Full Transcript

Dhruv Sharma: Hey there. Listeners, this is TON 113. We are streaming live. And I'm really excited about this one. Also, because my good friend Kashish is co hosting TON 113 today. And so, Kashish, like, it's great to see you. And how are you doing?

Kashish Sharma: Before we getting this alternative career choice that you guys keep very graciously inviting me over to

Dhruv Sharma: it looks good in you. And we obviously love having you and are you following the World Cup?

Kashish Sharma: I am. Unfortunately, the time zones are a bit of a doozy, right? But I mean, of course, last night's match was pretty crazy. France versus Spain, rooting for Spain. So okay, for me, at least.

Dhruv Sharma: And will you like depends on who's going to make it to the finals? But is there any one team out there that you're really rooting for right now? I mean, we saw Messi win last time around.

Kashish Sharma: So this time for me, it's actually going to be England. Right? I think it only makes sense, right? Let's get the Englishman World Cup trophy this time around. It's been a while.

Dhruv Sharma: Great. All right, let's welcome our guest for today. We have Harshan from the Interact group. And the you know, the product which is incredibly popular is Friend. Welcome to TON guys. It's great to have you.

Bhanu Pratap Singh Tanwar (FRND): Nice to be here.

Harshvardhan Chhangani (FRND): Yep.

Dhruv Sharma: All right. So I think let's just start with you telling us, you know, how you think about the company and how you describe it these days.

Bhanu Pratap Singh Tanwar (FRND): Cool. So I'll take a crack at it and Harsh, feel free to jump in whenever. So we started Interact. Interact is a parent company. Back in 2019, with this thought process that social media is becoming content social, and there's going to be a vacuum for interaction on internet where people or humans now that they are internet native, they would want to come on a platform with the sole intention to interact with other humans. So that was the overarching thought when we had started and we believe there's going to be a plethora of such apps and we are going to be holding up of these. We started FRND as the first product back in the day. And the thesis very simple and voice based avatar led platform where users from tie 2, tie 3, tie 4, which we call the next billion user base would hop on to a conversation with the motivation that they can speak something. It's a place where they belong. It's not a place where they consume content, but they will be heard. And fast forward to where we are, approximately 3 million users use it monthly. And the thought process is going to continue. I think we're going to double down on this thesis that when everything is becoming content, which plays on internet reminds humans to interact with each other. So that's about it. Harsh, feel free to add if you want.

Harshvardhan Chhangani (FRND): Yeah, so through the core insight was that when we were using existing social media products, we were trying to be in the shoes of the users that we were catering to, we could see that there was a strong need to interact with each other. It is just that the infrastructure of the existing products like Facebook and TikTok were not able to cater to that need because but they were trying to do start a conversation in comment section, even though the section itself was a very small part of the overall screen. And we could see that there was a steep demand the users coming from smaller towns, they don't have that kind of social circles. And they would want they would appreciate a product which is able to cater that particular use case.

Dhruv Sharma: I think will be very interesting to know is how you know, how the product itself shapes a user's identity as they interact more. So, you know, you mentioned Facebook and others, you typically see a photograph and whatever people have to say about them. But in the case of FRND, you see enough that and it's also it's social discovery, but it's audio first. So how does that identity get shaped over a period of time?

Bhanu Pratap Singh Tanwar (FRND): I think to start with, it removes all the baggage that someone would have in their life, which essentially shapes their behavior on socials as well. So there's no baggage, it's a judgment free zone. There's no pictures, essentially, your voice determines how you'd be perceived. So you can shape it, you can shape the platform according to your use case. So that's how I see that the judge, if you remove the fear of judgment, you know, people actually speak about beautiful things, they can be truly vulnerable. So this is how I see it. Feel free to add harsh to it from your so.

Harshvardhan Chhangani (FRND): So then, so one more another aspect of shaping the identity on social media is how the social itself rewards like what kind of behavior does it reward? For example, Twitter would reward being very intelligent and very, very quickly with your votes. But on FRND, once a user has come in, we know that the appearance in itself is not the primary identity of that person. There are certain features which will shape your new identity. For example, how good listener you are, how much respectful you are towards new people. For example, we had a feature, we have a feature called respect point. So if we both are chatting, and we know that we are anonymous, and it is the for the for we are strangers, there's a feature of respect points, if we both give each other respect points, that will shape our new identity, how respectful we are. So that in itself is also a very strong component in the product that once the identity is removed, what is the other what are the other parameters which are being appreciated by the product.

Bhanu Pratap Singh Tanwar (FRND): And I'll add one more thing to it. It is essentially, let's say a lot of social graphs would reward how engaging content you create. FRND social graph would reward how chatty you are or how warm a conversation you can actually hold. So that's slightly different from the way we have been perceiving socials in the last 10 years.

