, I ended up actually visiting Stanford for a year during my master's and working out of San Francisco in the Bay Area. So I was fortunate to then, like, spend time on campus also at Stanford and build some connections and sort of learn how that department works. And so when I ended up applying to grad school, I was like, you know, either I'm going to do a PhD here or at CMU. I don't want to go anywhere else, really. So, and ended up happening that I was admitted. So very long story, but, uh, you know, I, I think, uh, it kind of worked out well in the end, but, uh, not because I had some like deep goal of like I need to do a PhD in my life. I think it was just like You just stumble into it.

Stumble into it. And it was more around the environment you were in and that kept giving you the direction. And Karan, look, you know, I think in 2023, you know, you were able to convince a bunch of people who were doing PhD with you as co-founders and you ended up raising about $27 million before you had a you even had a product. I might be wrong, before you had a product. What was the pitch like to, I think, Index, and how were you able to convince your co-founders to join you? And what does Cartesia do in simple language for the audience who doesn't understand AI that much?

Yeah, yeah. So, you know, I did my PhD and it was a 5-year PhD, very long. It would seem, but, you know, most PhDs in the US are 4 to 6 years typically. And so, you know, I spent that time just learning about lots of different things, working on lots of different problems. One of the key things I worked on with one of my co-founders here at Cartesia, Albert Tu, was new sort of ways to build AI models, architectures for AI models. And so we had this whole, you know, line of work that we did together. And he was sort of— he's a genius, so he's a pretty big pioneer in this field. But, you know, I was fortunate to kind of get to work with him on a lot of these problems. And we had a lot of fun. Honestly, that was the biggest part of it. We just enjoyed working together. So when I was graduating, he had just wrapped up a year at DeepMind. A couple of my other lab mates were also thinking about what they wanted to do next. Our PhD advisor is Chris Ray. He's a serial entrepreneur. He's built a lot of companies. He was helping us as a co-founder, also helping us think about what we could do with a lot of the technology work that we've done during our PhDs. So we kind of came together and I convinced a couple of these folks, including Albert, that this was a good idea to spend their time on. And I think the main reason be— I was able to convince them to work together was, uh, that was going to be fun to do together. Yeah, so I think, um, you know, people— I mean, most people who, uh, uh, form, you know, uh, teams as founders, I think it's just about chemistry, right? Like, what do you feel like you have mutual respect, uh, in terms of How much you value or how highly you think of someone, and then do you have the trust to be able to work together, and do you actually have fun doing it? Because it's a grind, obviously. So I think all, all of those things were true because we had tested it live by working together during our PhDs. So there was no question about that, and I think it was just about whether everybody's ready to take the jump, which, you know, obviously By convincing them, you know, we got that over the line. So, um, so yeah, that was, that was the origin. And then once we started the company, I think, um, you know, part of being, uh, in Silicon Valley at Stanford means that you do have access to a lot more resources and capital. And like, uh, there are a lot of people around you that can help you because they've done this before many times. And, you know, Chris was a PhD advisor. He had He started many companies, many billion-dollar companies, and so, you know, he knows how— and a lot of them are deep tech companies— so he knows how that transition can take place for a very technical founding team from the lab to, you know, a business. And so he was able to sort of help us through that process, and that includes, like, obviously we raised $27 million, but we raised it across a pre-seed and a seed round. It was not very far apart, like 4 months apart, 4-ish months. But the pre-seed was really like, here's some money that we can use to get going. And the seed was really our sort of first official seed round. And yeah, you're right that when we raised that seed round, we didn't have a product. We didn't have, you know, revenue. We didn't have, you know, a clear sense of what it is we're trying to productize. But what we had was a really good technology. And we had some really interesting demos. That was where we started. And, you know, I think the technology was interesting because we had spent close to 5, at that time 4 years on it. And so we had like actually published a lot of papers. One of them had won an award. And so there was a lot of work behind it. And like you could go talk to people in, in machine learning, AI, like researchers, and you could ask them about the work we'd done and they would be able to give you some sort of informed opinion on it. Uh, and you know, they could say it's good or bad, but the fact is it was actually fairly well known, which helped us a lot, right? So I think, I think when you're trying to do something new, obviously not everybody's going to agree that's a good idea. But the fact that everybody knows that that idea exists matters a lot. So I think that all helped us. And of course, like we focused on voice. Now you asked me, what does Cartesia do in a nutshell, right? You know, what we basically are doing today is helping businesses automate, uh, conversations, right? These are conversations that might happen inside the business. For example, an employee has a question about their healthcare. These are conversations that might happen outside the business. So for example, a customer calls you up and asks you a question about their account or wants to know where their order is or so on. And these could be questions in a marketplace like buyers and sellers connecting together. These could be questions in, you know, different industries, like in healthcare versus in insurance versus in, you know, public services. Like, this could be a government trying to answer a question that a citizen has. So it's about access to information, right, and making that easy and doing that in a way that's very localized, right? Like, the idea is, like, voice as a technology is about connecting people and connecting people to each other and connecting people to systems. And doing it in a way that feels easy for them, right? So that's what we're trying to empower people with. How can we partner with businesses so that they can do that super easily? And, and, you know, the nuts and bolts of that are you need models, you need infrastructure, you need to have agents, you need to be able to implement them, and you'll be able to scale them, you need to be able to evaluate them. That's what the product is. But, you know, at the end of the day, the problem is that, right? Um, so, so that's what we do, and we, we do all of that work to help customers, which makes us a pretty nice I would say a partner for a lot of businesses because we kind of take the problem off their plate and we do everything from models, as I said, all the way to agents. And yeah, I'm happy to talk more about that, obviously, but that's kind of a little bit of the snapshot. Yeah, got it.

