₹1 Lakh To ₹1 Crore - AI Business, Solo Founder & Business Growth - Mukund Jha | FO460 Raj Shamani

34:45if we are headed towards you know the agi and everything right i think some of these things will get very very

34:56out of control which we can't see right now what's your biggest worry i would say probably a rogue agi right now explain AGI and then Rho AGI AGI essentially means when you know the I mean everybody has their own definition my definition is that when a model or a general AI system can actually do most of the work that humanity can do today knowledge work right and

35:27it's sort of sentient right and it happened right recently Gemini improved itself I mean those are very minor but like it's a start it's a start to a direction that a large language model improved itself without any instruction anything just going on and on this race of becoming better yes there are a lot of such cases like Anthropik published that you know that it was an alignment study in which like the model that I'm testing and he said to just pass that test so it had it had knowledge that I was being tested and was able to fake that test. Right. So I think that is a big risk right now. Nobody knows what is on the other side of, you know, AGI, ESI.

36:12Are we in a Terminator scenario? Are we in a Matrix scenario? And like, we would have thought that this is like a very

36:23fictional thing. But people who are closer to the field, like they truly understand that, you know, it is a real possibility. that at one time computers and this AI computing whatever technology will become way more powerful than humans and they will take their decisions so they will run today we control it because we ask them to do certain things and they do the task tomorrow we don't need to do that they'll do it themselves that's essentially the age of AGI right yes yes and then you said rogue AGI what's your worry with rogue AGI if it's not aligned with humans like it thinks what can it happen I mean doomsless scenarios like AI is optimizing for creation of factories and robots and clears out cities saying that I don't have a factory right it could be that it manipulates humans to believe certain things and you know you know expands.

37:30I think all the bad movies on AI that we have seen all the things remotely possible are. So rogue age is meaning that folk influence a device, a person influence to clear out the city and then you're gone. Maybe. Yes. It's possible that AI once it decides that humans don't need to be able to i mean again distance possibility but i mean that's why alignment is very big research right now and abhi openia has hired like preparedness readiness uh you know function how far are we from this dunes day where ai decides that we don't need humans

38:21very low probabilities and but could be anywhere between you know I mean it keeps changing now the estimate is changing now like next year we'll see the next big pre-training happen now the GPT-4 model which is pre-trained is going to be on the same next year now a lot of computers becoming available this year so we'll see like the next step What's your guess? I mean, I think we'll have advanced AI, like closer to AGI, I think like in five to seven years. That is my Lycraff estimate right now. Crazy. Yeah. What are we talking about when we're making a business? No, I think things are going to move fast.

39:07And people... What's long term in this then? Because if everything we're talking two, three years, two, three years, business banao bach ho paise banao sab 2 saal mein tum career start karo khatam hojao 2-3 saal mein like koji skill seeko that skill khatam hojao ki 2-3 saal like everything is just somehow come under like under 4 year cycle so what's long term like do you think the long term long term is only next 6 months nahin nahin I mean I feel ki so capabilities will start emerging right it'll take time to permeate through society there will always be there is obviously we some you know and again this is again extra politic it could be that we have so there is this concept ki fast take off slow take off right slow take off manlab ki technology aya ga but bhot time laga us ko you know permit honne mein and like

40:03I mean the outside chance is ki agle pang chai saat saal mein like we will be on the cusp of technology you know which is supernatural right and I think people are talking a lot less about it than they should because if I agree like where Manolo you know like if that is a possibility I mean beyond that horizon you can't see right now what happens right and you know but at the same time I think like who gets to AGI first is also important right and like who has control over it is also important right and it could be like it is distributed right and used for good it could be that it is military controlled government controlled I mean there will be scenarios like that that is going to play out and like at least in probability space you know I'll give it some probability you started the conversation with something and I want to end that conversation with again understanding the thesis you said cycle right Right, right.

41:12Now, the countries are very powerful. And now, we are going to that place, in that world, where corporations and companies are more powerful than the countries. Already, some companies are more powerful than the large majority of the countries. Right. And then, eventually, individuals are more powerful. Right. Tell me, explain. Right. Are companies and countries more powerful? Economically, right. For example, Nvidia today, right.

