AI Founder: The Right Way To Use AI To 10X Productivity | Sravanth Aluru | FO572 Raj Shamani

Can we make our entire organization AI. native with AI when I look I'm worrying. that people are getting dumber for sure. the people you can hire tomorrow are all. going to be dumb humans shouldn't be the. ones pulling AI AI should be pulling. humans. what I see is across the board leaders. are asking team members to get on a. co-pilot and then what they're doing is. they're doing them an injustice by. telling them you become dependent lot of. startups die not because of trend they. die because of wrong timing so how How. do you pick up the right time to enter?

Number one, you can't just be on. internet, right? You've got to be able. to create a advisor network. Talk to. people who've done this for a while. Second, make sure that you feel you know. the basic tenants and know what it. entails. If you don't know that space, please don't jump in. Sprint time. The. last, don't expect to be right on day. one. If a young person is watching this. today, what are the massive. opportunities that you see next, the. next inflection is going to be the.

influence of AI and robotics. So that's. one clear thing and there's just so much. opportunity. If you're starting now, AI. is too late. The ones that have to. figure out have already figured out. >> What's more dangerous for an. entrepreneur? 5 years entering too early. in a space or one year late? I I would. certainly say 5 years early. You're too. late if you're one, right? You need to. know the secret before the world knows. In fact, Squire has this whole concept. of saying if you have a secret and the. world doesn't know it, I'm betting on.

it. Like Sam Alman can shape AI the way. others will never be able to. And there. is that advantage. I would rather be 10. years late to the game and play an. incremental to that game or be 5 years. earlier and play the exponential game. Let's say young founders watching this. They have an idea and everyone around. them is telling them your idea is. stupid. How do you know that your idea. is stupid or the world is wrong and. they'll know? That's a very good. question. I have a small favor to ask you. I need. you to subscribe to our channel. The.

more subscribers we have, the better and. bigger guests we can bring and provide. you more value through these. conversations. And the full audio. experience of this show is also. available on Spotify where you can. follow us and listen to the new episodes. as well. Some of the biggest fortunes of. the next decade are going to be built in. AI and our today's guest is one of the. people building them. Shravant Aluru is. the CEO and co-founder of Avatar.ai.

He's an IIA Bombay and Watton Aluminous. with over $55 million raised. His. company just launched Varya, an Indian. video AI model 10 times cheaper than. global rivals. In this episode, we'll. discuss why does he believe AI will be. bigger than the internet, mobile and PC. combined, what kind of AI companies are. going to become big and which ones will. disappear and why did he give up. majority equity in his own company to. bring in stronger co-founders. Watch.

this episode till the end and to know. more about Avatar.ai and Varyia, check. the link in the description below. Let's say young some young founders. watching this, okay? And they are at. this curse where they have an idea and. everyone around them is telling them. your idea is stupid. How do you know. that your idea is stupid versus. either you're early or you're completely.

wrong or the world is wrong and they'll. know? Like what's the differentiation. between all three? Yeah. See um. that's a very good question. There are. two games you play in business. >> One is an incremental game. It should. not be stupid. The step jump game, the. paradigm shift game has to be stupid. because if the world already knows your. secret, then there is no secret to start. with. Right? >> So first of all, the the clarity that I. think a founder should have is which. game am I playing?

>> Cuz in one the risk is very high. The. reward is equally higher. Everyone wants. it and it's a headache and you're just. helping it happen online. But is is that. a paradigm shift? No. Will will it be. called stupid? No. But is it a bad. business? No. It's a great business. It's a here and now demand and you're. serving it. Will it be a you know one is. 2000? No. Probably not. Right. Um. >> if that's a game you want to play, be. very clear. Go make sure that you know. learn what not to do from experience. So.

the first thing I would advise a founder. is especially if they're young because. you want to start earlier in this new. generation. It's not my generation where. you know with AI and I there is no point. of you wasting time if you're an. entrepreneur mindset. Um. >> okay. >> in this generation if I were born 20. years late I would not wait till 30 to. start my company. I would have done it. probably earlier. >> True. >> Um I had a loan and I had to take care. of my paying back the loan and all of.

that. So I had to work a bit to do get. there. Right. Um today I would not even. take the loan. So that's the reality of. where we live in. Um. >> so you played two games. >> Yes. Right. So the other game is. >> one is where you get. >> like,000 or zero. >> Yeah. >> And one is where you go from 0 to 10 200. then 150 then 180 then 200 and you step. by step by grow. Right. >> And these are and I'm sure they're. shades of riskreward. >> incremental and paradigm experiment. >> But these are the two extremes right? People keep confusing these two. Um.

especially when you're starting as a. first-time founder. Uh you have to be. very very clear which game you're. playing because the game is it's like. football versus cricket as in the rules. are different. Uh on the right side you. want people to misunderstand you. You. want people to challenge you. The more. people challenge like when I started I I. knew people are going to challenge. Every time someone smart would challenge. me, I'd be like, "Thank God so much.". You know, I mean that's the secret you. and of course you need conviction.

internally and it has to be logical left. and right both together, not one side, right? You can't just go right and say. get over optimistic or equally go left. and become very linear. You want the mix. of linear and nonlinear conviction. within. Um but I I would actually find. it comforting when people would say. and I had left a reasonably pushy job. So there were friends who were all smart. similar backgrounds who were like are. you sure you're doing the right thing. with your life? I'm like yeah one life. who you know and I wouldn't even resist.

>> But you can't do that if you're building. an incremental business because you want. to be someone everyone can understand. everyone can relate to. And the reason. I'm saying these two games are different. is in this game you have literally the. first goal you have to have is PMF. >> in incremental. >> Yeah. Because you really need to figure. out how to get product market fit. because here and now demand. >> You're serving something today, right? You're not building something for. tomorrow. >> And and so the game itself is about. today. It's about creating product.

market fit. And you've got to be. learning from the best out there because. it's not easy to achieve. I'm sure you. and I know that from our own businesses, right? Um. on the right side, you can't expect. product market fit in your year one. Look at open AI. Look at how how how. many years they had to do the deep. learning to become the open eye that you. talk about today, right? Uh. >> and so there is that deep trenches work. where you're challenging the status quo. >> and you know it's not an incremental.

game. You've got to really challenge the. status quo and it's fundamental. Okay. >> So, that game is a very different game. out there. I would say just the opposite. dynamics. You're literally going for the. moonshot. Uh and you'd probably start. saying, "I'm going to focus on Mars, but. maybe I'll land in moon.". >> Uh, and that's good enough. Or I I can. at least get the escape velocity to get. out of atmosphere. That's good enough. Step one, right? You've got to take. break that down into simpler goals. because your ambition is so long-. winded. And that's why you see deep tech.

companies take. >> longer, right? So I I don't think this. is a generic answer. It's very very. specific to. >> No, I like what you what you're saying. right now. I can I want to go more. specific. >> Y let's. >> So it's like you said there are two. games to play incremental or. exponential. >> You decide which one. >> Yeah. >> Uh if you play incremental game then you. need to be understood. >> Okay. People should understand what. you're trying to do. You need to have. PMF from day one. You need to be. prepared for having small returns and.

you will keep achieving goals on a. short-term basis again and again. >> Yep. >> Then you if you play an exponential. game, you'll be misunderstood and people. will call you stupid. That's the price. early on. No PMF from day one. You can. expect either great returns or great. risk or no reward at all. And it's a. very long-term game. So you need to. break down how do you actually reach. that long-term vision in smaller time. buckets. >> Y. >> so these are right now the only.

differences or would you give more. differences between. >> they're more on the left there's. competition. So the game plan is very. different. So in incremental there are. competition. >> there's an likely an existing. competition you're getting into on which. you're doing an incremental margin. So. there is a reality of a landscape. >> h. >> right here probably in the first couple. of years you're a market of one. >> and there's no market. >> Yeah there's no market there's no. product there's no product forget.

product market fit there's no product. there's no market and it's it's both are. equally both come with pros and cons. So. don't don't think one's tougher or one's. don't glorify either side. Both are. equally challenging cuz on the left it's. a lot of relativity that you have to. absorb. And on the right there's no. challenger. So you've got to keep. yourself you know the paranoia needs to. be within the company cuz it's not. outside on on this on the other um. incremental game. The paranoia is.

natural. You're just reacting. So it's. more important that you're creating a. culture that's sensitive. >> to these external stimuli. there is no. external stimuli on the other side. So. you can happily sit and do nothing for. the next decade and and sometimes I feel. like I'm doing that. you know u but I think compounded. learning on the right side I think. compounded learning is what matters on. the right on the incremental game. compounded PMF is what you want to focus. on because it should ideally not need so. much learning.

whereas on the right side you probably. you don't have any other metrics so. you're replacing compounding PMF with. compounding learning. >> as your until you hit PMF you that's. your yard stick that you're measuring. measuring your own success on and. hopefully you're paranoid about it cuz I. don't think if there's no paranoia and. you're on the exponential game then good. luck to you cuz. there's no pressure on you and you can. be a labs and you know keep become a.

nonprofit. >> and just stay there forever. >> stay there forever. >> one day will never come. >> but these are what you said these are. the characteristics of the company okay. and what you you need to be aware what. game you're playing. >> Yeah. >> A lot of time it starts with an idea, right? How do you know which game my. idea lies into? Yeah. If and anecdotally, I don't know. the exact percentage, but I'd certainly. know from my own experience of being a.

banker, um. probably more than half companies change. their idea that they start with. H. >> ki if you already know everything before. jumping into the swimming pool right uh. then there is no real opportunity to. start with. >> you I my suggestion in both cases is to. never be closeed-minded. you an idea is a starting point and an. idea means nothing in this world if you.