Dhruv Sharma: Can you also help us understand, you know, what scale FRND is operating at? Maybe the kind of, I don't know, the number of users, the conversations and what really is your North Star metric as well? Do you do you track the length of audio messages or like conversations back and forth? That'll be pretty interesting.

Bhanu Pratap Singh Tanwar (FRND): So we have around 3 million monthly active users. And they spend around 30 minutes per day engaging on the platform. In terms of what is a North Star, I think it's a combination of two, three things. It's of course the scale of the platform matters, but also how deeply engaged the community is. So how long they are speaking to each other, let's say another would be how many virtual gifts they are sending to each other, which in a way correlates with how many conversation minutes are actually happening on the platform. And in addition to it, there are North Stars around trust and safety, because when the product is pretty dense, and think of it like you have to manage a house party where there are 3 million people in a month. And it's a tricky task in that sense that they are the North Star would become, okay, how can I ensure a time well spent rather than let's say, increasing my time spent. So these are how the North Stars look like for us. If I missed any, Harsh, feel free to add.

Harshvardhan Chhangani (FRND): So I'll add another point there, which is very like the product perspective of the answer. And it isn't exactly the North Star metric, but what we have understood from the user, sorry, is anyone speaking? Yeah, so what we have understood from the user psychology is that there is a very strong barrier when users are fear of judgment, they have fear of judgment, the first 30 seconds or the one minute are the most crucial part. And so if you are able to cross that barrier, you can see that only after that you will be able to see the whole engagement and the time spent, time well spent. So what we do at Friend is we try to ensure that the first 30 seconds we are able to match with the right set of users, the product in itself has the kind of features will make someone who is vulnerable very warm in terms of let's say the suggestions being provided by the platform. So these are some of the things that we highly optimize on. And once these things are in place, the engagement follows. So we don't chase engagement right away. These are some of the important metrics that we critically work on and all the engagement that we get, which is I think around 30 minutes on an average user spending on Friend, which this engagement time is the second order effect of the effort we are putting for the first 30 seconds or the first one minute of all the users. Anyway, what I think to summarize it would be, let's say, how well do we icebreak when a user has just come out of the platform? That also becomes a daily north star.

Dhruv Sharma: Yeah. And so you're saying the average user spends about 30 minutes a day. Do they spend 30 minutes? I mean, help us understand that they spend 30 minutes talking to just one person or the two, three, four friends that they've made on the app. Also, what's the single longest chat any two individuals have had like uninterrupted that you would know and give us a sense of user behavior from that as well? Ash, why don't you?

Harshvardhan Chhangani (FRND): Yeah. So what we see is that users come and they come with an exploratory mindset where they're trying to find out people of their own community, their culture, specifically, the language and the region that they belong to. And they talk to multiple users trying to find out who is the right listener or the right conversation list for the kind of things that they want to talk to. And once they find someone, they will be interacting with them. They will find a belongingness with them and they'll stay for that belongingness. That is the kind of behavior we see. So users will come. They will come with an exploratory mindset. You will talk to, let's say, five, 10 people. They'll find two to three folks with whom they'll connect very well. And probably they'll stay for, let's say, weeks. And these people and early stages of the conversation are more like an exchange of voice notes that eventually can turn into actual calls that are a lot longer.

Bhanu Pratap Singh Tanwar (FRND): So there are no voice notes. It is more like, it's very similar to this kind of setup that we have, right? It's like you are joining a live stream. And these streams can have multiple formats. For example, this specific format have two hosts and two strangers who have come and you're doing a pretty good job in breaking the ice for us, making us comfortable. And hence we are able to chat, right? These are multiple formats that we have on FRND. For example, there is one format where there is one friend, RJ, the host will break the ice for you, make you comfortable and break and will introduce the two strangers. And then there are some rooms where there are only two people. So users can keep on joining these multiple formats of audio rooms and they can explore different type of people, communities, conversations. So there are a couple of things which I want to add to it. So one is a nuance that language is a very big nuance here. So if a user is speaking Telugu and he is or she is connected to another user speaking Telugu, it essentially unlocks a great cultural context, as well as a great linguistic context for these two. So I think this was a big unlock for us. And the second thing is that even in the matching algorithms, how we identify people who are warm and ask them to or in a way prioritize them to welcome new users so that the ice breaking happens in the right way is actually also a catalyst for a new user journey so that when they come on the platform in the first five to 10 minutes, they really think, okay, this is a warm place on the internet. I like being here.

Kashish Sharma: Interesting. I mean, Bhanu, Harsh, you've been building friends for, I believe, seven plus years, right? That's an insane journey. And clearly, you know, like as a technology ecosystem, you have undergone a lot of like, you know, like shifts per se. And there have been a lot of apps that have emerged. And you know, like, some are still there, of course. Like, for example, there's a whole clubhouse phase, right, where everyone was exploring audio as a way of just kind of, you know, building communities, chatting, etc. What are some of your learnings that you realize now with the benefit of hindsight, that, you know, you've learned from and maybe incorporated in friends to kind of, you know, like, mitigate the risk of, you know, shifts that are ephemeral, or, you know, like taking good just product innovation, etc. And, you know, implementing them in your own product as well.