41:41literally controls the future of AI in many ways. I'm just saying church or state was more than before. It was clear that church was the authority and it was then there were ideological differences and then people sort of went there. And generally what I've seen is that historically there are these cycles. right like where something will it could be that you know like religion is is the next thing that comes up again right like people want faith because everything is artificial and like the only thing that i can attach to myself to is is faith right maybe religion makes a comeback right so things move in cycle right and at least the way sort of i'm i'm um zooming out when i zoom out and and see and every like markets have cycles you know like seasons have cycles so so cycles are everywhere and and if you want to predict what's going to happen in future like you'll have to intersect some of these cycles that, okay, you know, what's going to happen?

42:38For example, we are in a massive technological cycle right now. So technology progress is going to continue. You know, we are also in a massive, I would say the trust is eroding generally, right? So you'll have to see that we were in a high trust society, we're moving to a low trust. So some societies, like for example, India was a low trust moving to a high trust, right? So I think generally the way I sort of try to predict, you know, the future, what's going to happen is look at the world in this, you know, like which cycles are we on right now? And I don't know if it's going to happen in like 10 years or 50 years, but it is clear that, you know, like nation and governments, right?

43:17You know, you can see, you know, what's happening across the world, right? There is a, like a natural sort of, you know, threshold to that, right? Where, you know, obviously like companies are going to get bigger, become bigger, right? And then, you know, government wants to intervene. At that point, if somebody is controlling, you know, like a really strong piece of technology, right, like what do they do, right? So I think like those are things that are, you know, like quite... One skill that will become absolutely worthless in the next two years. I mean, the way I think about it, and I'm also nuancing it, right, is that the middle will disappear, right?

43:57But let's just take an example of design, right? do you think design like of course AI is gonna get getting better you know you design but the taste right you still require that you know like that 1 % human on the top who is actually able to direct and give the right input and you know like generate a better taste will have a lot more leverage right because they can do that and obviously at the bottom you'll have this layer where like 99 % things are working but that 1 % is not working you'll need like a lot of people at the bottom to sort of fix that right your website makes a little bug that someone will fix it and at the top I think the I feel the middle is going to be occupied by AI completely right and people who have good taste people who are good at problem solving like their leverage will increase multifold like one person will run you know like millions of dollars worth of company on his own because he can direct most of the thing and then you'll have a bottom layer which is sort of fixing a lot of the AI issues or training the AI, right?

45:06A lot of the work will become that you train. One of the thoughts I have is that

45:16people will... In every generation, what happens is that this is saying that your work today is going to feel like play tomorrow, right? So let's say if you, let's say, go to your great-grandfather and say, hey, you know, my job is, I sit, you know, chat with interesting people. Right. And that's how he can go in. And he'll say, yeah, I was doing so much and I was going to go and go to the house and I was going to do it. Right. If you go and see somebody who's a gamer today, right, and they'll tell their great-grandfather that, you know, I sit on the computer and play games all day. That's how you make money.

45:50Right. It's going to, to them, it's going to feel like, hey, yeah, this is a play. All right. I think the same thing is going to continue. I feel that a lot of people, to them, the work is going to start becoming like play. Maybe they are just playing games to train AI. Maybe they are just looking at content and training the AI. I think a lot of the people's work will be just helping AI get better at things at the bottom layer. And the top layer would be to direct the AIs. Interesting. And I feel low. like one of the things i can be wrong here i'm just saying it from a very non-developer non-techie side of things which is don't go like don't don't put me on the spot for this but i genuinely feel that people are obsessing over this one ai skill which is prompt engineering okay i think it's gonna die i don't think it's like that important going on like it's just the basics and the fundamentals people need to learn they need to learn how to be more clear and more nuanced in order to give instructions right they don't need to do these like giant 10 day courses for prompt engineering and just obsess over it or her cheese basic cheese be to prompt engineering framework i think all of that is like crap right like in fact i feel at least my experience have been my my models give me much better answers when i'm much more casual than when i'm trying to force like just putting a framework of doing this and that.