don't execute well so we the world. overvalues ideas right founder. tag. >> all of that right it's over glorified. but honestly the real value is getting. created by executing it well. >> right. >> so I I wouldn't even I wouldn't even say. I need a perfect idea I need a vision of. what is a supply demand. >> ecosystem that I'm going after and I. need to have some sort of a emotional. connect with that supply demand problem.

statement that I'm going after that's. enough to start a company. >> so for example if your idea is. something that exists. >> y. >> and you can just do it better and you. have a better solution for something. that already exists. It's an incremental. game. >> If it's something which doesn't exist. and you think it should exist because it. will solve larger problems but it. doesn't exist. It's the exponential. right. That's the right framework. >> But who should do what? >> So the exponential I'm talking. exponential guy the incremental guy has.

to be very leftrain. >> logical. >> Yeah it has to be leftrain. the game is. left brain. >> So in fact the other way to think about. it is and I'm sure there are shades of. this. Okay, there's a 40% incremental. 60% exponential. >> flavor to the game. Uh but if I were to. just explain how I see it, I would say. the exponential is where the right brain. it starts with the right brain. Hopefully you're moving towards left and. on this it's left brain and you better. stay in left. Don't go right in the idea.

or imaginary fuzzy world. don't go right. in that game cuz honestly um you know. efficiency for example is a very real. value creation. >> Yeah. Yeah. >> It's you know maybe someone doesn't find. it very aspirational to help with them. >> for me as long as I care about it and if. it's real value and I'm creating it from. day one so I could be profitable in year. one. >> One of the most profitable industries in. the world. >> are the physical ones. >> The guys who say that they know how to.

make it efficient. and they don't even. make it efficient. If you want to become. an entrepreneurial founder, >> what kind of person should aspire to. build an exponential company and what. side what type of person should aspire. to build an incremental company? Because. there must be some kind of. needs, want, characteristic profile. >> Yeah. >> Which will fit. a certain type of game, right? >> Yeah. I'd say for me if I I'm playing.

the exponential game and for me I think. curiosity was very important. Um in fact. we just have three values at Avdar and. they'll represent my answer for the. exponential game. >> So it's learning mindset. People also. call it growth mindset. The same thing. it's effectively you grow 1% every day. at the end of the year it's three. >> yeah three path 365 right the. compounding of that learning. Second. love was curiosity because.

if you don't have curiosity you don't. get into zones which are unknown and the. whole game of the exponential is to get. out of your comfort zone. Go into. unknown unknowns experiment I trade. learn move forward. So you you've got to. have that little bit of a curious nature. to get into uncomfortable zones and bet. um and be okay with the failure because. your curiosity says okay I learned what. not to do maybe not what to do but you. know what not to do is as effective in. decision-m tomorrow. So having that.

curious mind that values learning what. not to do is important. >> Okay. >> Um you don't want to do that in the. incremental game. The sec the last I. would say is collaboration. And I think. finally see we are all I don't think AI. is anywhere close to human ingenuity u. and the the crazy brain that we have we. our judgment abilities are very. nonlinear. so it's even an AI native company uh. like Avdar for example is still a set of.

people with AI behind yeah maybe they're. superhuman today compared to what they. were before AI uh all of that accepted. but it's still a set of brains so we we. don't call all hands in a we call it all. brains because finally we are a set of. brains. >> True. >> Um so you you've got to also have. collaboration there. You've got to have. the right environment. You've got to. take the fear of failure out because. it's an exponential game. So you've got. to give the ability for people to. experiment and fail and it's okay as.

long as there's learning, right? So I I. would say that's the exponential game. On the on this part, honestly, you got. to be structured. You've got to be a guy. who loves process. You've got you got to. be okay with monotony. And you've got to be okay with realizing. that world is also 80% boring stuff. while there's this 20% aspirational. stuff. >> So I actually think there are archetypes. of founders who fit into these buckets. and it's a fit. >> It's a founder market fit. >> Yeah. >> Right. where I I don't think I can do a.

linear ex incremental game cuz um I I. think I'll get bored very soon and I'll. be like. what else will I learn right so you've. got to have that innate. nerdish plus you I want ambition of. sorts to play the exponential game on. this side you could play a very. calculated game um you could just. literally learn the craft work for a few. years in that industry that you're in.

Uh, and then just go and improve a lot. of obvious inefficiencies that you've. seen, which one's better? I actually. think my game might be, it's very tough. to actually judge, but I I actually. think that if you're a left brain guy, it makes more logical sense to do the. incremental game because it's real. >> on day one on day minus one. So, >> there's some reality to it. >> On the other side, I mean, nine out of. 10 companies may go bust for all you. know. So. >> true. >> Do you have that ability? And I think.

it's more intrinsic. >> The the founders will get attracted to. one of these games or a shade in between. which is their mix of how they value. monotony and value creation. >> and a mix of that scale will define. which founder would fit which true. >> So I I wouldn't even say which you. you're not deciding the game bases the. idea. You're deciding the game. This is how. you want to play. Like a web of Suryani.

playing a Rahul Dravid's game doesn't. make sense. >> Fair. >> So it's same cricket. >> though founded by Rahul David. >> Yeah. Right. The mentor I took that. example for that reason. I mean but he. has to play his game not Rahul Rit's. game. >> He needs to be thankful and grateful to. Rahul's uh role of hand right in in. where he is today. But he has to play. his game and it's the same cricket. field. It's the same opposition likely. It's the same field whether all of that.

but you have to play your game. >> And I think that's the best analogy I. can show. So I don't I think the game. you're playing is a founder game. I. think cultures and companies are a. founder culture. If you end up with a. issue internally, don't blame the. employees. It's a reflection. It's a. mirror for you. >> cuz these are very founderled things. because you're the first guy. I mean. >> true that you set some of those. >> if you're absolutely chaotic.

>> Yeah, it's. >> you see that in the company. If you're. extremely structured, process driven, you'll see see it in the company. >> Both are important. So I'm not saying. something's right or wrong. It's just. >> absolutely just do different types of. >> each their own. I would say um as an. answer to your question. >> So so based on all the things that. you've said, do you think. if someone. before starting a company or before.

because a lot of people want to become. entrepreneur by choice, right? That they. want to start a company and it's a. >> they don't know if it's a calling, it's. fame, it's money, it's aspiration, it's. looking at other inspiration. You don't. know but somehow something is telling. you that you want to become an. >> there's a drive you want to become an. entrepreneur. Yeah. >> So the first question you should be. asking yourself is what game to play. exponential or incremental? >> No. The first question I think you. should ask yourself is am I ready for. it? >> Yeah. After being ready after being.

ready. >> because the drive is not enough. There. are realistic stuff. If your bank. balance is just 10 rupees you want to. jump say a responsibility that you don't. do that to yourself because 3 months in. you going to give up. So structure. yourself to first be able to be in a. position to start up and that's not a. small thing. I've seen so many great. smart friends smarter than me get that. little bit wrong. So you've got to be. realistic to yourself saying am I being. given the opportunity I'm in in a. position if not first get yourself in.

the position and then yeah. >> true. >> because I I do think it's very important. for a young founder because this is all. very glorious. >> True. And you and I know that grass is. green from the other side but when you. come here. >> years. >> yeah same here four I would say for me. even smarter. >> So based on what you said the risk. profile is if you are one of those who. is excited by creation and you're like. it's okay if I. don't make like 10% and 11% and 12% year.

on year and I'm okay it's fine. one day. I might end up making like a lot and. there's something which will which I. will create and the world will use and. it'll be amazing. If you're driven by. that and you're okay to not make money. in the short term, go for exponential. one. But if you're somebody who's like. by next year I want to have x amount in. my bank and then after that I want to. have x more and then x more. You. probably want to stick to incremental. way because you're not thinking zero or.

100. you're thinking 0 to 1, 1 to 5, 10. and that's how so that's a better way to. choose which. profile or which game to play. incremental or exponential and then. there's a list which we'll put out as an. edit on the screen which we just. discussed right now and then you said. one more thing to add. >> so I I would say. >> what does success mean to you that's. that's an important question people. should ask before starting. >> yeah because I think every every. decision is a riskreward ratio and you. need to know the risk you need to know. the reward.

And what you just said is in a way your. way of approaching your inner stance on. risk but equally you want to know your. inner stance on reward. See finally. small wins is what keeps you going. And trust me that's the biggest driver. of motivation in my 11 years at least my. experience of it. And the small wins are. about your expectation and then the. result. It's not just the result. The result being greater than your. expectation is a win maybe how however.

small it may be and that keeps you. going. >> The result being lower than your. expectation is a lose. >> and you have that every day. Any good. guy will have cognitive dissonance. saying. right? >> So you don't want to put yourself in. cognitive dissonance. So I I at least. I've learned in my life that every. decision that you make if you want to be. true to yourself first there are two. questions you need to ask. First assess. risk which is am I you know the. questions I just laid out right am I.

even in a position where I can take this. risk. >> and if I do then what kind of risk am I. comfortable with am I betting allin am I. playing poker am I playing you know far. more game of skill where where do I land. or the other other part is the reward. cuz it's finally riskreward the reward. part is what I framed as saying what. does success mean to me or rather in. other words when Do I feel a particular. moment is a win. cuz finally it's my expectation and this.

is the reservation price I'm trying to. ask myself and then the outcomes right. and I think these two questions itself. will tell you which game to play. >> true because it'll give you the profile. that you you are at some. >> I'm thinking of what game I play and. even though I'm in an incremental. >> no you're not incremental right at some. level you are. >> in an experiment Because I play a. 10-year game minimum and allin game like. I'm. >> and you have a first principle which is. very abstract but so profound.