Bhanu Pratap Singh Tanwar (FRND): I usually focus on things which, what are the things when a lot of things, the world is in flux, what are few things which don't change. And I think for us, it's been identifying the pain point of a user and solving it in the fastest manner, in a business efficient way. I think this is something which has been Bruce's last 100 years and is going to be true. Majority of what I see us in terms of functioning is still the same at a meta level. Of course, there have been trends, I think harsh, you can throw some light on it with respect to the specifics.

Harshvardhan Chhangani (FRND): So I think this is something that we live by, I mean, ultimately, you will have to find a specific zero use case for which the user will come on your platform. And then you can always build something around it, right? technologies will come. For example, let's say we are seeing how JNI is evolving, right? But every founder must understand that ultimately JNI or the AI that we are getting is ultimately a tool. It is a fabulous tool. But the core principles of identifying the problem and finding solutions still remain same. JNI is not going to help you find, identify new problem statements, it will not create a market for you and then you are able to basically it is not a solution that will fit on a problem statement, the process of finding the zero use case and then building the solution around it still remains the same. It's just that with JNI, you'll be able to do it much in a much faster way. So these are some of the core principles we have lived by. Clubhouse also was a very good product. But we always felt that the zero use cases in like, for example, a specific user of a small town, who is having a very small social circle and is lonely in his life, not in a negative way, but because they don't have anyone to talk to was the specific zero use case around which friend was built. And of course, in future, we can have group formats where there are hundreds of listeners actively listening to different topics. But the identity which was identified, let's say the specific use case that was identified a couple of years back is something on which all of this is anchored.

Kashish Sharma: And now like seven years into the journey, how are you monetizing it today? And what is the monetization journey also been like over a period of time?

Bhanu Pratap Singh Tanwar (FRND): Cool. So with respect to monetization, I think we took a call back in the day that monetize this product has to be monetization first, and it has to stand on its own feet. So that was a very conscious call, which Harsha and I and sometimes this call in the hindsight, of course, I can say, you know, this was the case this but I think it was more like an instinct that this is not going to be a platform where you're creating time sync for users that come and just, you know, get lost in the infinite content stream. No, it's not going to be that which essentially means you'll not have crazy ad inventory. And if you would have seen in India, in terms of the depth of ad markets, it's still growing, it's not at a peak of what it would be in, let's say, us. So in a way, you need to now think of how do you monetize this platform directly from user with respect to subscription or something else. So this was thought back in 2021. And I think almost everyone at that time had question is why you're doing this? Why will you not just you know, max out the MAUs and create? We'll figure out monetization later. But we thought that no, that's not a fundamental way to go about it. Because crafting a digital product also has to follow certain rules, which I mean, it's not gonna be that the user would pay for something without understanding the value of it from thin air at some point in time, or you'll introduce a basic friction and the user would pay for it. No, a user has to understand what value this product is serving. So you create something, you ascribe a value to it. And then eventually, if it makes sense, the user would pay. So I think that's the thought process which started back in 2021, 22. And at that time, we felt the first ones to do it, the ecosystem was still not agreeing to it. But I think now the entire conversation has shifted to know that monetization should be done on in the early days, that's part of PMF. I think what Harsh and I felt that engagement PMF and monetization PMF are two different things. Your engagement PMF does not guarantee that you will have a monetization market fit. And I think that is how we still think about the future products that we would be launching. Another thing which we would add to I think the scale up has been cool in the sense last financial year, we did around 200 crores of revenue, we have been profitable. So the business has stood on its own feet, and it is growing sustainably. So that's something which we believe that in India, particularly because our GDP per capita is still, you know, in a medium range. Digital products would be amazing businesses, because you don't have any cost of delivery per se, and they are inherently high gross margin. So we believe that this is going to be a space which would become crazy big in the next five to 10 years. And an added opportunity to it would be if we are able to export few of these products internationally.

Harshvardhan Chhangani (FRND): I'll just add one more point here. I think this answer given by Bhanu also ties to the previous question about this last seven years of learning that we have had. So it is a very important point. See, ultimately, founders also have to understand what kind of product they're building. For example, interactive product like FRND is a very active platform, you will have to put in an effort to speak it is not a very it is not a passive platform like a content consumption app. So the ultimate matrix being changed and the business that will be built around it will be very different. And that open mindedness has to be there in the founder, we cannot be hung up on the idea that every business like this will be around ads. So you will have to have that open mindedness that because it is fundamentally a very different kind of product and the user behavior is very different, the nature is different, you will have to find out new ways to create a business model around it.