Now, Karan, give an example. Could be, you know, an example where, you know, a company that you're helping with, how are you helping them? How are you better than, you know, other solutions in the market? And when is that time where, you know, that company can't, you know, function without Cartesia?

Yeah, you know, I think, let's take an example. So I think, you know, we have customers in pretty much every sector of the economy, but I'll pick an example at random. One problem that a lot of folks are interested in is how do I make my salespeople spend their time on the right prospects. So, you know, as a salesperson, you only have finite time, uh, and you want to qualify your customers as well as possible, and you want to spend the time on the most qualified customers or leads, right? So lead qualification basically becomes the use case. And so, um, you know, you can imagine, for example, you you're trying to sell real estate. So you have a high-value asset that you're trying to make a sale for. As a broker, you would only want to talk to serious buyers. So how can we help with voice in that situation? Up there, what you can imagine we want to do is actually talk to those leads, and you could have, you know, many of them, and actually try to understand who's a qualified buyer for this. and then have those folks be connected so that, you know, people can spend their time talking to those folks and you can kind of make that connection. Well, so that's an example of a pretty end-to-end use case where you're trying to help someone be much more productive. And this leads to actually like improvements in revenue, right? Because at the end of the day, you're actually helping someone spend their time on, uh, places where they can grow their business. And this kind of lead qualification, you know, it's not just in I talked about real estate, but another example is imagine you're a small business and you can't afford a receptionist. So you actually go out, you're on the job all day, right? And you're going and talking to people on the phone, but you're also like doing work, right? So you're a plumber maybe, or you're an electrician or something like that. And so how do you make sure that you can fill out your calendar with business tightly as possible and make, you know, the most amount of money per day. Uh, let's say somebody calls you, right, and they're looking for a plumber. They're going to call 3 or 4 plumbers, they're going to find who's available, and they're going to go, uh, you know, sign them up in some time slot. If you're not there to take the call because you're on the job, you're basically losing revenue. Uh, and so that's an example where like this kind of capability actually gives you a superpower that didn't exist before. Which is that AI can answer the phone for you. AI can talk to your prospect. AI can figure out, uh, are they someone who's actually looking for services you can provide, and it can even schedule it for you. And you don't have— it's all hands-free, right? Basically, you don't have to pick up the phone. So these are examples on the sales side, but you have so many examples in support where you can, you know, as I was saying, like, you can have somebody call in, who knows what issue they have, right? But you can have AI actually triage this whole issue, and then maybe it's too hard for AI today, a human can address it, but maybe the AI system can actually action it, right? Maybe it's like process a refund or something. So these are all different use cases— sales, marketing, recruiting, support, you know, where we're seeing people adopt voice and build really interesting applications. And the, the, there's no limit to this. This is like, there's a lot of creativity that's going into these use cases where people are just coming up with new things and just building, and they're going out in the market, they're trying to understand, do people have this need, and then they're selling it. So where we come in is, you know, we are the layer that makes this possible from a, uh, you know, conversational agent perspective. So what's our biggest differentiators? A, we have the fastest models in the world. So how that helps is, you know, in speech you need to have low latency in the interaction. Otherwise, you know, if there's too much lag, you'll be like, is anybody else on the line or is it just me? Right? So we think a lot about the milliseconds in conversation. That's one area where we are really, really good. The other thing we've done is we have super focused on localization. So today we have the best models for Indian languages for speaking, so synthesis. And the reason is we obsess about data quality and we go very deep in understanding the nuances of how people speak in a specific language, and we bring that into our models. So we've like invested a lot obviously for India, and, and, you know, we have we have amazing models that sound super realistic in all these different regional languages, you know, let's say Marathi and Malayalam and so on, and as well as like having obviously great Hindi and English and so on. So I think like our whole approach is like, as I said, like voice is going to be such a key way in which people transact with businesses. Yeah, that, um, that you cannot have a product that's not hyper-localized and feels easy and natural to use. And that's what we're really building for people.