>> Yes. >> Right. So it's not someone like me. coming up from outside can't predict. your first principle but the moment you. say it I can connect the dots and. everything makes sense. Right. >> So you have that with you which is where. I think exponential game is happening. already. >> And even in an exponential game finally. it's incremental. >> compound. >> compound. Yes. So you can always micro. zoom and say it's an incremental game. >> True. >> But yeah, because. in my head whether it'll be a success or.

a failure, I will judge it in a 10-year. window. I won't even judge it. I don't. even see like today anything at all. because I'm like. >> it's just all in game and we everything. we make we put back in like. >> yeah some. >> exponent like compounded make it bigger. make it bigger and all in game like it's. a very. how sometimes I think it's it's a very. wrong type of race it is going all in. >> it's an archetype Raj. >> it's a founder archetype like are you.

wired like that sometimes. >> yeah you are because it it finally comes. down to probably was saying um at some. level you enjoy the journey more than. the other. >> I agree and that's the fundamental route. that manifests as I'm thinking 10 years. The point is you're enjoying your today. The world may think I would stop if I'm. not enjoying. >> right but you are driven by that. So. it's a very absolute stance of saying. I'm enjoying my today. So problem.

>> I agree. What do you want tomorrow? >> What will I do if I achieve whatever. your so-called outcome is tomorrow. morning? I'm enjoying today and I will. enjoy it tomorrow and that isn't that. what matters and if I can contribute. every of that moment, wouldn't it lead. to what you're asking me? >> So, it's a very different mindset. But. there are other people who are very. relatives, >> right? For that's a very intrinsic. mindset. And I'm not saying they're. better than us or worse than us. >> True. >> They are them. They're as unique and.

special as probably our archetype would. be. But they want instant gratification. And who said instant gratification is. bad? The whole world's running on it. So. fine. I mean, if you're in for instant. gratification, >> you want to play that game. >> because otherwise what will happen is. you being. >> judged every day. You'll get frustrated. cuz you'll be like, what's why am I. wasting time doing recordkeeping of my. journey? I should spend time making the. journey more worthwhile. >> True. >> Right. So you you will not fit into that.

that frame of thinking you know founder. and someone who's starting right if a. young person is watching this today what. are the three specific. massive opportunities that you see next. the way you saw AI years ago. >> what do you think is next what should. someone. look into today. >> see I I personally feel that AI has come. to a point where a lot of the. yesterday's.

um avenues are going to be disrupted. >> Okay. Like what? >> Like a lot of the jobs that we do uh I. keep saying that but the coders are the. first to go. So karma, >> right? But it's not going to stop at. coders. It's going to go. >> disrupt a lot of the knowledge worker. >> concept that we live with today. Um. >> so it's a I would say a paradigm shift. that one needs to make if you look at. the next two decades of career.

>> and I I was actually reading up the. industrial revolution. So I was uh in. this world economic forum initiative um. and they framed it framed AI as the next. industrial revolution. So I actually. said let's let me go back and read up. and what to my horror I really I got to. know that the last industrial revolution. when say machines beat human body took. four decades of pain. >> 40 years of pain. >> Yeah but I'm sure it was much slower a. world than we are in today. So I would.

say let's assume that we at least 4x so. it's at least a decade of pain. Um and. it's bound to happen because and I'm. seeing it today in say the enterprise. customers that we serve in US or say the. Middle East where the boards have. extreme pressure to show to the market. some proof point that they're doing well. on AI shift cuz otherwise their values. are impacted next day morning right so. there is a clear pressure coming top. down and in some of the cases we're.

seeing probably 60% of what humans were. doing yesterday is now just getting. Right. And by the way, the remaining 40%. uh most of these architectures are. self-learning. So you're creating a. central organizational brain. That's. what we do at Avatar for our. enterprises. And it's learning from. every usage that the humans doing. So. the humans are helping their jobs get. displaced at some level. That's a. reality. Now. >> see if we don't do it, it'll happen. M. >> so I I don't take ethical dilemma there. because.

>> but I'd rather have contributed in a way. that I can help people transition. >> Now how do you so coming back to your. question now with that context um what. does one do a young. inflection is going to be the influence. of AI and robotics. So that's one clear. theme and there's just so much. opportunity right uh it's a no-brainer. if you're starting now AI is too late. >> uh right the the ones that had to figure. out have already figured out they have.

their distribution modes and they have. the recognition they have brands now and. and so you know it's a self-fulfilling. prophecy there might be a bubble. somewhere in the in that journey like. internet.com. >> but you know the technology is so. profound that it's large it's going to. be larger than the internet opportunity. that we see saw in my belief more than. the mobile or the PC even uh because of. the nature of it's the first time. technology is a challenging human brain. and that's not a small shift right so so.

that's going to happen I think the next. in inflection is while AI moves towards. AGI there will be models like world. models and capabilities coming from the. AI side coupled with say two cameras for. a robot being able to you know and the. diversity of the complexity of the world. is such that if you see today robots can. can't even fold a like folding a cloth. is a complex benchmarking problem for. robotics right for us it'll be like yeah.

probably. how how complicate but that's the. complexity so there will be so many use. cases that will need niche. >> application and each of them would have. huge addressible markets. >> um. >> I'm think of agriculture uh and how you. could really make lives of a farmer. highly productive, you know, chill or not have to do all. the hard work that goes in uh and. hopefully use their time to think about. how do I innovate the practice of.

farming rather than just keep doing the. grunt work of farming, right? And that. could be a great purpose for someone to. start with and it can affect the whole. world's uh agriculture in a meaningful. way and that's bound to happen. So. opportunities like those will be. vertical specialist companies that'll be. needed. There'll be horizontal shovels. companies that will be needed. So you. got that's one theme I would say is a. no-brainer. The second theme I actually. think is what happens when all of these. guys are I mean at some level people are.

talking universal basic income and all. of that, right? So I do think there's a. big I I at least spend a lot of time. thinking about it. I think there's a. pre-agi world and there's a post AGI. world. And the post AGI world is a is a. whiteboard. Go paint. And if you truly go deeper in. that, right, you'll realize that the. post AGI world is a whiteboard. >> But give me an exact opportunity in the. post AGI world. >> So, you know, you could start thinking. of space as a new frontier.

Um, and today's space tech, we haven't. even reached Mars, >> but there's a lot of action happening. >> Yeah. So I'm because of the same reason. because now it's becoming viable, right? You SpaceX has done a phenomenal job. Um. I think autonomous driving has done a. phenomenal job. So there are proof. points that have led people to think but. that's just the start. I mean space. could be for many other reasons. Um. thinking about energy for example the. ideas of satellites for example with. solar cells and the Elon Musk's idea.

right it's a white board hasn't hasn't. been thought before maybe not. So you. literally have a whiteboard uh and you. can come up with these very novel ideas. which now you can aspire to do like one. thing that I keep telling my 8-year-old. daughter is build an iron man as your I. mean why not. but see. trend is one thing y okay the difficult. part is timing a lot of people can see.

this trend that AI is disrupting the. world all the old things are not going. to happen the way it used to happen. So. there's an opportunity in the world. where it's whiteboard. We don't know how. it's going to be. Maybe space, maybe. robotics, maybe. >> you have no idea, right? Something. health like something that's it's a. trend. The trend is all the old things. in the world the way used to happen, it's going to die. >> Yeah, that's the larger.

>> scary thought. >> Yeah. >> But what about the timing? A lot of. people, a lot of startups die not. because of trend. They die because of. wrong timing. Either they're too early. or too late. >> Yeah. >> So, how do you pick up the right time to. enter a space to learn and then. eventually start a business? Number one, um you can't just be on internet, right? You've got to be able to create a. advisor network. The first thing that. you want to do is talk to people who've. done this for a while because what.

happens is outside and it looks all. rosy. Mhm. >> But once you go get the devil's always. in the detail and the best way to vet. that is to go to people who have spent. the 20 25 years in that journey, right? Um so I I brought in a lot of CTO's for. example who are thinking about AI and. how AI might impact their companies from. large companies um global CTOs as. advisers on day one because if you don't. >> um there are many things that you'll. assume which then you'll have to rectify.

because you've gotten a bit deeper and. then realize oh yeah for the hara so. there are those elements of simple stuff. that yesterday also people used to do. you got to continue that so gather. experience of people who are topical. experts. Second, make sure that you feel you know. the basic tenants and know what it. entails. So. let's take an example healthcare. Um. health is one domain aspect of the depth. you need and then if you're doing AI or.

say AI plus robotics for healthcare. those are two different domains hardware. and soft software elements of it. If you. don't know that space, please don't jump. in. Spend time cuz what you know don't. know falls in unknown unknowns. >> and you you cannot predict how that will. go. >> True. >> You want to spend that time before you. jump in full-time because full-time you. won't jump in anyways. Yeah Raj, you. know all of this happens. Most nine out.

of 10 people who feel like I want to. startup don't. By the way, that's the. other anecdote I was told in Barton F by. prom by a proper case study that she. did. Um, so you know there's that in. inner battle that stops nine out of 10. people who are not in that. The last. don't expect to be right on day one. I. think this is as important an answer as. the question the answer I was giving to. your question because your question. assumes that you need to have the. perfect idea on day one. What you again.

I'll reiterate what you need to know is. that even if this idea is a failure or. simply ask yourself this question right. this is a theme this is the first idea. in that theme the first mission. statement if this fails will I continue. the theme or will I give up. >> because in your first principle of. recognition podcast is one means to that. outcome. >> if say thank god and congratulations on. the success here but if you Would you. have given up?