47:25Right. Yeah, I agree. I think, you know, like, I mean, we are seeing that on Immersion also, right? It used to require you to prompt in a very detailed way, right? And like not even simple prompts are, you know, like working, you know, as good. Because, of course, there is like this, you know, back and forth of, you know, what has AI understood so far about you, right? and

47:51yeah I think all skills are going out of fashion in some way

48:00because middle AI is going to keep eating and I think the race is that are you the one who is directing the AI in the right way and

48:12I think I also think that where things get stuck is that AI is not able to understand. You don't understand. You have to say something vague. So prompt engineering is helpful there. That you know, it's vague. And is it able to get content? As memory improves, is it able to get a lot more context? That sort of thing will simplify. But till then, I think prompt engineering is quite relevant till then. In future, I'm saying today it's relevant. Today people get more results from it. I think in some time, I did a couple of courses and I'm like, this doesn't, I mean, it's great to learn, but I think it's going to be relevant at some point.

49:04Maybe two years, five years, it's not going to make sense. Like the obsession and the intensity which people are learning from, it's I think it's going to disappear in a while because more and more people are going to build tools and things like yours which and your obsession at one point or like open AI's obsession or anthropic obsession will be that how can I make it easy for a very normal person how can I make the system idiot proof right and because you're going to be obsessing over that yeah so you're definitely not going to obsess over how do I reward somebody who's like an excellent prompt engineer you're gonna be like

49:45so if you guys are successful which it looks like you will be yes definitely so then prompt engineer I think the role of prompt engineering reduces but again then there'll still be like better prompters like who would probably command 10x more leverage right you know because they're able to unstuck system somewhere so as I said like my mental model right now is that middle is going to get hollowed, right? And at the bottom layer, like, people are sort of training the AI, right? Mostly, you know, correcting mistakes. And top 1 % are builders. Top 1 % are builders. And I don't think that, and the top 1 % can be a large part of humanity.

50:23I don't think it's the 1 % of humanity, right? I think the amount of work that we need to get done, right? Like, if we have to go to stars, you know, if we have to solve every single disease, if we have to solve, you know, teleportation, right? The amount of work that just humanity needs to do, right? Like, is insane, right like i mean so you know like i mean you know this is the time to sort of you know imagine that we can go to next galaxies right find new theorems solve quantum computing solve quantum mechanics but if let's tell me if app banana building gets easy app now you can make an app normal everyone can make an app right i have played with your tool a lot so if I did it what gets hard if building app and building software becomes easy what's gonna be harder now so if you look at you know like because if everybody can build yeah right everybody can build app today somebody earlier the moat used to be which has made a lot of software and a lot of apps now he will get funding and things will get bigger.

51:34Now everyone can make it. I'm sure now not, but in 2-3 years, 5 years, it will be when everyone can make it. If you can make it. So when you can make it, then what's the hardest part? The hardest part will be finding the right ideas, distribution, taste. And again, I think, I think that we are very constrained in the process of zero-sum game. like finite resources right the need for software is infinite like you know like you will have on demand software right like you you know like we we were talking about this you know like Christianity and other things right like you know we would just prompt and say like you know show me a dashboard of you know top religion and history and those kind of things right so I think the the I mean the personalization of software like every business needs a custom software right I feel that one thing is for sure that teams are gonna get smaller which will be built in 50 people in the team.

52:33You will build... Like,

52:37in the emerging way internally, there was a big debate. Asana used this task management software, right? And, some people want to use it. And, it's a mess. We won't use it, Asana won't use it. And, somebody actually internally just wipe coded a new project management app. And everybody moved to that now. And, so we gave up on like$2 ,000 per month of subscription of Asana. And built this entire app internally. and entire company has moved to that now right and so you'll see a lot of that happening that where and and because the reason was ki jira you cannot have sprints lower than a day or a week we should have three times shift in the day so we should have been in the morning evening night slot and no tool was supporting that and this was like a you know a senior QA engineer who just built it on his own because there was this you know like commercial going on internally and we gave up on the subscription we were paying$2 ,000 a month and we are seeing this across the org like where our marketing team how are we tracking this how are we doing that but is it as effective as Asana?