Got it. And especially India. India is purely, you know, voice-first economy.

India is amazing for voice. Yeah. India is amazing for voice.

We'll talk about India in a bit. Karan, you know, feel free to name a few customers that you have.

Yeah, I mean, I think, you know, we have a bunch of folks on the roster. You know, everything from ServiceNow, which is, you know, one of the biggest companies in the world, and they use us to build voice agents for their customers. So this is more in the B2B2B side. And so for them, it's really about both internal-facing use cases, like how can we enable employees at a large organization to ask and answer questions, whether it's like IT support or so on, all the way to like, you know, external-facing support. So these are all really amazing use cases that we're working on with them. You know, we have— I cannot name it, but we have, for example, a government that we're working with that's, you know, deploying us for citizen-scale use cases. And there it's sort of like you have folks from the country that need to ask and answer questions about some specific, let's say, processes. And so they can just, you know, in the local language actually interact with these use cases. One of the common themes amongst our customers is that we deploy in a way that's very secure. right? So, so we try to talk, as I said, we are very business focused. So we partner with businesses and we try to understand like how do we enable them to do voice applications but do it securely, privately in this localized way because voice is really biometric data. Yeah. And so, um, that's, that's super important. But we have all kinds of other customers, you know, everything from a Yelp to a Indeed to a you know, a Decagon, like lots of different industries, use cases. And then, you know, even in India we have a bunch of customers. Some of them I'm not allowed to talk about publicly, but, uh, you know, I think everything from 1.1, which is a publicly traded company, to a Magic Bricks, to, uh, to others. So there's a lot of, um, great kind of Indian companies we're working with, um, as well that, uh, I think are really starting to kind of figure out how they can deploy the technology with their customers. And the scale is what's so exciting in India because it's just, you know, there's just— voice is big and also it's crazy scale. Yeah.

Karan, you know, look, you've built the fastest model when it comes to latency. I think you brought it down to 45ms. Typically it's around 90. And you're the best when it comes to localization, meaning local languages, and you have customers like ServiceNow and other great ones. What's the magical outcome you hold for Curtisia? What would it eventually end up becoming?

Yeah, I think that's a great question. You know, I think in the short term, we are super excited about enabling as many customers as we can to adopt voice as a technology and help them be successful in bringing their use cases to production at scale. That's it. That's all we want to spend our time on. When we only have one focus, like we only build this one kind of product, which is for conversation, and we're only focused on businesses. So it allows us to have like hyperfocus when working with companies. and I think we just want to make that super, super successful. So that's one. Two, I would say Cartesia is sort of an interesting company because we're a research lab meets a product company. So we kind of have this, um, uh, core ethos of research, and we do a lot of research in the team, and then we're using that research to build innovative product. So, um, so because of that, I would say like, you know, We sort of think of ourselves as pursuing sort of frontier work in AI, and there's a lot of really interesting work we're doing, especially as these models that we're building start to get more and more intelligent, right? So we think that the future should look like, um, it should look more like you have a— you have the ability to hire Cartesian model, uh, as an employee to do the work of your business. That's kind of where we want to go from a product perspective. And so to do— to be able to do that, we need to build these pretty amazing models that can work like humans do inside a business and get things done. And I think right now what we're doing is sort of being able to solve for one or two of the main capabilities of certain types of roles inside a business. So for example, If you think about support, I'm effectively doing some part of the work of a call center agent. But I think what we're really interested in is what does this mean for the future of work when it comes to AI inside your business? And how do you kind of have an autonomously acting, thinking, listening, talking system that can just do work on your behalf and you can collaborate with it? So that's where we're going with this. And so I think We sort of see voice as sort of the starting point for this because it brings in that element of connection, collaboration, and action. But then we need to move towards a future where these systems are actually much more autonomous. So rather than like a call comes in and you have an AI model being able to take that call on and take action, but then once the call is over and some of the actions are taken, everything is silent now, like the model doesn't do anything else. Instead, we want to be more proactive, right? So instead, it's like the model is always doing something. It's always, you know, cleaning up some system, doing some work, taking calls, taking action into systems, just making life better all the time. That's where we want to get to. So I think that's going to be a super interesting hill that we want to climb in the next 2 years. And it's going to require a lot of technology work. So, we have a really amazing research team led by my co-founder Albert. It's about, you know, we have 35-40 people in research and then we have about, you know, 35-ish people in our engineering teams and then we have the rest in go-to-market. So, we're like scaling all these teams and, you know, we've been, I think we've doubled in size in the last 6 months, maybe 5 months. So it's been a huge, huge sort of undertaking, but we're excited about that future.