>> No. >> That's the answer. >> So the theme needs to be. >> the theme needs to be. >> very solid everlasting problem to be. >> left and right. There has to be a. homeostasis in your head saying yeah. theme no I'm I'm I'm in for it. And so. your all-in question I would say is not. at the idea level. It's at that first. principal level. And if if you're not. all in on in the first principle, I. would strongly advise founders not to. look at success stories like you and. jump in. Please don't cuz the the.

reality is um you know the journey isn't. that smooth. Um you will make changes, you will learn and thereby you should. make changes. >> and you should learn otherwise what's. going on right something's wrong. What. you're saying is whenever you're. starting a company or you're thinking of. starting a business, think about what's the theme. What's. that first principle big large vision. that I am going to chase for rest of my. life or rest of the years, >> right? What is that big theme that it.

doesn't need to be. a telltel business problem or a business. opportunity needs to be a large problem. Let's say for you it was what was it? >> It was effectively the eye and the. brain, right? the computer vision and. the AI. >> Yeah. >> Right. So let's say for you computer. vision and AI is a large theme that you. need to solve. Now you might have to. walk through graveyard in order to reach. there and you're okay with it. >> Similarly give me one more example. Let's say someone start thinking of.

>> I want to solve depression example just. random people are thinking about it. Right? >> So the theme is. loneliness and mental. I I'll give you a. sense right depression is your first. principle. >> u one means to that outcome is medicines. one means to that outcome could be. meditation one means to that outcome is. just TLC tend to love and care. >> right which is social. >> but there are many means to that. >> to reach that. >> first principal outcome now you start.

you say okay you know what I'm going to. try all. >> to see which one is impacting my first. principle now say day one your medicines. for some reason while you expected it to. be the easiest to get because there's so. many offthe-shelf depression medicines. in in valley. doesn't work out to the quality you of. life and side effects that bar you have. >> got it. >> whereas meditation maybe starts showing. you magical returns and TLC is equally.

good. >> you have to take that honest decision of. saying yeah medicine is maybe not the. direction and maybe you have a small. team sitting there with skills of you. know some doctors and. >> all of that that side of the the actual. hard skills of medicine right which you. don't need tomorrow. >> and at that point what call will you. take so the idea was medicines for. depression at that day. >> but maybe the idea changes. >> yeah medicine didn't work out meditation.

didn't work out. >> I'll provide social structure and love. and then create that environment. >> right I don't know if you followed Simon. Synk I think is a very smart guy I've. had in podcast. >> Yeah. Oh okay. Excellent. >> Good to be here sitting here. >> But no I I really think he's he talks a. lot of sense and the why what how. framework if you look at it idea of what. >> the first principle is the why. >> And I think he has a very profound way. of explaining this. Right. Start with. the why.

>> Yeah. >> Cuz you persist at the what's in the. house. I mean don't get stuck. Don't get. attached to those as much as you can. because let data tell you, let the. universe tell you if. what's more dangerous for an. entrepreneur is. 5 years entering too early in a space or. one year late. >> I would say one year late. >> Why? Cuz you I mean which game? Although.

>> what's dangerous? I said. >> what's worse? >> Which game? Hm. >> Which game are we talking about? >> No, let's say for you, let's. >> for me, >> we're talking about exponential returns. >> We're talking about I. >> I would certainly say 5 years early. You're too late. If you're one year, you. need to this you need to know the secret. before the world knows it. In fact, Squa. has this whole concept of saying if you. have a secret and the world doesn't know. it, I'm betting on you cuz there's a. serious value in being able to shape.

like Sam Alman can shape AI the way. others will never be able to. >> Sure. >> And there is that advantage. Now there. is a leadership dilemma part to it, right? So I'm not saying there are pros. and cons but between the two um I would. rather be 10 years late to the game and. play an incremental game to that game or. be 5 years earlier and play the. exponential game. So you're telling me. for an entrepreneur for to get an. incredible return or to have success.

success it's better to be 5 years early. than to be one year late. Don't you. don't you think a lot of people say that. it doesn't matter about like first. mover's advantage doesn't matter? >> I agree with you. I first. mover's advantage. That's why I said. leadership dilemma. >> So that's the problem. >> because the game is if you're better in. any industry you can just crush it and. you can be really you can reach it. >> equally. The next guy has the benefit of. seeing you and learning from your. mistakes. Right. So there is an.

advantage of being the second mover to. your point. So I'm not referring to. first mover advantage. I don't believe. in it. What I'm referring to is the. ability to learn when the noise is less. I'll be very honest with you. My. learning has gone down after AI became. AI. The real learning was happening when the. world wasn't shouting AI because there. was no noise in LinkedIn. Where if I. today scroll my LinkedIn, right, Raj? >> Nine out of 10 AI articles are.

inaccurate. are not anywhere close to what the real. truth is. And it's that in your face. >> Yeah. And they're written by AI. written by original thinking. >> It's an eco chamber now, right? But you. know it's that bad honestly. And so. that's what noise does. The the my value. of the years before the noise is just. the ability to be very very. valuecentric. Um. >> like. >> but for exponential you would rather be.

want to be early and fail. It's okay to. fail. It's so amazing when you when you. talk to a young founder, they always say. that 3 years, 5 years early is way. better than than being the third or the. fifth or that 20th in the market, right? >> Yeah. >> And when you talk to a big industrialist. and conglomerate, >> they say you do all the ground work. We. want to be the last ones in. >> because when we in then there'll no one. there'll be nobody. large conglomerate with capital behind. should play the game and they should.

never be first because you'd rather be. the last who comes in and says okay you. all made all these mistakes thank you. for teaching me now I bring the capital. to show you how game is played. >> and I drop prizes to one10th and then. bye-bye. >> but if you look at Google versus say. anthropic and opi today that game is not. playing out the way you'd expect it to. play out so. >> they don't but they don't. >> there are some counterintuitives to. >> but they don't operate from this. mindset. I think Google is also very. innovative mindset. >> They may they are maybe because of the.

size slow. >> but they don't operate from this mindset. that we'll be the last ones in. >> Oh yeah, for sure. Yeah. >> Right. So it's it's about the mindset. >> Yeah. It's I think the mindset of last. one in is the incremental margin. It's a cash flow business, not an equity. business. It's a free cash flow. business. So you're right, you've hit. the nail. Whereas the exponential is a. pure equity game and it could go zero.

for all you know, right? the riskreward. issue is such and. you want to be the last one in. >> sure. >> and so they these guys are saying I'll. wait until the exponential game becomes. an incremental game and then I enter. >> but it gets so difficult for a young. founder to then. >> differentiate what what advice to follow. because you know when you talk to these. large conglomerate business owners who. are inspirational. >> they get biased so they don't realize. that the reality of a young founder is.

they don't have access to capital the. way they have. >> right so they will tell you. you can kill any competitor right so. young founder is not thinking about yeah. maybe that's true. >> awareness. >> awareness and he or she's not thinking. >> that he's saying this with an assumption. that capital is unlimited. >> y. >> and for you capital is not unlimited. >> see I I I actually think one shouldn't. go and take that advice to start with. and in fact one should have the.

conviction within saying. you see what I'm saying because you need. to have that ability to h you need the. conviction to believe in yourself to say. I'll prove you wrong especially for the. exponential game as I said if those guys. are telling you you're playing a foolish. game it should give you a bit of. conviction. so you got you got to make that fair and.

I I believe in this. If you want to put. it in a framework or like a moto to. follow, it should be like if you're a. young founder, you should. follow and learn from the journey of. someone who's five or 10 steps ahead. than you. >> Yeah. >> And only take inspiration from someone. who's thousand steps ahead. Don't follow. and take advice of someone who's. thousand steps ahead because his world. or her world. >> Yeah. is very. >> is very different. You have no idea. >> what that world looks like. Also, don't. you think our world is 70% internal?

Like we sitting here together, right? But I see you and you're behind. You're. seeing me and my behind. We're not even. in the same world. We think we are, >> but I have a completely different. experience in the same. >> of what's happening in here and you have. a completely opposite view of it. So the. the reality is I think after a point cut. the noise especially in the exponential. game. Uh and if everyone understands you. you're doing something terribly wrong.

>> cuz people shouldn't be understanding. you otherwise it's not going to be. exponential. So let people think you're. a nerd. Uh. >> right. So was Mark Zuckerberg when he. started. >> but look at where he got finally. So. social media wasn't bought in like it is. bought in today. Many people thought. instant gratification really but the. people will sit and keep swiping. Nobody. thought that. >> Look at the world today. Uh look at the. gen Z and look at what gen alpha is. going to bring. So these generations. were born in it. So their perception of.

so users will also evolve and this is. always in a change. >> Am I making sense? >> So you've got to be adaptable and you. got to be open-minded. These very fixed. mindsets I think are arrogance are. people thinking I know honestly I don't. know what's going to happen tomorrow. morning. Raj in EI and I've spent 11. years. >> Okay. The so-called IIT and whatever you. started the podcast with with all of. that so-called tags. I have no clue what. will happen tomorrow in AI. One lab.

somewhere might change attention based. mechanism to something that you know no. one saw coming and everything's back to. square one and GPD suddenly from you. know where they are to okay now what? So that's the world we live in and. accepting it. um you know if you this. this is one thing I think both of us and. I'm sure you have the same experience as. me should tell your young founders. do not get into the game expecting don't.

be the optimistic founder tell yourself. tell the investors decade. be very. be conservative in your head and the. reason I'm saying that is coming back to. my earlier point the founder journey is. very lonely. Okay, you don't have many friends. You have to fight your battles. Um, and. in that journey, a lot of it becomes. intrinsic. cuz you're not sitting with, you know,

support of 10 people. You. confidentiality will stop you from being. able to access 10 people, right? Uh, so. the real tough decisions are going to be. lonely. Yeah, you may have your board. with you. That's about it, right? Practically for each one of the board. member, it's a lonely journey. So with. that context the world intrinsic. battles become far more tougher to deal. with than exttrinsic and if most. entrepreneurs give up because they give. up not because they needed to. It's not.

as much. Someone else would look from a. distance and say. right. >> but the entrepreneur sitting there won't. be able to take that perspective because. he's put in his life hard work right. even if you become detached with extra. absolutely you put enough effort in life. and energy into it that you will be. attached true. >> and unless you're a Buddha if so you. shouldn't be doing a startup but as long.

as you haven't reached enlightenment. I. think you're going to have that struggle. internally, >> right? >> And therefore, the point I was trying to. say is I do think that. you win or lose in your head. >> And the feeling of win and lose is the. feeling that'll keep you going or you. give up in this journey of as a founder, >> right? As an employee, you can change. the job. It's it doesn't have the. associated story. As a founder, it's not. a job. You it you've committed. you're.

all in in your words. So when you're in. that mindset, what you really need to. make sure is you're able to control your. intrinsic. battles and that's purely expectation. outcome. That's it. And what most people. miss is your expectation is in your. control. Outcome isn't. >> But every single expectation you have in. your head is in your control. So the one. strong advice learned it the hard way. that I tell people is start with a very.

low expectation. expert and an industry expert to come on. Raj Shamani show and talk AI right I. shouldn't do that in second year of. start because if I'm doing it I'm just. faking it cuz I haven't spent the time. to get there so you got to have a very. low expectation of the near term. >> true. >> equally you got to have a crazy. expectation in the long term and what.