Kashish Sharma: It makes a lot of sense. I have to ask, when users come aboard on the platform, are they just looking for friends? Or is the intent and motivation, you know, some sort of retaining use case, etc? Have you seen something like that kind of, you know, like people kind of supplement because you do have guardrails around, you know, like you said, respect points, etc. So how do you kind of curtail, you know, such, I don't know, you know, such events and actions on the platform to ensure that it's a safe environment, we're safe and suffering, that's a very big USP of your app as well.

Harshvardhan Chhangani (FRND): So, so the platform is not designed to date, because it's not a dating platform. People come here for they come for entertainment, they come for conversations, they want to be part of a community. And they will stay for belongingness. Romance is something that we leave up to the users. There have been cases where people have been married because of FRND app. There have been cases where people have found real helpful communities that have really helped them coming out of some very serious situations in life. And because the community was helpful, they were able to survive that particular phase, right? Ultimately, what we try to do is we try to make sure that since people are coming from a very different kind of background, a lot of innocence is also there in the users like they don't intend to do any bad stuff on the product. So we have designed the whole trust and safety mechanism using community, basically the other users that are on the app, the product nudges, so we keep on priming the users what kind of platform this is. And then of course, with the help of AI, we are able to like put in a plug in a lot of tech to make sure that the overall trust and safety is on point. We are able to create that community which is very conducive to these kinds of conversations where people are able to talk about, let's say, if someone is shifting to a new town, coming from a village, they'll be able to find a community on this platform, they'll be able to talk that they feel lonely on this platform, right? So this is something that we have built and essentially, it is very similar to like we keep on talking about this example that metro stations in India, they are very, very clean, like, and the kind of users or the passengers who come to metro stations are very similar to let's say, railway stations. So though the situation has improved, but there is a stark contrast on the cleanliness level that is there because from day one, metro stations are kept clean, right? And that is I have a theory. By the way, which is like metro stations are clean, because they never allow that first pile of garbage to ever form. Because God forbid, if that forms, next thing you know, the whole platform is going to get littered. So if you don't allow a single instance of littering, you know, littering will not happen.

Dhruv Sharma: Guys, thank you so much for coming on and chatting about the story. I think we have one quick final question for you before we move on to our next guest, which is, if someone is in the same shoes as you guys like seven years ago, and is trying to raise, you know, an early round for an NBU story, what would be your advice to them? And then, I mean, maybe let's just address this question to the seed founder. Harsh, do you want to go first?

Harshvardhan Chhangani (FRND): I think I'll go back to my previous answer that you don't have to optimize for fundraisers, you will have to find a good niche use case. The social media space is very big, the time spent by Indian users is humongous on platforms like Instagram. And if you really need to create something, which users really love, you will have to find a specific use case that the current platforms are not solving. And we feel that is how the next set of social media platforms will be built around specific niche use cases. And then people will socialize over it. One such inspiration is Strava. Strava is a very specific use case of, for example, running. So that utility is there on which there is a social graph being performed. And there are profiles, people are following each other, because they were able to find that specific use case, which will never be catered by a platform like Instagram, because it doesn't serve the purpose, right? So my message would be that don't optimize for fundraisers, rather optimize or try to find out all these niche use cases, which are still unsolved.

Bhanu Pratap Singh Tanwar (FRND): Yes. So we, quickly, I'll just add, so we believe there are going to be a lot of social networks, which will come out of India. They need not be 1 billion plus user kind of stories, because those are ad networks. These will be much smaller utility driven social networks, which would be monetized upfront. And there are a lot of such opportunities. A good insight would eventually lead to a fundraise. So the insight is what you should focus on. Yeah.

Dhruv Sharma: Well said. And again, thank you so much for coming on and sharing your story. All the best.

Bhanu Pratap Singh Tanwar (FRND): Thanks.

Dhruv Sharma: Great. We're now going to say hello and welcome our next guest for today, who's Devender of TrueFanAI. Devender, welcome to TON.

Devender Bindal (TrueFan AI): Hi Dhruv. Hi Kashish. Thank you for inviting me guys.

Dhruv Sharma: Our pleasure. Did you gatecrash Virat Kohli's wedding? Is that a true story?

Devender Bindal (TrueFan AI): Actually, my co-founder did that, Nimesh Goyal.

Dhruv Sharma: Okay. Tell us more.

Devender Bindal (TrueFan AI): It is a true story. He had a like a office meeting there in that building. And then like he got to know there is a Virat, there's like Rohit Sharma wedding going on in the same building. So somehow he managed to pass the guard and went inside. And then he met the all the like cricketers like Rohit, Virat. And he mentioned like, I'm a big fan. So Rohit gave him a like selfie, which he forwarded to everyone next day.

Dhruv Sharma: But I think it does raise an important question, which is what should the relationship between fans and stars be like? There's a version of it that, you know, everyone knows from Bollywood, but I'm sure there's more to it. What are your thoughts on it as a company?