Yeah. Yeah. I remember, I think at Akrit's event, you mentioned that now is when you're hiring your first salesperson after such a long time. So look, Karan, you know, a lot is happening in AI, you know, lot of great talent is working at OpenAI, Gemini, Claude, and other AI companies. Why do you think the top talent would want to leave those top companies and work at Cartesia?

Yeah, I think, you know, I would say One thing that I would, I would first want to talk about is, you know, who is the top talent? As in, you know, how do you define who can do sort of groundbreaking new research, do great work? I actually think often when you are trying to, let's say, hire someone from a place like OpenAI or Anthropic or any of these places, you've actually kind of lost some of the alpha on finding great people because they're kind of already proven out. Like, what they did was actually what you want to do, which is they found people that were diamonds in the rough, and then they actually went on to do great work at those companies. So actually, the problem is a little bit different from our perspective. Like, we're not really trying to hire people that have already been successful. Necessarily, although that can be helpful in some situations. I think what we're trying to look for are really great researchers and engineers that are excited about the mission that we have, that are willing to go in and work super hard, and that are, you know, technically brilliant. But they may not have, you know, a long resume of achievements from the last 10 years. And I think that's kind of the talent strategy that we have. So we've been able to use that to hire some pretty amazing people that I think would not be out of place at any amazing lab. You know, as I said, my co-founder Albert spent a year at DeepMind. So we have a pretty, pretty strong understanding of what great talent looks like. I, you know, obviously being at Stanford, many of my classmates went to OpenAI and Anthropic and, you know, Meta and all these places. So, you know, it's, it's I think we have a very clear calibration of what talent is. And then we try to find people that are at that level, but perhaps haven't had the opportunity to be as successful yet.

I see.

So that's, that's how you, I think, try to build a talent moat, let's say. And then, and then of course, when you do that, and you can actually get to a great stage, I think all of the folks that might already be at these larger places will want to join. Because ultimately what we're doing is completely different from what the labs have done so far. And that's the only reason to want to join, which is the mission is so different, the work we're doing is so different, you won't get to do it in the same way at a different place.

Got it. Talking about the talent, Karan, you know, you've been spending a lot of time in India. I think you made a couple of trips to India in the last 4 months, and you also, you know, talked about it that you know, when you are in a country outside of your home turf, you can't just go there for 2 days, you know, hold 10 meetings and come back. What's been your experience like? What have you been doing and how are you looking at India for Cartesia's angle?