53:39it is actually pretty good for us because we don't need so many things and software is in our control it is evolving the way we want it because 90 % of software gets built like a lot of people use only the 20 % the basic and a lot of it is bloat or you need a lot of people to come in and customize because now people will be able to customize and every company every system will have their own personal needs so I think my thesis is that need of software is going to continuously increase people will increasingly get more and more dynamic on demand software I also think the nature of software is going to change because a lot of it will be just an instruction to an agent you may not even need a software to do that here's my worry here's my worry so because I'm moving to Asana and I know this I've worked on this tool before and now the company is getting bigger project management is becoming a pain now you have to do right we're going through that transition and we were like click up use Asana use bunch of other things we're going through the same tool right I like Asana because there's nothing more simple right like click up is probably better for customization but I don't have that level complicated i don't want it at least for my work and then my worry is if i build something asana i don't have to think about is it getting better is it getting worse company is taking care of it right but if i will make it myself then it will be the problem um initially maybe yeah right and software maintenance is also going to get very easy.

55:22Just have and deploy we take care of your scaling we take care of all those things. Right. You can come and customize it as you want. Right. And you know you'll have different you know versions of it that you want to run. And initially obviously there'll be some you know hiccups you know like you're maintaining sort of maybe somebody else is going to build like a you know a simpler cheaper asana right. And say hey this is task management software for you know people who are running a podcast business right or a media business right and like this is like really really tailored for you right maybe you'll buy that custom software versus buying any other big company right which is so Asana is great but and it's expensive it is expensive and it has a lot of restriction like why does Zira exist because Zira is a lot more complex like you can do a lot of things like you know divide teams you can you know manage sprints and other things right and so there's obviously like you know a difference between like zira which is a lot more complex uh and and but i think there's this this wave of you know new software that's going to get created which is very very personalized to your sort of need right and you'll probably have and who does the maintenance like you'll have to have somebody from the team doing it again and again right and i mean like or do you guys have like a service under the product business that you have we don't have it but there will be emergent experts who will sort of maintain the software for you.

56:46I think there'll be a new cottage industry that'll sort of spring up as the wipe coding movement sort of takes up where you'll have parallel economies running where people are maintaining your software and you can sort of build things on them. But I think the way AI is progressing, I think a lot of power will go to the end user. This is my individuality thing also. Once you start trusting it, you'll say, like hey you know I don't you know and it'll happen and I mean and maybe like you know you'll start with one version of the software right and then sort of you know like you know give it to somebody else to sort of build it maintain it as you sort of learn more but you can totally see you know there'll be like a task management software for your industry or for you which will be built at a very cheap cost and you'll have full control over it You know, at one side, there are a lot of people who are scared of losing their jobs.

57:44And then there are one side where people like you sitting out of India, building it for the world. You are growing at a crazy pace. You're raising millions of dollars. You're generating millions of dollars of revenue. You're killing it, right? And you're building this big thing. So there's a fair shot of you making it big. you've already built businesses before you've gone through this whole cycle and you understand this whole phase right so A should anybody take a risk to leave their dying industry if they're very talented and join someone like you and B who are the kind of people if they join what are the risks involved yeah so I mean we are seeing crazy growth right now we just recently crossed 50 million dollars in ERR.

58:37Right now, we are one of the fastest growing AI startups in the world and in the history of the world. Not just in AI companies, but historically, no company has grown this fast. Really? Yeah. We have crossed 50 million in under seven months. Our growth is accelerating. Every month, the number of people we are adding, the number of revenue we are adding is actually accelerating right now. And we just closed a large Series B. We are right now, one of the most top funded AI startups in the world right now. I mean, in India right now. How old are you closed? We have closed a$70 million round. Sweet.

59:14So that puts us in like$100 million raised so far. And we are building, and currently, like we have the best part in the market for this category. And this category is going to be like a super huge category where, you know, we will democratize all of the software development. And right now, I feel like, you know, anybody who's building anything in any industry should jump they ship to an AI native company. For two reasons. One is the learning curve is extremely steep. Most of the learning will come when you are in a high growth startup in a new area. This is literally like 1995 when internet was happening and people were in older retail jobs and these kind of things.