I've noticed is humans typically. underestimate the you overestimate the. near-term. they completely completely grossly. underestimate the long term. Right? So you got to re rewire your. natural. tendency and as a young founder tell. yourself.

be very brutally honest to yourself. because what you're doing with that is. tricking your brain as a founder. You're. setting the expectation lowber. yeah great do it earlier that's great. then. your expectation was even lower to. whatever you got so you'll only be more. like motivated. >> don't do the other part because that's. what most people I see get wrong.

>> and outcomes we can't control. >> the 100 things right who the hund 100. things beyond you that will play as a. founder in the journey. >> Well, >> but your expectation is completely in. your game in your control. >> You work with lot of enterprises, right? >> Yeah. >> What is one industry which is dying and. which will die? >> Yeah. It's I I think it's still a. question in my head right now because. see um. let's take one step back. Um there's an.

incumbent. and there's a new age AI native startup. right the classic question for an. industry to die the new in new age AI. startup should ideally make these guys. obsolete and kill them. >> but if you honestly look at the real. world in the industry right foundational. models aren't enough because say a. healthcare in US context one it's. regulated so you can't get. hallucinations so to your point if. anthropic just gets under pressure and. starts lying then in in sue right and.

there's legal ramifications for that you. simply can't do. Um. >> equally in an insurance call the. complexity is exact medical code for a. particular diagnosis and that code will. define everything thereafter and thereby. the general LLM level training is not. enough. So to give you a sense if you. take the typical insurance code medical. code that's assigned in US healthcare. um state-of-the-art open AAI or.

anthropic model will be 20% accuracy. It's not workable. So in each of these. cases you will have to create. specialized models SLMs like you have. done for your marketing team. And the. reality is all there is in it is data. So now let's come back to the. comparison. You have the incumbent with. years of distribution. You have a new AI startup. Ideally who.

should win? >> Uh you would say probably the incumbent. because they are sitting with data and. >> they should ideally convert that. historical distribution into a day one. state-of-the-art advantage versus not. just an open but AI native startup. that's coming up. But the challenge I. see is agility. The challenge in fact. the biggest challenge I see is change. management and these decisions and these. large incumbents while they're sitting. on a gold mine are political in nature. So you'll have someone who's feeling.

like either I I'm not getting to own. this credit so I'll create some chaos. So the human behavior at that incumbent. and the inertia to change and the. insecurities and all of that somehow. make me question whether they will. actually be able to use that data and. win the game. So to be honest, I don't. have a specific answer. I think we'll. have to wait and watch. I think good AI. native startups who execute well and. with agility should be able to beat this. incumbent distribution advantage because.

I don't see incumbents really moving. with agility. Incumbents who move fast, who have a culture where the leaders are. cohesive, they're all putting the. company first, which usually doesn't. happen once the scale goes beyond a. point, right? Right? It's your bonus. that matters, not really the company's. outcomes. And we've all seen that if. you've grown in the corporate world, right? It's and and unfortunately I. think um so the incumbents who execute. well should ideally win. I mean if if I. were running an incumbent company with.

my learning right now, I would win. I. would rather be on that side. But. unfortunately what I see in the world. right now is incumbents have a lot of. change management issues. The challenge. is not getting AI right. challenge is. making the organization change and. there's an inertia to that change. >> and that change management is where. they're losing the problem isn't the. model or the agentic solution it's the. change management which becomes an. obstacle. So. >> so what are you building? >> So right now what we're building is. we're trying to solve that puzzle. We're. saying that hey ideally these incumbents.

should win. So we've over the last few. years built a deep plum platform. Um it. has two components Raj. One is what we. call a central brain. Now if you were to. simply take a step back what is an. organization? It's just decisions being. made right some data being the source of. it. >> If you just reduce every all complexity. there some decisions being made and some. actions that those decisions lead to. >> So it's effectively data which helps you.

make the decision decisions that you. take bases data and then actions that. you take bases the decision. And if you. take any workflow in any company, you'll. see that it actually come decomposes to. these three elements. >> What can AI do? AI can actually help. every human become superhuman. Both in. terms of how much work they can do, but. more importantly, which I don't think. people think enough, it helps human take. better judgments and thereby better. decisions.

Because think about a marketing kind of. context. If your AI is doing all the. hard work and giving you the best. insights right in front of you, you will. make better judgments of that in the. earth team. If you you had a marketing. team and they were just creating excels. and trying to dump some information, they may not be as holistic as what you. see today and thereby you will make. weaker decisions. So there's a clear. advantage of the organizational learning. becoming faster. >> and in the process the decisioning. quality and the clarity of thought.

getting better and that's the. competitive advantage in the new world. that we're getting into. So what we've. taken as a vision mission statement not. a vision. >> it's one of those medicines for. depression. >> but coming coming to the question I. think the the work that we're doing is. effectively going in and saying don't do. these point solutions right now most of. enterprise is saying broad problem. there's a heat map in my organization or. this is where I have 10,000 people I. want to attack that we go in and say.

please don't do that let's think before. we do let's create a central brain. under which all workflows will come in. so that there's crossworkflow business. model level learning happening at a. workflow level the central brain we it's. central intelligence layer but we call. it central brain so it's easier to. understand is actually orchestrating. both the humans and the AI agents in. every single workflow that you on board. it has a data and an ontology layer I. don't know if you understand that but. it's effective think of ontology as a. business meaning to the data that you.

have. >> okay. >> right u so customer is could be a data. table but customer has a business. meaning to it. Customer has multiple. linkages. If you do this to a customer. this needs to happen right a refund to. what what all all of those aspects. So. you map all of it up front and then you. keep transforming the organization and. what you get at the end of it is every. single employee has a chief of staff. bases their role and the access. credentials that they should get access. to. Central intelligence becomes their.

chief of staff. It limits what they can. see and then within what they can see it. answers anything correlated. right it equally has a control tower for. every employee now because it's also. saying I need to orchestrate you now so. what we are saying is human on top what. we're trying to do is make these. incumbents be able to shift humans to. superhuman it's a better way to position. it because then humans don't get. insecure and then in that superhuman. it's true too and in that superhum human.

capability. What we're allowing them to. do is effectively become orchestrated by. the central brain. >> So we're removing all the process. management in efficiency. And you won't. believe it what we're seeing today. Um. and these are large enterprises lot of. them are 10,000 plus employees, right? Um. we seeing that the process fat the. process inefficiency. of yesterday itself is 40 50% of the. jump. just eliminating humans doing this.

middle management layer and instead have. AI tell whom who needs to do what give. the right alerts just cohesively make. the team do their job itself unlocks for. us 50% odd versus. >> but you've not even gone to a decision. layer. >> yeah we haven't even got to the AI layer. and I don't understand why because. yesterday digital transformation would. allow you to do all of that but that's. the world we live in. >> so the first layer is honestly just.

purely getting everything sensibly. cohesively architected and orchestrated. >> process efficiently. >> So that is a big big jump that we see. immediately. The next jump that we see. is we are probably eliminating today I. would say about 50 to 60% of human. judgments. and allowing the judge where humans come. in the loop allowing them to take a far. more informed decision by giving them a. far better construct. The other thing. that we're doing is we're reducing the. complexity because remember I told you.

change management is the real challenge. humanity has to go through this new. revolution. Um what we've realized is if. you reduce the human job. are experts. There's hardly any human to. human inefficiency in that point. So we. are able to eliminate a lot of the. change management by simply making task. clear saying up you relax the central.

orchestrator will coordinate everyone. So you forget the process management. part. >> but this is a place where I need your. acumen. I need your judgment and. therefore and here's the flavor right. here's the various different inputs that. you need to make the decision better and. that's what we call as human in the. loop. So what we're doing is effectively. saying instead of point solutions let's. create first the architecture the. foundation of an AI future and then we. create a blueprint of saying this is. your end state and then we simply say.

let's go transform one by one in a path. of say 2 to 3 years a a deterministic. business to that. Now to your earlier. question that is how we see them winning. over the transition that's needed. The. ones that we won't be able to convince. to do this I don't think have a chance. to stand the AI native challenge that's. coming because if you look at the other. side VCs are going crazy. The funding. that's happening right now on niche. solutions in AI very.