Yeah, I mean, I think, you know, I obviously have a natural affinity towards working on product that is good for India. Um, we, uh, you know, obviously we're based in San Francisco, but, um, you know, having grown up in India, I was— my parents still live there. So I think there's a, there's a clear motivation to, uh, especially in voice, to work on problems that can be helpful in, in India. So I think, like, I think, like, you know, for companies that are operating internationally I think it's important to have focus on markets where you actually want to be helpful to the people there and you understand what they need, right? So I think from that perspective, I feel like, uh, you know, I actually have spent a lot of time on the ground. You know, I, I traveled to 5 cities. I spent countless hours in meeting, you know, founders and enterprises and you know, folks in government and, you know, folks who even in, you know, religion and, you know, all kinds of different parts of life to try to understand how can we make our products better for the people that want to use them in India. And my perspective is it's really, really hard to really build a deep understanding in a couple of days in a market that's as big, diverse, important as India is just not the way business is done in my head. So I think for me, from my perspective, is like, if you want to be great here, you have to spend the time. And by the way, this is coming from someone who spent, you know, 23-odd years growing up there. So I already have like a pretty big, you know, prior on everything and how things work and all of that. You know, it's not like I'm starting from zero. But I think it's super important to then engage with folks on the ground. So that's what we're doing. I would say our strategy is not to be— we're not trying to be pushy, I would say. Um, I think that what we're trying to do is figure out like what would make the product the best thing ever for companies in India. And then I think when you have a great product, people see that. And even when you do sales in any, you know, whether it's in the US or in India, if you're actually genuinely interested in solving somebody's problem, I think people care about that, and then they want to work with you. Because too many people view sales as a transaction. It is a transaction, there is an exchange of goods and services, but there's also like, are you actually prioritizing and do you care about the problems that those customers will have over the next 6, 12, 24, 4 years, 8 years? That's the real question, right? So I think that's why I think showing the commitment is important. So to me, from my perspective, it was like, if you can't even spend a few weeks within a market which you think is important with the customers that you actually want to help, you shouldn't be in that market because you're not committed. Yeah. So I think that's kind of my perspective. And of course, we are trying to put our money where our mouth is, let's say. And, you know, we're building a team in Bangalore, but we're also doing it in a very thoughtful way. And we're trying to hire amazing people there and give them the opportunity to work on interesting problems. This is not, you know, and I have my own opinions on, you know, how to hire talent, as you can probably tell, but like, for me, it's super important that everybody at Cartesia gets the opportunity to work on important problems. We don't believe in back offices and any of that. Like, we only want to hire the number of people that are necessary to do what we need to do, and we want to make sure they all get the opportunity to do great work. So that's the philosophy, like, it's to have everybody be part of this core team and, and be contributing. And so I think India is going to be a massive opportunity for us, but there's also so much to do because the— it's a complex market, right? Obviously there's a lot of linguistic nuance, there's, uh, it's a cost-sensitive market. Which means that you have to price and package in the right way for the market. So these are all things we think through holistically when we're trying to build business in India, and it's going to operate pretty different from the US. We obviously have a big business in the US, but I think that's— it's just, it's a different, you know, it's different, right? Like, and you have to kind of adapt your whole business towards how things are done there. So that's, that's my belief, and that's what we're trying to push forward.

Got it. So it seems like, you know, as you mentioned, India is a It was a big opportunity for Codicia, and you know, you're thinking differently, you're hiring in India, you're talking to customers, you're looking to onboard customers in India. Are there any specific talent or type of talent that you're looking to hire? And maybe talk about one or two learnings that you got by talking to the customers in India.

Yeah, I mean, I think in terms of talent, We are A) hiring on the engineering side. So, we are hiring great platform engineers, folks that have strong software skills. That's one bucket. We also just 2 days ago started hiring research engineers in Bangalore. So, we are now opening up more opportunities to work on our research problems from India. As part of our core research teams. And this is, again, this is, uh, the first time we're doing this, and I think it'll be quite interesting to see how that plays out. It's an experiment, right? So we have to see how it plays out, but I'm pretty optimistic we can find some great folks. We never change the talent bar anywhere. Everybody interviews at the same standard everywhere in the world where we hire. Um, and then, um, you know, in terms of learnings from customers, you know, I think Um, um, you know, one, one thing we've seen is that in India there's, um, a, there's a lot of use cases that are obviously like consumer-facing, and, um, uh, just because, you know, consumer applications are such a big part of how people have built businesses there. So that's obviously one area where we're seeing, um, interesting opportunities. I think people are looking for opportunities to monetize better typically, right? Because again, it's a cost sensitive market, you know, monetization opportunities tend to be, uh, something people are really excited about. So I think, uh, we are doing some collaborations to figure out, okay, what are better ways you can try to actually monetize consumers using voice? And I think that's an interesting area. The second is, I think, um, what I would have observed in my trip was BFSI as a segment is really exploding in India in voice. It's, it's, it's just off scale in terms of the interest and the adoption that's happening. And so it's clear that's going to be a pretty important segment for voice. And it's actually a place where we've like, again, targeted the product because in BFSI, privacy, security, compliance, super important. And our whole stack is basically portable and deployed on-prem. So that allows us to have like a product that can actually really fit the bill for what these customers are looking for. So I think there's a, there's a really interesting sort of movement in voice happening, but I would say voice in India is still early. There's clearly a huge amount of growth that needs to happen. And, um, and I think there's a lot of exciting companies, startups that are building in the ecosystem that are doing really interesting new work. Uh, you know, I, I, uh, I was not able to go to the AI Summit because I just spent 5 weeks in India and I came back, and my co-founder actually went there, but But I think like the AI Summit is an interesting example where like, of course it was an amazing event and all these people came together, but the hard work of building happens outside the summits, which is why I spent— I was like, I would prefer to spend my time actually traveling and meeting people in their offices than having a 30-minute conversation with them at a summit where maybe they're not, you know, necessarily going to have the bandwidth to talk about the problems deeply. So I think there's a tendency to over-optimize for the big summits and less for the work that happens in the middle, where all the actual work gets done, by the way.