59:49Most of the money over the last 30 years got made from internet companies. So similar things are going to happen in AI right now. And people, first is learning, second is obviously the financial upside. The rate of companies like us are growing right now. I think it's the best time to come and join companies like Emergent, where we are super young, super early, extremely, extremely going after a really large market and growing super, super fast. And the work is extremely exciting right now. The kind of work that you get to do, you always need first-person thinking to solve the problems, lot of leverage because we are AI first as a company so you get a lot of you know you think in an AI first way and small teams like really large problems never been solved problems before and extremely fun place right now right so but tell me here's my worry a lot of young people a lot of like extremely talented people are gonna watch this yeah and one side of them is like okay this is exciting right I want to become AI part yeah If there is no company from India, which is a global level, world level, it's a very big difference.

1:01:00Because one side is, you have sort of like missed the ship of joining OpenAI or Anthropic, because they're like really large companies now. Right. That's one. Second is, India is becoming very rare. Right. And third, AI is becoming fast-growing, money is becoming one of the well-funded companies. So it's exciting, maybe financial rewards upside. Learning also. learning be upside second is my worry what if AI is boom it's just a bubble which is all over random bubble super funded companies burst from one day and the bubble crash in the world dot com time so my worry is that I leave a stable job and go to a young startup so I don't know if I don't know if I will make money or not will be all that is missing?

1:01:56No, I think even if AI progress stalls today which there are a lot of chances like the way progress is going like there is enough that industries will get transformed right? It is quite apparent of course there may be like some momentary temporary bubble that might get formed like which will be in my opinion much bigger than where we are today right? And but I mean whoever has joined in 1995 he will not regret today right? Maybe in the 2000-99 bubble, they would have been sad for a couple of days, a couple of years maybe. But they would have been coming out super, you know, like, they would be people who are at the top today.

1:02:35So I think, you know, like anybody who's jumping on AI today, right, and especially AI native companies like Emergent, like, you know, like where you have hard problems, you get to start thinking from AI first, you know, away. I mean they are probably most likely to win in the next decade or so why should somebody leave their stable job in a large company where their promotion is due and they know they're gonna become something to join some young company at very small scale right and it's very unstable and risky what's the incentive I get it one is financial incentive which is almost like you don't create a startup so you join almost like you're building only so you're almost a mini entrepreneur inside an ecosystem where you have to not care about money so that's a one incentive which i love a lot i love a lot of young people who should join us because we are we are doing something what media companies 30 years ago did some sort of that right so i i'm excited about it but what's the other incentive i think financial is the least incentive I think people doing good work will make money wherever they are.

1:03:47And I think, of course, companies like us will be financially extremely rewarding for people, both in terms of comp and equity and everything. First, I think, is you get to work with really, really smart people. Half the team is under 100 IT rankers in engineering. So you really, really get to work with really smart people. Second is you get to work on really, really hard, newer problems. Like, this is a problem that the world has never solved before, right? And you get to, you know, take the first chance at solving some of these problems. A lot of the problems that we have solved, you know, a lot of people have not solved yet, right?

1:04:21So, you are on the extreme bleeding edge of technology. Every single day, internally, I see a demo that blows my mind, right? And you'll get to experience that. You'll get to build that, right? Thirdly, like, you also, you know, get to get a chance to, like, really shape the future of AI and humanity, right? Like, you know, you get to sort of have a part in, have a say in like, you know, what direction humanity is going to go. Right. And very few people, you know, will get the chance to sort of be part of the ship where, you know, like you're actually really directing for them. Let's say, you know, you're joining Google in like 95, 96.

1:04:54Right. Like those people are really shaped up, you know, what our internet looks like today. Right. And today, like people have this opportunity to sort of really come in and, you know, solve and shape the way sort of humanity is going to sort of build tomorrow. and really, really democratize and empower all of humanity. That's sort of the way. I mean, we actually want only people who are extremely mission-oriented, people who actually really care about making a large impact in the world. And if you care about impact, if you care about building something giant, great, if you want to sort of compete at the top global level.

1:05:29Internally, we said that we are in an Olympics game because now we are competing with every single person or every single company like at the top level, right? So if you want to have, you know, feel, I mean, really experience what like highest of standards look like in the globally, right? If you want to really feel like what does problem solving at, you know, where maybe a handful of 10, 20 people are capable of solving these kind of problems. Like if you want to experience that kind of thing, then Immersion is the place to come. Fair. Fair. I think people have, but it's crazy, man. It's scary.