Devender Bindal (TrueFan AI): So we started our like we started in 2020 with the idea that India has a lot of fans like which wants to interact with the celebrity and that cannot be done because there is no way like scarcity is there. Celebrity time is very important. So we started solving it using the technology. So we thought, okay, we will create the, at that time, it was not called it, it was called use the technology to solve this problem. Like fan and like celebrities can interact with each other and fan will have their like to fulfill their dreams. So we started with this vision in 2020. And then like we added like a lot of celebrities on our platform. Fans also came. But then since last two years, we pivoted to the B2B mode, like we got to know like there is like a lot of application of our same technology in the B2B where we started with the Zomato. And then now we have like I think more than 70, 80 enterprises in India and globally where we help them create AI content of celebrities or any person whose time is very important. So we take two minutes of their time, create the avatar. Using the avatar, they are able to create like end to end ads, which can be like shown on the TVs, they can be run on the social medias, like Google, Facebook or any other channels, then they can do the like translated in different different languages.

Dhruv Sharma: To be clear, is this done with the public figures consent?

Devender Bindal (TrueFan AI): Yeah, the brands takes the AI rights from the celebrities, they give their rights to us and then we create the avatar and then we help them like create the ad and every video which gets generated is getting reviewed by the celebrity team.

Kashish Sharma: So is the value prop because of course, as you mentioned that the brands are getting the rights from celebrities. So is the value prop that the production cost, the production time and time to market of the content gets compressed because now you're using AI to churn out the content, not faster.

Devender Bindal (TrueFan AI): So like the time to go to market is very low. Like previously, like people, teams used to think, okay, we have to think of a script, then we finalize the celebrity, then there is a shoot will happen, direct makeup and all this. But with us, nothing is required, you just give us the script, we will create the video, celebrate, and see it, approve it and then it goes live. Second thing is the difference between the shoot and the videos which get generated by our platform. There is no difference. So you will not be able to know if it is generated by us or it is like actually shot. Third thing is you can translate it to different languages, all the big banks like SDFC, Bajaj, Zipla, Intas, Goibibo, Zerodha, Azomato, they all have users from different dialects, right? Celebrities or like a lot of people knows one two language, right? Now, brands wants to communicate in their language, in the customer language, because it will make them understand more, it will increase their ROI on the ads which they drive. So it helps there in that sense. And the third thing is they can do it personalized communication. Let's say like you're booking a flight from Delhi to Mumbai, you drop out, brand can send you a message. Hey, hi Kashish, hi Dhruv, you were trying a flight from Delhi to Mumbai, you dropped out, what happened? Let us help you with some kind of coupon code or something. So all these use cases and use cases are not limited to just marketing. There are a lot of learning and development use cases are happening where brands has to initially they used to give a lot of text content to like new joinees. But now all this has been moved to the video content. There is a like a lot of our clients who translate the educational content to different different languages. Okay. So they are like wherever you see like the utility of video is there. So that utility is getting enhanced because you don't need to text, you can move everything which was being done in the text to the video now.

Dhruv Sharma: And Devender is, can we call you Dev by the way?

Devender Bindal (TrueFan AI): Yeah, yeah.

Dhruv Sharma: So they've, are brands finding that the ROI on these localized personalized advertisements is way better than anything generic? And before you answer, Kashish, do you know what's my wish for equity list for 2027? That you guys use, that you guys use Trufan to create, I mean to get Steph Curry ads to expand to the US.

Kashish Sharma: Yes, I was thinking too local, but Steph Curry, Devender can, I mean, but we have to procure the rights for it.

Devender Bindal (TrueFan AI): Yeah, definitely. So you have to like discuss it that way. So in terms of like the brands see a lot of good ROI because you understand like when a brand spends, let's say a hundred rupees on marketing, five and 10 rupees is the content creation cost. 90 rupees is the distribution cost. We are saying spend like five instead of five, spend just six, seven rupees and you will get ROI increase on the 90 rupees where like the, your customer will understand the better, whatever you are trying to communicate with them. Whether it is an ad about your new product, new face you are bringing, new services you are bringing, new features or some kind of offers, whether it is. So whatever you are trying to communicate, communicate in the user language and you will feel great. It's just like, you can assume like when Kattappa was like any South Indian movie got released in the dubbed versus they actually make the movie in the different languages. You know, the experience is very good.

Kashish Sharma: And is your use case only limited to celebrity, you know, like generated videos or are your customers using them generically for internal purposes as well?

Devender Bindal (TrueFan AI): Can you repeat like I lost for a second.

Kashish Sharma: My question was that, is your use case currently only scope to celebrities, you know, like videos with celebrities or are your customers using you for like internal videos as you mentioned?

Devender Bindal (TrueFan AI): Yeah, internal videos are being generated for learning purposes, for HR purposes, like for like professor videos are being translated to different languages. So technology is give us any content like two, three minutes of video content. We create the avatar of that person and then you can generate as many video of that person. You don't need to record. You don't need to, let's say, translate to different languages. Everything will be done automatically.