>> wild is wild. >> So these guys don't have time. They're. sitting with the laurels of past, but. they really don't have time because what. they are not realizing is they may have. taken 10 years with a human-led. way of getting there. These AI native. startups are going to spawn 10,000 AI. agents, right? So what you had at the. end of 10 years as human acumen, these. guys can just click a button and just. >> But tell me as a go wild. >> as a founder. Okay, I'm sure a larger.

company will have larger problem. But. let me tell you with a small founder. with two 300 people team, right? >> It's. >> the biggest problem is with AI when I. look think of solutions like these that. can we make our entire organization AI. native. >> and make every team member. essentially give them some AI support so. that they become super human and their. judgment gets better. >> Yeah. I'm worrying that people are getting.

dumber. >> for sure. >> and there's research on it. >> Yeah. So people are so I have seen I've. put AI in one department. >> Yeah. >> Which was a very crucial department. I. made them use all the tools in the world. probably uh led last six months forcing. them to start using this and now they've. become so dependent that they've stopped. taking judgments and they're just. letting AI do the work and they've. become absolutely dumber. So the ideas. which I used to get before.

>> are not coming. >> are not coming. >> Yeah. >> There used to be an element of. stupidity which I really love. >> Yeah. Like until unless there's an. element of stupidity. You don't have an. original idea. >> Right. Y. >> that is not coming. >> Everybody's giving the same thing. They've become dumber there. So my. brightest talent. >> y. >> is now not producing that kind of. results because their thinking has now. been you know compromised by um there. are no two. >> So how do I so if I implement this in my.

company how do I make sure that the. entire organization doesn't get average? >> Yeah. So let's let's actually that's a. brilliant question Raja and I think I. have a very real answer to it because. I've experienced it. First of all um. even if you don't do it here the people. you can hire tomorrow are all going to. be dumber. >> because that's the that calculator made. all of us. >> weaker on math. >> Okay. Um and similarly.

artificial intelligence once it reaches. artificial general intelligence levels. is going to render a lot of our. abilities to think to analyze to judge. >> Yeah. So if if the p the onus is now on. the person to be the ones who are AI. native and the AI native guys won't be. this bad by the way and I've seen the. difference at my scale already where the. ones who are curious the ones who are. more risk-taking outside comfort zone. seeking folks will probably become.

superhuman and their judgments will. improve because they're not using AI as. they're saying AI team member. And there's a big difference in the. mindset once you so if you're. outsourcing what you need to do you will. get slower if you so first thing that. you can do internally is to try and. avoid people outsourcing their work. what they can do is make a team member. in their 12 org obsolete because that.

portion of work is getting done by AI. and if someone's work is 90% being done. by AI um. hard decisions remember so and it's. better for them too because they you're. making them also they there's no growth. and learn learning for them doing that. job so to your point they'll only go. down so ever put my answer and I've. thought a lot about this because I've. had the experience of having to do it at.

Avdar. My realization was that you cannot. control this and there thereby you don't. want to have roles which are getting. redundant by there's no easy answer for. two reasons. One employee. you're not going to deserve that. employee because you don't have growth. for them. What I learned is don't let. humans use AI tools. Humans shouldn't be the ones pulling AI. AI should be pulling humans.

>> Explain. >> So there don't let humans use AI tools. >> as tools. Yeah. Historically, if you see. chatboards and everything is humans. using AI is almost like a software tool, right? >> You got to flip that completely. You got. to make AI orchestrate humans for. judgments and decisions needed. AI.

that's not AI native. You're giving the human the wrong task. What AI native means is humans are doing. tasks which you need to in the new. world. So if word do. that's one place where you can actually. trust Lord. >> So you don't need to worry about it. So. don't have a human who's being asked. create me this word document. The human's role should be if that word. document requires certain judgments that.

are critical and have some ambiguity. where a human acumen is required or has. some sort of a lack of perspective for. your data historically and thereby you. want a human to come and train your AI. that those are the points. So you have. to completely rethink through roles. You. have to completely redefine how humans. play a role. And by the way, you're. doing them a favor because you're. telling them you need to make judgments. Judgments that matter. You're not here. to create that word dog for me. Right?

Right now, what I see is across the. board, leaders are asking team members. to get on a co-pilot. And then, you know. what they're doing is they're doing them. an injustice by telling them become. dependent. And by the way, and by the way, your human in the loop. will be the reason why. So you're teaching it to take your job. So you have to first of all redefine. human jobs. >> I love this. >> as a leader. There's no other way out. >> I love this. >> And that's the advice I give my.

enterprise clients and that's why we're. succeeding. >> Um. >> you just wow. I I've lost the thread of. podcast at this point and I've just got. a crazy idea which I just want to go. implement. >> You can and trust me you will see. immediate response to this. >> which is it makes your employees will be. happier. My I was that is so true. I was. asking. an employee to do XYZ dock. I can ask an agent to deliver that dock. to that person and ask that person to. just now choose the three things which.

are the most important out of this. person. Right? That person should have a. dashboard. >> You are actually telling say an AI agent. saying I need this answer. The AI agent. then figures out who is the right person. whose judgment is required in your team. and it send then puts it in the. dashboard or a slack message or a. WhatsApp message or some sort of thing. Hey, Raj needs this from you. At this. point, I need your judgment. >> and this agent will do automat. And in. there, what you've done is you get you.

don't need to do random stuff. You've. optimized your time. You're putting in a. prompt somewhere to your central. intelligence. >> My agent is like. >> right chief of staff. >> That person doesn't need to talk to you. on a Friday evening because the agent. already knows the entire knowledge. graph. So he can ask questions to that. eliminating the time wasted between you. two which can now be offline. So you can. be remote, you can be in the opposite. sides of the world. >> How expensive is this? >> Very very cheap. This is very easy in. today's world.

>> And how much time will it take to. implement? Let's implement. >> less than a less than a month and it's. just inertia to change. This is exactly. what we do at Avatar. So what we do and. there's one more thing that we do which. is deepse like what we do is we've. realized also that high transaction. volume use cases the costs of an. anthropic and open are becoming. prohibitive. So we've figured out a way. to create these small language models. like you which simply make the data be. your competitive advantage. So data. never goes out and those that data keeps.

feeding back into these small compact. footprints that saves a lot of token. costs. So these are self-hosted within. their own garden and they simply don't. have to pay any tokens to the last. >> But is it is like deep as good as. anthropic? I don't think so. >> But the distillation mechanism you see. distillation is simply saying there's a. teacher model there's a student model. Student model smaller. >> teacher model obviously knows about what. Trump is doing in Iran. >> But then my student model only needs to. know marketing in your case, right? M.

>> so I will only distill marketing related. reasoning. >> which is in technical terms called. trajectories. It's input output based. chain of thought in the middle. Right? >> So you're distilling a task specific. intelligence from a large 3 trillion. parameter models at the top frontier. down to say a 30 billion or a 7 billion. >> ranges of that kind of gemma. >> But for me the large the teacher is more. important because of content world. >> Yeah. For you would want. >> it's not doain specific. I want that.

>> cleaning to be done from that three. >> until you have data. >> even after data. >> no. >> we need those large things to then for. specific context. >> then marry our data to find out what's. the true we need both but larger things. will happen on our data. >> yeah but the decision engine can slowly. become SLMs so what's evolving now is. something called mixture of experts. >> but then if a go ahead. >> so mixture of experts effectively means. that say decisions you need to make in.

your new AI native world is 20 humans. >> and say 50 decisions. >> The 50 decisions can each be a. specialized decision specialized model. >> Yeah. >> And there is a central router that bases. the decision is routing it to that. model. So it gives you the breadth of. your business but it brings the scope. down to this. It has multiple. advantages. less hallucinations cuz it. doesn't even know what's happening in. Iran. So it's not going to hallucinate.

in trajectories that are beyond that. given task. So it reduces hallucination. Second inference costs it's probably. going to be 1x 10 the cost of what you. were spending otherwise. Third most. importantly is accuracy. So you will see. that in the critical decisions what's. your competitive advantage versus. someone else uh competing with you. It's. the decision making ability. >> True. >> Which is nothing but data. So you want to acrewue every human. judgment that's happening in this new.

world as a judgment. All of that should. go back into the same SLM and it's only. yours. So what you're in other words. doing as a leader is to say that I'm. going to make my humans take more and. more mission critical judgments as I. look forward. I may not have the. linearity of growth as I used to have. yesterday where my employees have to. grow linearly with my revenues. I'm. going to flip it by saying my AI will. make these guys more and more superhuman. and thereby will start being able to you.

know address more and more demand. So. you're creating nonlinearity and by the. way you can use it for cost savings. which most leaders. tend to take which is layoffs and all of. that that we hearing but the real studs. that I see are actually thinking of this. as nonlinearity saying that okay maybe. if there's fat or if there's. non-meritocracy I'm going to fix that. but the real opportunity for me is. nonlinear supply for demand as I can. decouple supply and demand equations of.

yesterday. >> and so I'm going after growth. >> so that That's where the successful guys. I think and some some case studies I've. seen in the last two years is like. phenomenal. Of course, we have a hand to. play but I can take the horse to the. water. Right? So the real start leaders. are today not looking at AI as cost. savings. >> The start leaders today are saying AI is. nonlinearity. >> There's a growth engine for me. >> and then they're realizing that the only. way that will happen is a true AI native. framework. In a true AI native.

framework, no human should be doing. anything that an AI can do. And so we call it humans on top of AI. So humans should be just the judgment. layer. The process. uh you say you want to podcast shind, right? You need some briefing pages on. what did some briefing. Yeah, >> you should not talk to a human. Do you. think humans in today's world deserves. being given that task? Yeah, >> they are also actually going to some.

model and asking this only. They are not. doing the hard grunt work which it used. to do. >> Instead, what if they actually see. certain aspects which their acumen will. truly add and then pick up. >> Yeah. Raj, you've freed up their time do. instead of making them do something that. tomorrow is going to eat their job on. You're putting them on the things that. actually matter the most because those. business judgments and the value of it. is never going to go. >> Yeah, I agree. So you're shaping stars. of tomorrow by flipping their job.

description to what's actually going to. survive AI disruption. You're making. your organization therefore off. >> free to the extent you can. You're not. gone, right? Nuclear war. It's a. different state, but to the amount that. you have in control, you're giving the. best chance for your team to become. >> um successful in that era. And each. individual is now winning. You're. winning because you don't need to waste. time explaining basic stuff. The guys. who are coming joining new have a friend.

that they never had buddies buddies try. But in the human world every incoming. employee feels lonely for the first 2 3. months and then there's a induction. You. can try your best to fix it but there's. always friction. In this case the person. gets to know whatever question they have. they can ask. You know what I'm seeing. right now in Avatar the people find hard. asking silly questions to humans. They. don't find it hard when it's an AI work. >> True.