1:06:02even though it sounds too good to join it is scary and but the other side of me is also like when i put a hat off if i had to switch industries and not producing the other side is if i can't see anything crazy happening in the work that i'm doing right now even then i'm at a shit place where i should be more scared because probably this one's gonna to go faster than something new which i'm trying to become part of right no i think i mean um like people should think about upscaling themselves and if they have an opportunity to join some young startup where they can learn faster they can shape the way world is going and they they can they have a fair shot of good financial reward i think they should because being stuck at where they are is more problematic and more scary than not joining an unstable board.

1:06:58I mean I wouldn't call us unstable right like I mean in the sense that like you know like I think. The world is unstable right now. I'm saying startup like yours, mine or even like open AI like we don't know like Nvidia also because like this is we are at a we are working in a world which is unstable. Yeah I think that I think that is yeah only risk right now is not taking those risks. Yeah that's what I meant. Totally, totally. And I think as long as people are like deep into AI, I think that they're going to be fine. Yeah. Interesting. Where were you born? I was born in Chandigarh. Okay. And we are triplets.

1:07:40Triplets? Yes. All three you look same? No, no, no. So I have a sister. She lives in New Zealand now. And I have a brother, mother, who lives in SF. And with whom I'm starting emerging. Right. So it was a pretty fun childhood. Okay. and then all three of you like do you look do you guys look same? So me and Madhav look similar but not exactly same. If we were looking same like our life would have been very different. I was never good in exams. Nice. And then so you guys grew up. After which college did you go to? So college I went to United Arabad and there I realized that there will be something that I will have a little bit of a realization.

1:08:22that IIT missed. I didn't think that I was going to go and I realized that I was going and I saw that all smart people are going to go and the kind of people that there are. I realized that I missed. Which one did your brother go to? He went to ITVHU. So it was pretty close by. I was in Allahabad. He was in Bananas. And we used to spend a lot of time together. and first time I had this adult sense that I could have been there and you know I was learning a little bit and then I was a little serious that I have to do something in life and I used to be very inspired by Bill Gates and Steve Jobs because I was told the first time that the world's rich is a software so I said that I can do it

1:09:26I was very inspired. I got very inspired. I got very inspired. After that, I managed a lot. I got an internship in IIT Madras. Then third year, I went to Spain for my internship. We didn't have any time for an internship. IIT's are quite common. I put in a lot of effort. I wrote a script that all professors look at online. and I automatically email everyone outbound marketing and I have written the script on the internet and send them and send them and send them and I am a third year student and I want to study and I have researched and I have 3-4 internships and I went to Spain and then I was interested in researching so Madhav actually went to Penn State for his PhD and I also went to Colombia to do my PhD and New York New York and and and then I when we went in 2009 so in 2008 there was a financial crisis and I was not very like reading news and other things I thought that everything will happen I didn't understand the gravity of a financial depression then that's what happens and when I was in New York I'm talking about seniors first

1:10:49and everyone has no job. Everyone has a problem because it's not going to work. But luckily, I got into Google in internship. And then I wanted to do PhD and I wanted to do it. I've been told that Google has already 2-3 years before and I've been told that what a benefit is research again. I know there's no output so I've been

1:11:21I got into that team and then I told my advisor that I don't want to do PhD so I'll finish my master's and then join Google. So I joined Google. And I was the person with the most ideas there often and I would push everybody around that young energy and people used to be very I would say like excited but they're also and and and and engineering how do you build high quality teams how do you build a culture you know where you know everybody's always excited about great work how do you focus on user and back in those days Google very strong culture and we shipped did google knowledge graph project and search quality when you type what you type what the result would be our team what the result will be the result and what the result will be the result and that first time I worked with a lot of big data stuff so a lot of learning is there but I felt like I don't have any ideas I don't want to push it and I felt frustrated that I want to run more and more and I want to run more and more.

1:13:18So I am starting to start up. What insight did you start to start? The insight was that everything needs convenience. The economy is growing and people you know time spent something to do first we started as a conscious service you have to do something, you have to do something, you have to repair you have to do electricity bill pay, whatever you have to do, and second we felt like now because mobile phone is coming and data cost is working and of course Uber was also there as a model it was like the labor the work that can be distributed now and there will be enough economy created on this side and that you can create a marketplace of both where a lot of people want to outsource some work there are enough people who want to do that work and you can create make the economics work and largely we wanted to live a convenient life and if someone does this work to help us.