Dhruv Sharma: All right. So, Dave, you know, in 2024, Anil Kapoor was in the Time magazine for, of all things on earth for AI because he felt very strongly about personality rights. Can you talk to us more about how these public figures feel about personality rights and how platforms like you, like, what are you guys doing to prevent the deepfake menace?

Devender Bindal (TrueFan AI): We take the, when we have the right, like we make sure like we are properly compliant and we are ISO compliant. We are SOC 2 compliant. We make sure like the team who is working and handling this has the very specific data. They don't have the data of the access to anything else. And we make sure whatever goes out has to be reviewed by the team. If they agree, if they allow, then only it will go out. If they don't allow, then we say, okay, it is not like brand can generate any content. It's like first we create and then celebrity reviews it, then it goes to the brand. So there is no scenario like where deepfake can happen because deepfake is something which is not accessed, like which is not authorized by the celebrity, but here everything is done with their permission.

Dhruv Sharma: Can you help us understand the scale of Truefan as of today? Right. How many customers do you work with? Are they usually enterprises or you know, companies and generally the videos as well, right? Like what kind of, how many millions?

Devender Bindal (TrueFan AI): So to give you like, we have like more than 80, 90 enterprise customers in India. Okay. These are all big brands. Like generally they are B2C facing brands. Okay. So B2B2C, we help them to the direct customer in terms of videos. We generated like around 20 million videos in last one year. Okay. The models which we use to generate the videos are owned by us. It's not like we use any third-party models because the model are used by us and we know the complete tech, tech stack as well. We own the tech stack. So our cost is also lower than our global competitors that like Synthesia and Agent, like our cost is low. So this is the scale right now in terms of scalability, like we were the first company to generate 5 lakh AI videos in a day for Zomato. Okay. In May, 2024. So till then, like we have came a lot ahead and right now, like we are working on some few new products as well, which will come in the market and, but video creation of our face video creation is the main thing which we do.

Dhruv Sharma: And speaking of new products, you guys also, do you have video agents in the pipeline as well?

Devender Bindal (TrueFan AI): We have video agents in the pipeline. It will be part of our studio, but in terms of a big product launch, we are working on something called real-time avatar, which is like something, suppose right now we, I have to be present in this call later, like maybe once we launch, my avatar can take this call as well. Okay. It will be that seamless. And then there's another product image to video, which we are working on. Okay. Yeah. That's like, we are scaling well in terms of existing times.

Dhruv Sharma: And that's such a fascinating thought, right? Like say we have, this is TON 113, say when TON, I don't know, 257 happens next year, all of us send our avatar to attend on our behalf. I get that the avatars might look like us, that they might sound like us, but can they also think like us?

Devender Bindal (TrueFan AI): No, no, they cannot. Obviously they cannot think like us. It's like a driven by LLM and the LLM plus you like your memories. It's not about like identity we are trying to replicate. It is more about like, how can we solve the use cases? Let's say in the banking, you don't need to always talk to a person. You need to know like a lot of your other detail. What is your insurance policy? What is your nominee details? Some other process you want to change. So for that, you don't have to wait in the line to talk to the actual person. You can have the avatars there. You can talk to the avatar and a lot of your queries get done. Let's say you're coming to a hotel, right? You have to check in. You have to just ask, okay, where is the pool located? What is the time for the breakfast? I have to check in like I have this request. So few of these things can be done by this agent.

Dhruv Sharma: It's a predetermined workflow. There's a script and that's when they do best.

Devender Bindal (TrueFan AI): No, no, it is script is not required. Okay. But it has some guidelines has to be there like what like what brand wants to communicate in what scenarios there like some things which initially like in every agent like whenever the agent starts, it can agent can go in any direction. So we have to make sure like there are guardrails which for the safety and compliance purpose for the brand and which actually which communicates what brand wants to communicate and how much about their actual product. It cannot go like other way around. So it's more about like increasing the efficiency of companies, whether in our hospitality, whether in the banking and there can be a lot of other use cases.

Kashish Sharma: Interesting. I mean, they mentioned that you're creating your own models. And of course, there are some global counterparts. I'm curious, right? As a as a customer or a sales prospect that is benchmarking, you know, true fans video quality was some of the possible vendors. You know, what are some of your learnings or some of the spikes that have, you know, like just gotten you to win a lot of those deals and get their comfort?

Devender Bindal (TrueFan AI): So when people see, OK, the scale which we work on is the most comfortable comfort they get, like versus working with us, like when they go, OK, you can generate, let's say, 100K to 500K videos in an hour. OK, that gives them a lot of comfort in terms of scale, because sometimes like people think, OK, can you do the scale generation at the speed which they want? OK, that is one comfort. Then the second comfort is when they see, OK, these many big celebrities, OK, given the approval on the content, there are a lot of like there are some companies in India and globally where they have tried to convince the celebrities, but the quality of their model was not good and they didn't get the approval. And that brand has came to us again because celebrities said, OK, true fan is a brand which we want, which works and we like their content. You have to talk to them, see the quality which they produce. So this is something like which brands like a lot. And the third thing is they know like we are a very like working in this space since last six years. So it gives them comfort. We can work on the scale. We have the quality of the models. So we are happy to go to extra mile for the customer satisfaction as well.