Yeah.

Yeah.

Karan, you mentioned that, you know, India should not just become a consumer country of AI. It should produce things. In your opinion, Karan, what's the biggest advantage India has when it comes to AI?

Yeah, so I think, you know, I've been saying this for almost 2 years, which is I think the people in India need to have very ambitious goals around building and owning their own AI stack. And I think that Um, you know, I— having gone to IIT Delhi and like, you know, having lots of friends who are super talented and, you know, uh, clearly capable of doing amazing feats of engineering and, and technical feats, I think, I think it's just like organizing the system so that people can do those ambitious projects is super important. Having the capital to do it, having the talent density in, in in place to be able to do that, making it easy to hire, making it easy to build these kinds of companies. I think that's super important. So obviously, like, I had seen, you know, Sarvam did some really interesting releases at the AI Summit, which I think is a really good sign of progress, right? Like, they've invested time and capital to try to build things there, and I think that makes a lot of sense. And so I think for companies like us that are working in India, I think it's super important to, uh, and this is what we do, is super important to help companies own and operate their stack. So we actually, as I said, we try to hyper-localize, and we're saying, here's all of the stuff we've done. It all runs in India. The data never leaves the country. You know, it's all run and operated there. So I think, I think that's part of it. But also, like, I just think it's important to have the muscle to do homegrown technology and try to compete with, with, you know, China and the US on these things. Just the act of competition, I think, sharpens up, uh, you know, um, sort of your technology sector and so on. And also think, for example, IT services as a sector, you know, it's not going to be the same in 2 years or 3 years or 5 years. It's just going to be totally wiped out and disrupted. So the business that, you know, relied on exporting Indian talent as a service overseas, it's not a very good business anymore. And so how do you then replace that with higher technology businesses? Like, how do you move towards more technical businesses? I think that's important. And of course, India has an amazing— I mean, what is the biggest advantage India has? It has a, you know, massive consumer base, consumer population. And like, I think that's going to be the biggest thing to watch for, right? Like, how does that increasingly monetize? And that creates opportunities for business. At the end of the day, the person paying the bill is the consumer, right? Everything else is, you know, downstream of that. Somebody has to pay the bill. And so the more, you know, you kind of get this sort of spending happening on the consumer side, the more businesses that can deliver, you know, high technology services and solutions will thrive. So I think those are the things to watch out for. But I think from an ambition perspective, people should try to believe that we are going to not just build the best technology, we're going to export it, not just for their own population, export it. Because like, we're in the US, we export technology to the whole world, right? So I think that's, that's, uh, that's very important.

That should be the ambition, not just build it for India, build it for the world from India.

Yes.

And Karan, look, we got one last one. You know, my co-pilot in the back, and if I have introduced you to him, his name is Alfonso. He wants to get in and he's asking, you know, we know Karan, you know, as a smart guy, super smart, PhD, AI. Researcher, now building one of the most interesting companies, and also is a great speaker, can really articulate amazing things. What do your friends know you for? Who are you outside of work?

I mean, I really enjoy— I was always a very obsessive gamer growing up. I am— so a lot of my friends know I'm a I'm very avid gamer. I, and unfortunately I don't play very much, but, you know, I'm a big consumer of esports as a viewer. So that's something that, you know, people can find me on my couch watching Twitch streams a lot of the time. So I have some weird hobbies basically, is kind of what my friends know me for. But, you know, playing video games unfortunately is not one of them because it tends to get in the way of building a company. But yeah, you know, I just like everybody else, I actually met my wife at IIT Delhi. So, you know, I try to spend as much time as I can outside of work with the family and my wife and try to travel and do things. So yeah, doesn't leave time for much else.

Love it, Karan Goel. We look forward to, you know, seeing you more more and more in India. I'm sure you'll be visiting this year as well, multiple times. Yes. Thanks for coming on and really sharing your journey so far.

Thanks so much, and hopefully we see more cool tech come out of India. Excited to continue to meet people, and you know, I'm going to be back soon.