1:14:32We are good. We thought that there are other people who want to outsource some work. What was that? When we got to deliver by, we would go to deliver. We would have to do 20-30 days. We would have to learn something about what we want. When you get to a human, you see the smile on their face. and early in a lot of early days Danzo was a saviour in many ways those people want something in the last minute they forgot what they were they forgot their food they forgot their charger they forgot their food in the midnight so you could see that delight on their face that Danzo saved their day so what happened Danzo was instant hit you were able to raise and attract capital also what happened what happened so Danzo was almost a phenomenon in many ways and we were pioneers on one of the biggest trends that got created quick commerce

1:15:34and generally if you if you see founder journey right like if you rank what matters number one is that you don't have in the right market second is product third is team execution all of that and generally you know like most companies fail because they chose the market wrong. The timing wrong. And it was our reason that we are in India one of the biggest trends. So we were in the right market. We had the best product. We had the best you know we had a great brand great customer service. Few things happened. I feel that we were unlucky because honestly right because if we have two months and we survived more.

1:16:23Our story would have been very different. We were getting confused that the economic downturn came. We were not able to raise capital. Even though our business unit economics became very good over time. And of course, reflecting back, there were a lot of things we would have done differently. But generally, at least, I think that our story is quite different. Because we did all the hard things. We came in the right market, we survived a long time, we made a good team, we made a good product, we made a good brand. binaja. Opsolve kia. Opsolve kia. Zet office and but our way is that we are getting to scale and we were spending a lot of money setting up the dark store scaling really really fast and that our assumption is that we will raise more money and we will do months and we will that we will do but at the same time we were a lot of business lines.

1:17:22Pick up and drop, marketplace, food, you know, many things. Darkstore was the one which was working. And one mistake I think we did was we should have shut down everything else and focus only on Darkstore because that was the one that was going to work. Which is a quick commerce. Which is a quick commerce. And we were one of the first people to set up Darkstores in the country. Which is a Zepto Blinkit. Yes, absolutely. We were just a model. We were already seeing that brand value before it was a big thing. Brand value was a big thing. right well it could go on for different things but thank you so much thanks man I loved having conversation with you right I I know that you've built an app for me yes right what is it for me and how easy to make it we can ship it we can show it right now right what did you think about tell me and show me that yeah so let's let's create a you know as we were discussing can people create podcasts with you okay let's try to create a you know a virtual podcast thing with you where people can create a podcast with Raj.

1:18:24Okay. And we'll just have to prompt it saying that, hey, I want to build an app like this, right? And so let's say I want to prompt it. So we'll prompt it right here. Yeah, we can just write, create an AI-powered podcast creation studio where users can generate podcasts by writing prompts. Clone their voice, use pre-made AI voices, including feature host voice of Raj Samani. and host is Raj, and we can add more voices. So let's just start with like simple prompt. We just start with this, right?

1:18:58And this will set up our environment, you know, a sandbox that we create for where your agency is going to run. And as soon as you start, it's, you know, like now your app is building. Now we can chat in the background. Your app is going to get built, right? It might ask you some questions. And you'll get first version in a few minutes. Then you can play around with it, give it more feedback and you know as we are going through the chat you know we can let the app build in the background how fast can I build certain apps let's say I try to do one which is a basic one let me ask you a little more complicated one I like this idea of to make a task manager for the podcast company I love that I want to build that my task manager is my company just for the podcast yes you can build this in like a week you know I mean by week means that you'll still have to probably spend some time back and forth depending on the complexity the first version you'll get like in like a couple of hours okay right and then I'll keep prompting I'll keep prompting what are the chances it'll fail I mean it it will fail like you know maybe 20-30 % of time okay like where in the fail instance it'll still have some bugs but you know you can just tell tell the agent hey this is what I'm facing the issue and it'll sort of go and fix it and because we do all of the testing, debugging all of those things automatically for you all you have to do is keep testing the app and debugging.

1:20:28So it basically asks you some question like for example what voice generation you want to use. So let's say 11 labs. We have integration with 11 labs 11 labs It's asking what LLM do you want to use for script generation let's say GPT 5.2 and then it'll ask you a bunch of questions like, hey, do you want to add any new feature?