>> So people have started asking the. questions that matter. True. >> for their growth and. >> where are they getting answers from. existing data set or combination of. claude GPT Gemini Deep Seek bunch of. things together or is it one? >> No, you can't go one. Um. >> it's like a multi. >> Yeah, it's a multi- aent system. U each. workflow has multiple agents. They're. reusable agents like a marketing and. merchandising and e-commerce will have. sharing stuff. >> So there will be reusability but.

>> so they all will have like a access to. data set of the company. >> y. >> data. So each employee will have a role. right a marketing team will get to see. >> a lot. >> the marketing funnel. Um a finance guy. will see a very different set of the. data. The central intelligence at the. CEO level will be entire. So each layer. and each functional so in the matrix you. will have exactly the chief of staff. automatically because it's cutting the. data view that you have based on arbback.

which is role based access credentials. and then it's cutting. your um asks by a nondeter you can. natural language ask any question you. want and if you have access to it you'll. get the answer even if you don't the. limited data whatever you've asked. you'll get the best possible. >> but this is this would have been. possible before also no before AI also. this was possible but people didn't do. it probably because it's easier now. >> no I think an analysis and understanding. and reasoning was missing and a lot of. these things.

>> only data was coming. >> and then why this is coming this the. reasoning was missing. >> yeah so I'll tell you the big difference. I see is it's from dashboards to. actionable insights. >> earlier humans had to do that. >> actionable experiments this will give. that okay these are three things. probably which you can. >> you can act on the judgement of acting. is being left to. >> out of three which one should I put. first is a human layer. >> and not just that you in your case you. can actually then tag the human action. to outcomes.

because finally this action will go to a. podcast and how it converted. So you. could start correlating without any. human in the mix saying that decision. did this decision help the outcome. eventually. and therefore the constant learning is. implicit because the next time the same. employee comes he's not being given that. option because it didn't work say 10. times in the past. So it's nondeterministic. It's an. intelligence that's actually. understanding the context, the domain,

the actual workflow task and correlating. it to the final outcome. Could be bottom. of the funnel, right? You went podcast. live, but you can connect the data of. the performance of the podcast to a. decision that was made on day one of the. pre-shoot schedule that you had. And. that is very very powerful. Um the and. what that does is create nonlinearity. again for you because. >> what you're then doing is tomorrow you. could do 20 podcasts with the same team. >> What you've also done is earned your.

right as a leader for them by making. their role descriptions become AI. disruption free. So your team is now AI. native. And in fact that's what I tell. clients. If your organization has a. single job description that can be. disrupted by you're not native, stop. fooling yourselves. You're just kicking the can down. True. >> Right. And then you won't be able to. compete with that AI native starter. because that guy doesn't live with. these, you know, uh degrees. He has. degrees of freedoms that you call.

constraints, right? You're telling me. out of all the data. >> Okay. The AI will start giving you. documents. AI will start giving you. possible experiments to run and AI will. give you reasoning behind it. >> Y right. >> And it can go one step ahead. It can. also tell you what works, what doesn't. Ah and it'll tell you the output. It'll. match the output. Right? >> Now this will all be done based on your. past data. So a first you're missing the. newness where you are just like absolute. random new thing you will add which will.

not be there in data. So that's first. thing that you're missing. >> Second is month one it'll be great for. every employer. >> Y. >> month two it'll give you predictable. thing. Y. >> month three it'll start hallucinating. because the data will be so much and. context will be so much. >> that context will start getting mixed up. >> and it'll pick up few things and start. probably hallucinating. >> that's an easier problem because that's. a technical problem and there's answers. now. >> no but like how do you the. >> so here's here's how you do it so. >> do you keep giving new context every 3.

months every 1 month every two months. >> every second every action refreshes the. context and it has to be automated. >> but then will will it not mix all the. context together and start giving you. how does it know. >> that today the research that I'm asking. for. >> y. >> okay. >> is with the context of today. >> it doesn't need to mix the context that. I gave for let's say Andrew Hubman yep. >> right it'll mix because these are two. different types of people and problem. that I have to deal with. >> y so the way you do that is something.

called knowledge graphs um so there'll. be a at a customer level knowledge. graphs they'll be at a functional. workflow level knowledge graph. And finally when you're talking context. you're getting context is nothing. knowledge graph in simpler terms is um. all the input output the data that. you're referring to stored for a. particular context. Now contexts are. varying right marketing is a different. context similarly leadership at the CEO. level is a different context at last.

mile is different context. So context. vary bases your role bases the function. bases the exact task. Now what central. intelligence does is it persists every. action. It then goes and says maps every action. via ontology to the relevant entities um. in the business. Customer is a. >> ontology object. >> So it has all linkages that are. happening. So ontology is effectively.

objects like customer. >> It's relationships and then the actions. you can take. So you structure all of. this up front in the central. intelligence layer. Memory thereby the. knowledge graph is actually different. layer tiers of memory um the basic. episodic and semantic level but then you. can go deeper by saying I'm going to tag. it to the framework that I have by. saying for this action for this object. this is the memory structure. >> Got it? >> In there you can build rules of saying.

recency overrides history. M. >> so effectively for example you've. learned a new thing today. >> or a human overrid. >> experimented something you never saw. >> as AI. >> and it actually produced good results. you want to pipe that back as saying. this is a new learning. >> it should override anything that. conflicts with it previously so you'll. put rules business rules effectively we. call it safeguards and guardrails but. the idea is you can put a rule saying in. this particular context if I learn.

something new and and there's a. conflicting history, clean up the. history. If say I need in a particular case you. want more human originality, right? Um. you know that my AI ready is not ready. but you still want this capturing of. data. You also want the little bit of. productivity unlock that AI is going to. give. So in that scenario you will say. that AI will not recommend. AI will give will stop at assisting.

>> but then the humans will have to now. take that data and then judge and in. that variations of judging you are. creating the data that's needed for. tomorrow you being able to elevate. yourself from so you can put multiple. different workflow level guardrails. depending on where the AI accuracy is. depending on where the team right now is. >> but then the AI between all the. different workflows and knowledge graphs. >> won't create a certain level of latis.

work between them to come up with a. better decision which is combined. It. would. >> why because I'm writing a rule that. >> X outcome if I want. >> then get access to only Y knowledge. graph. >> y. >> right. >> but as a human. >> because I have seen layers of let's say. 20 different knowledge graphs. >> whenever I take X as a decision I'm. biased towards accessing one knowledge. graph but my decision is influenced by. 20 other knowledge graphs as well and.

then that's how I take decision if I do. a hardcore rule No, it's not a hardcore. rule. >> Then I won't be I'll always stuck to one. and I will never be able to come up with. an original scenario like this. You. know, do your hardcore rule is. weightage. >> So you're effectively saying here's the. prioritization of these 20 knowledge. graphs and there's a precedence of who. overrides. >> So weightage let's say 60% weightage. I'll give to this 10% I'll give to the. old one. 10% I'll give to the recent.

trends which are happening. >> and 10% will be just something random. new which will come out from some. pattern which I which we don't even care. about. >> y and you can keep it open and see you. have full flexibility to use human for. whatever judgment and the judgment right. now in a particular workflow could be. think of what we should do. in which case AI stops at allowing. giving you everything you need to think. and then captures what you decide I did. in certain cases like create a dock you.

will. >> let AI do it. So there is a varying and. you there's no one sizefits-all. So I. can't give you a general principle. because you have to work backwards from. the true north. What is the true north? The outcome of the business. >> So it has to be an applied AI mindset. You can't take a foundational or an AI. mindset to it. You have to say what is. my true north? That's a problem. statement. And then say okay how do I. architect? >> That's true. See I I I see where we are. going. >> and there is some merit to it. There's. there's a merit to experiment this. Maybe this works.