Dhruv Sharma: Great. Yeah. Have you guys announced your, I think your A round last month, a 10 million series A led by Bering. Talk to us more about that round. Where's the money going to go? What are you hoping to achieve over the next 12 to 18 months?

Devender Bindal (TrueFan AI): Yeah. So we are, we want to go in like three direction. First is like we will scale our current existing products in terms of we are adding the team, we are adding the sales team. So that revenue will increase the current product revenue. Second, we are adding new products as well. And the third thing is we are adding new geographies. So we are adding like Middle East, Southeast Asia, and obviously like us is the primary market we want to go. So these are the three main things which we want to do with the new countries.

Dhruv Sharma: All right, great. Those are all the questions I had. I don't know if Kashish has, has any closing questions.

Kashish Sharma: So I mean, I'm just, I'm just curious, are you at any point of time going to provide your APIs for let's say a creator to plug and play at any point of time?

Devender Bindal (TrueFan AI): Yeah, we have a like a studio which is live, like creators can use it. But mainly like our focus is on enterprises. So it's a different market, like either you're focused on presumers, which is like creators, and versus the enterprise. So our product is more enterprise focused right now. But as a creator, like you can come you can try our product.

Dhruv Sharma: You know, I think we should, you should definitely explore this for dubbing in different languages at least.

Devender Bindal (TrueFan AI): Yeah, video translation, you will love it. If you face an issue, obviously, like, we can help you but still love it.

Kashish Sharma: I think we'll try this over the weekend.

Dhruv Sharma: That's great advice, Kashish. Dave, again, thank you so much for joining our show. Congratulations on the raise and all the best for what's ahead.

Devender Bindal (TrueFan AI): Thank you. Thank you. Thank you.

Dhruv Sharma: Thank you. All right, listeners, those were our guest segments. We do have at least two pieces of of news, which Kashish and I are going to chat about. Kashish, there's a there's a yet another new Frontier SA that everyone's talking about. Have you seen it?

Kashish Sharma: I have. I mean, you can't really scroll past x long enough to not come across some sort of a and bit of honesty, I think you you help kind of bring it up to our attention.

Dhruv Sharma: Right. So that's all. Yeah. So this one is a breakup. Yeah. Yeah. This one is by Demis Hassabis of DeepMind and of Google DeepMind. And it's titled a framework for frontier AI and the dawning of a new age, where he makes a case for defining what the frontier really is. So first of all, having like a standards body, which is kind of like a self regulating organization, public private partnership, it's got, you know, it's it's basic to use a model term, it's got a mixture of experts who define what the frontier is, and then use benchmarks to give frontier status to a certain model and the lab behind the model. But once you earn that frontier tag or moniker, then you have to act like a frontier, you know, lab in a frontier company. And, and, you know, with what's the super, the Superman or the Spiderman line with great power comes great responsibility. And so when it's, you know, when you're talking about really high risk domains like cyber, and bio risk, there'll be continuous evaluation of these models to see that they're not, you know, there's there's no unintended behavior or consequences. So that's what's there in the essay, the best part about the essay, Kashish was that, you know, typically anthropic, open AI, Sam Altman, Dario, they're always exchanging blows with each other on the internet. But for some reason, Demis, he's also Nobel Laureate, I think his essay was very well received by everyone who's a who's a key stakeholder in this, and hopefully it sets things in motion.

Kashish Sharma: 100%. And also have you realized that whenever, like, for example, this essay came about, and I believe, like, for the past few days, there have been a lot of news articles as well, right? Like, for example, Satya Nadella recently, I think he wrote a blog about, you know, like, even just consumers being very cautious about which frontier lab actually owns the data, and what are they really giving away? Yeah, you know, with respect to the value, the value that probably is not perceived as of today, but, quote, unquote, could be kept as hostage, etc, as IP, as you know, like, you know, as time evolves. So it's just like, isn't it weird, like, everyone is kind of, you know, talking about cautionary measures immediately over the past few days? Like, is it triggered by something? Because, yeah, it's not like it's not been discussed before or not been, you know, aware of,

Dhruv Sharma: I think, and also, when I chat about this often, Kashish, I think, because everything at the frontier is happening all at once, people tend to mirror each other's moves. Sometimes I think everyone keeps their essay ready. And once someone publishes, you know, they hit publish within seconds of each other. So there is that. I'm sorry, I think you did have a question in there, which I seem to have, which has seemed to have lost track.