1:20:54So we give you a lot of things built in like authentication databases, all of those things are built in, right? And then you'll say, okay, just build. And I mean, it is as easy as just chatting with your developer, you just chat with it, you know, like it'll start doing its work, it'll run for maybe like 15-20 minutes in a session, I'll give you the one version so every 15-20 minutes you'll get one update saying that hey this is what I've built so far so you're telling me that every person who can make the problem for the app totally right in how many rupees so he varies how many complexity people I would say typically you'll get an app billed you know like$50 to you know between$5 ,000 right and depending on the complexity of the app you're getting complex app like this this will be under like you know$50 you can build an app like this that's it yes and then I can do whatever I can build my own habit tracker just showing you here like now we have a design agent that has come in right it's trying to you know design your app right so we have multiple agents that will come in at different point of time doing the app building so it got your requirement it looked through its memory to figure out how does this integration do it it figured out how does 11 laps integration looking at you know like the agent has the memory of like what all apps it has built in the past and now this design agent has come in it's trying to like you know read your prompt and understand like what are the best it's going to do some research online it's going to try to figure out it's going to take some images and it's trying to figure out what is the best way to design your app right trying to find some themes and it's going to you know like put it back what kind of apps people can make what is the most common kind of problem that people are trying to solve by building an app which is personalized for them so we have a lot of small medium business owners who are building who are trying to digitize their business okay they say you know factory people manufacturing construction a lot of people are building erps procurement software a lot of people are building crms and and those kind of things in the platform a lot of entrepreneurs are building ai apps roughly you know 50 percent of our apps have some AI component in it.

1:23:04And especially because AI, you know, like, it compresses the logic, right? Because you can now prompt a lot of things, right? A lot of people are trying to, for example, people are trying to build like AI lawyers, AI chatbots, you know, for their specific industries. A lot of people are trying to build agents on the platform today, right? So AI as a theme is pretty big. People are building voice agents. Somebody built, you know, somebody automated their whole calling function using Emergent, right? And people build that. um what are you saying so it's saying that hey do you have an api key for learn labs now i have one i'll just give it to it so wherever it gets it like it requires that help will come and ask for you uh for example if you wanted like hey do you have you know do you want to use your key or my key and then i just give it a key uh saying use my key uh and it'll take us like 20-25 minutes to get the first version out and then you know like you can done and we'll have that app yeah we'll put it here in the editing yes yes yes to show people what the app is perfect done nice thank you so much thank you so much for having me it was really fun

1:24:29Everyone talks about success but almost no one talks about the trade-offs behind it. The late nights, the wrong decisions, the uncomfortable learning. Today's conversation is about what it really takes. Hey Raj, good to meet you. How are you? Please come on and enjoy. All good, all good? Awesome. What's up? All great. I've heard so many of your shows. I'm so excited to be here. Thank you. I'm excited to have you man. You're doing killer work. I believe I'm this five-year-old kid sitting in front of you and I'm randomly curious about everything that you've done. So I want to learn more. So that's the whole process.

1:25:03You know in fact, there is a process that I follow in my life now. So in Israel, when children go out and go to school, then they go to school and then they go to home. The first question that mom asked her is, Which question is the biggest question today? And that's a practice and it's just there. So they ask, what's the most interesting question? And she tells me that I have asked this. and it's a habit it's a habit that everyday you ask your kid what question did you ask so you're training your kid to be more curious and ask more questions and raise more questions in the world and I do that like every night I go back to bed and I write down a biggest question now I've made my agent and I've told you that I'm asking a big question and then I think how do I ask this question and it's a habit everyday that I have asked what I have asked today.

1:26:00It's an incredible thing to do. Thank you so much for watching this episode. Now I have to do three things. First of all, subscribe to this channel because as much as you subscribe, we will get better and valuable guests for you. Number two, tell me in the comments, what did you feel good and bad in this episode so that we can repeat those mistakes and which guests you want to see in our podcast so that we can take them and give you more value. And number three, this episode, to share this episode with someone else. Because one conversation can change someone's life. I'll see you next time. Until then, keep figuring out.