Kashish Sharma: You've kind of answered it, right? Which is, I mean, yeah, no, basically, as you rightly said, I think people are just trying to get their word out. And it's pretty important, I suppose. Let's see. Let's see how well, I mean, does this age like fine wine? And what essentially happens to this? I think that'd be.

Dhruv Sharma: Yeah, I, yes, you, you, you brought up Satya Znadela. And I think that's a point that many people have come to accept, which is okay, rent the model, but own your data. You know, it's okay for you to rent frontier intelligence, but don't start, you know, renting out your data. Keep it within the context of your organization. So there is that. And there was, I briefly saw this clip where Shamath went on, he was speaking with, what's his name? I think he was in Squawk Box, CNBC, etc. And he made the, you know, AI oil analogy. It was an interesting analogy, actually, Kashish, where he said that, think of all of these frontier labs as just selling oil. And everyone seems to have a different price per barrel. So what's going to happen six months or nine months later is there'll be one Fortune 500 company that'll miss earnings. And then the CEO is going to look at the CFO and say, hey, where did all of the, where did these incremental costs come from? Because down in the, in the trenches, people are token maxing all the time. And so that's when they, when they, you know, dig deep and find out that they'd been paying, use Shamath's analogy, like they'd been paying $50 a barrel for a barrel of intelligence, which let's say loosely defined as a million tokens, when they could have also bought the same barrel for $1.50, then that's when everything comes crashing down. So I guess everything's happening at a breakneck pace. And every once in a while, people get cold feet, but then they give each other confidence and decide to bash on regardless. I think we're going to see a lot of that happen.

Kashish Sharma: Do your credit, I mean, the Shamath parallel that you've, you know, the analogy that you just shared with the oil industry, the regulated or pseudo regulated standards body operates nothing like OPEC. And start setting a price cap on these things. That'll be pretty interesting too, if that happens.

Dhruv Sharma: And then the second bit of news, Kashish was Elevation announced Fund 9, which is like a half a billion dollar fund. You've got thoughts on that?

Kashish Sharma: So from, of course, for our users as well, for our viewers as well, this is predominantly for the early stage bets. And I think they made it explicitly clear that you have to look at this fund in conjunction with the previously announced late stage fund of, I think, roughly around 400 million. So cumulatively 900 million, covering the entire spectrum of early stage to late stage. A, great stuff for early stage momentum and investing horizon for sure. B, I mean, for example, like clearly there are a lot of announcements that are hitting the PR cycles over the last, you know, just in general, over the recent times. Do we see, or would we see a lot more early stage investments, right, in India especially? Is that a telltale sign or maybe larger checks that are being cut out? Right. I mean, that's just like, that's at least one thing that you're going to look forward to.

Dhruv Sharma: Right. No, it's great. And I think it brings, it makes Elevation now the second largest venture fund in India after Peak. There's this, there's a chart that I'm guessing we're showing right now from ED Tech, where I think Peak has the most amount of money at their disposal to deploy, followed by Elevation and then the Axel India franchise and Nexus and then everyone else. So congratulations guys. I mean, they of course have a very storied track record with Paytm, with Nisho, just so many wins under their belt.

Kashish Sharma: Good for the ecosystem. A lot of, I think, great early stage bets, right, over the last few years. Murph.ai, Composio, Grey Labs AI within the FinTech voice space. So yeah, I mean, Elevation is just one of those storied funds, right, that you do expect great winners and self-prophecies kind of playing themselves out. Yeah. So, I mean, fresh capital being induced, you know, partners cutting checks, that's always good for the ecosystem at the end of the day.

Dhruv Sharma: It is, yes. We had Grey Labs on the show as well last year. The other two, we'll go find them and find a way to bring them on.

Kashish Sharma: So sorry, I don't even want to close this up, but isn't it worth noting that this year there's been some insane, like just announcements per se, right? Like Metal Capital, a lot of, you know, a lot of spinoffs from tier one funds per se, you know, like just a lot of newly announced. Those are not, you know, they're not small funds. They're massive funds. Like Metal is big. Ambition is big. Again, I don't know if Ashish or TJ or anyone listens to this. We'd love to have you on the show as well to cover what you guys have going on in the kitchens, what you're cooking. But yeah, it's been almost like a, yeah, like a huge shakeup and re-rating of the entire venture landscape in India, as is to be expected. Yeah. Out of curiosity, are the performance benchmarks for their last funds performance with respect to DPI and MOIC, or is this mostly premised on, of course, you know, the list of companies that have been public, Misho notably, etc., right?

Dhruv Sharma: Well, there's a handful of funds that publish them voluntarily. As for the others, they come out in the form of leaks here and there when, you know, somebody lays their hands on them. But no, I think it might be an interesting idea to cover that as a story on T1 in a subsequent episode. Sounds like a plan. All right. Great hosting this with you today, Kashish. We'll both be back again on Friday, listeners. Thank you so much for tuning in. Have a good rest of the day.

Kashish Sharma: Take care, folks.