>> No it's working. It was working in US. healthcare. >> Okay. In a in a highly regulated. >> patient health information like kind of. context which is legally regulated. >> No. So I understand I I I am not saying. that this is not working. I am seeing. from a business lens of lens. >> Okay. You mean from a. >> from my lens of figuring out figuring. out perspective? >> Yeah. I'm sure. >> So I'll be a little skeptic as a as an. entrepreneur right which I should be. So.

tell me I see merit and I see certain. level of risk where I can my whole team. and me can get dumber. >> y. >> but there's definitely merit to try. because it's obvious what you're seeing. is obvious why more and more people are. not doing this. >> people are starting to this is uh. >> because this should be the most obvious. thing for any organization which is of. decent size like and decent I mean. >> if you have crossed like let's say first. couple of. >> 20 30 40 crores You should just hop onto.

this. >> Y. >> forget like 20,000 cr or 50,000 cr after. your first 10 12 crores where you have. reached a certain level of margin where. you can you know put in experiment. See, I'll give you an answer. >> why people are not doing it. >> I'll I'll maybe rewind us back to where. I started my career. Internet uh.com. >> See when when I started in n I was 99 to. 2003 in it Bombay. This was the birth of. internet practically. Um I still.

remember in my first and second year. HTML coders were getting paid a million. dollars back then. >> Okay. And similarly today u some of the. best ML engineers would be getting paid. 2030 some packages are even 100 plus. million right if you're talking an Ilia. level talent. um all of that is going to get. democratized. to come back to your question the reason. why it's not happening right now is you. need some serious deep plumbers to get. to this but I can guarantee you in 18.

months like today does anyone think of. HTTP or TCPIP We no do does anyone even. need HTML coding? Cloud will do that for. you. So you know the world has come to a. point where what was a million dollars. back then and very tough to do and. therefore only 3 to 5% of enterprises. succeeding. will get democratized where it'll be as. easy as telling claude just create me a. website. Um that journey is not done.

yet. So if I were to draw the contrast, I think we're somewhere in the a few. months before the comish. position, I see a bubble um on. valuations. I equally realize that this. is the most transformative technology. I've ever seen. So I have a slightly. more middle view on this. There is a. bubble and I think whether it corrects. in what form it corrects is a separate I. don't have a crystal ball there. But. it's there's certainly a bubble. Um. equally is it is it just hype? Oh no,

please I see results every day. And the. gap between the hype and the bubble is. effectively not everyone having. democratized access to the what we're. speaking right now and the talent. availability right now is limited. >> It's not like everyone knows these. secrets we're speaking. Um having said. that I do think central intelligence has. become fairly accepted now as the path. I do think many industry folks have. realized we will need to get into one.

protecting our data so that we're. competitively. >> there's a mode in life. >> second the data will be needed if I want. to have task level accuracy for my. industry use case. >> so there's no foundational model and AGI. will solve everything so everyone's. already started moving where I see the. gap still is a lot of companies are. choosing to do point point solutions and. the real magic like you would probably. understood today won't happen if you. take a point solution view. You have to.

take a organizational view. If you want. to be AI native, you have to reook at. the job descriptions. You have to take. that little bit of an outside comfort. zone. You may do it in 3 years. You may. sequence it and not create panic, right? Do it smoothly so that every human's. taken care of and is upskilled to the. new role that they need to get to. All. of that is viable, but you do need to. take the hard decision of saying, I need. to step out of my comfort zone. look at. the organization. So the two reasons why. you don't see everyone succeeding here. is one the natural inertia that human.

leaders have and second I would say it's. the lack of democratized access to this. capability. But today if you look at any. valley talk uh a lot of what we're. talking is becoming more and more. accepted uh across the board. >> Yeah and you're right open models like. deep sea aren't the answer. So I'm not. saying foundational model isn't a game. It certainly is a game but an applied AI. is act equally a game. Look at. Palanteer's valuation and we compete. today heads on with the likes of.

Palunteer in some cases winning. So there are the people who are winning. know the secret and that's why even MIT. right. one and a half years back MIT put this. paper that spooked the whole world okay. saying I think it was 95 um but we. should correct me if I'm wrong but. somewhere in the 90 to 95 97% pilots in. enterprises are failing. but what people don't realize is MIT is. equally acknowledging there's a 3 to 5%. winning now who are they and what are.

they doing and if you look at that. bucket and see the pattern. There will. be one certain pattern in all of those. AI is not some, you know, putting a band-aid on a. sinking ship. AI is being treated as okay, I need to. go back to the workshop. >> Interesting. And some companies are. taking green field views now. >> which is the new evolution I've seen in. maybe the last 3 to 6 months where. companies are saying maybe I'll never be. able to do it in the brown field way so.

let's create a competitive company. these guys shouldn't know all of that's. happening the green field will get its. own funding and full degrees of freedom. doesn't need to. >> can use the data and the distribution. but beyond that doesn't need to follow. the rules of the game. >> and I think that that might also show. some results in. >> again early days we have just gotten. into it in the last 6 months but there. are companies big large companies with. billions of dollars of uh annual revenue.

saying you know what maybe I won't be. able to win this by trying to shift this. and change the current state to that so. let's just start another startup within. and give it the advantage of the data. and the distribution we've had. >> and let it have the strength of an agile. startup Right. >> I I right now I'm seeing multiple large. fortune 100 level enterprises taking. this decision internally. >> If a young founders watching. how to raise funds and tell me the game.

don't tell me that uh be this do two. things and three don't give me the you. know basic founder definition. Give me. the IB game that you would be playing. before making sure that your valuation. goes up. I think for me what my learning. has been over this journey from a. fundraising perspective is first get. this myth out of your head as a founder. that capital is the reason why you'll. win or lose. >> True. >> In fact, capital can be very toxicating. for an organization if you have more.

than what you need. >> Um and if you don't know what to do with. capital, don't raise capital. >> Agreed. >> So the first thing that you really want. to figure out is how do I make a rupee. 10 rupees right? And you don't need 55. million for that. You just need a little. bit of a corpus that you could play. with. Um you obviously need to be very. honest to yourself on. >> things being twice the time that you. imagine cuz you know there's we always. in my experience we always overestimate. the near-term and we underestimate the. long term. Uh as humans I think uh at.

least I I can talk for myself and I've. seen the people that I know well all of. us tend to always overestimate the near. term. So if you're thinking you need X. dollars, be honest to yourself saying. that I'm I'm an optimistic founder and. you know my execution as much as I want. would involve people and as people grow. inefficiencies will come in because. they're all rightfully their own. opinions, their own and that's the way. it should be. That's the beauty of a. team, right? Um and so apply a discount. factor to come up with the real number.

first. ensure that you know how to make. some value out of that money before you. even think capital raising because if. you don't have that sorted capital. raising can happen there many good. storytellers in the world and many. people who've done some amazing rounds. that I can't right um so there is always. the potential of you doing a fund raise. but what I've seen in my Wall Street. experience as an investment banker. watching tech CEOs um and now as last 11. years as a entrepreneur myself.

if you raise capital on the wrong. reasons um for fame and for. >> PR and there are many other advantages. to. >> but do people raise funds for just fame. how. >> yeah because. >> like why take external capital for fame. no it's not about absolute fame I'm not. denigrating anyone here. >> but sometimes what happens is the. relative competitive dynamics are such. >> that you are put into this prisoners or. the leadership dilemma or the prisoners. dilemma there you know it's a game. theory end of the day. >> so you're just one part of that game.

theory and you've got to react to what's. happening in, right? Um and sometimes. these are real constraints that one. loses their own clarity and vision on. Um so the the ones that I've seen are. all genuine. The ones who probably would. have done a bit of this are all genuine. They're doing it for the right reasons. in their head. >> No one does something thinking I'm doing. something wrong. Right? But what I've. learned is don't get into that zone as a. founder. U be very clear saying that. And what it does is it gives you. conviction. And what I've seen is when.

you have true conviction, earned. conviction, right, with clarity of. thought with um some sort of a karma. that's given you that clarity of. thought, which could be say learning. over X years, right? Um you would. probably story tell much better as a. person than you would if you were to try. and fake or you know project some random. appearance. So if you can get there, that's ideal. And if that happens then. honestly fundraising is about making. sure that you have a investor view which.

is a ROI view. These guys are not here. for incremental outcomes. >> So get it very right. You've got to show. them a 10x or 100x right return. Um so. think addressible market. Think about. what that journey would be from the. capital that you're raising to that. point to the extent possible. have as. much clarity on the tough known un I'm. not talking unknown unknowns but at. least the known known. >> and to the extent possible known. unknowns and unknown known is where you.

want to get crisp about oneline elevator. answers. >> without complication cuz what I've seen. is people get into storytelling which is. very very verbose um. >> it's far better to just have oneline. answers um it shows clarity uh it shows. conviction. it's easier and then pause. All right. Let the other guy, it's okay to have. that uncomfortable pause in those. meetings and let the other guy absorb. If needed, spend the next 1 minute uh.

which is uncomfortable when you get used. to it. And the moment you start doing. that, discussions become far more. streamlined. Interest becomes far more. crystallized. U you need to have a good. story. Of course, uh the story is. important. Don't take me wrong. But if. you have the right ingredients. underneath and if you have done the work. needed for you to be convinced first, I. think it's quite straightforward and. there enough smart people who see an. opportunity. Let's flip this, right? Um we're talking. founder lens, but think about say a VC.

>> You think he's getting great ideas every. day. >> He's probably looking at thousand ideas. a year, okay? Out of which 10 are. probably worth. >> So if you can fit into the 10, right? >> The game's already done. It's the it's. switched the hourglass is flipped. >> They're the ones who need you now. >> And by the way, you have 78 funds to. pick from. So. >> never never think of you as um getting a. favor from the VC. You're giving an. opportunity.

>> It's it's their job. >> True. >> Right. So if you could just flip that. little bit of a mindset of saying I'm. the one receiving money. No, you're the. one giving someone an opportunity to. ride on you. >> And these are equally right. It's half. class full, half class empty way of. looking at the glass. But. >> you would probably operate much better. when you have that little bit of a self. uh conviction and on that. self-conviction. um you know it's as much your loss as. it's mine. >> True. >> flavor to the mindset of yours. >> So you you don't believe in having this.

thought process that this is a. responsibility and they've done me a. favor. It doesn't help. >> if Sachin Tendulkar gets on the field. and starts thinking oh my god if I go. for my cover drive and if I miss the. ball and hits the wicket India's gone. right and they were at least in my. childhood they were at least five six. years I remember where suchin goes off. TV's off. >> right so that's a real pressure it's not. some fake today we take care of it but. back then it was that situation.

>> if suchin goes in with that burden he'll. lose the potential he has u So he'll. start having second thoughts the moment. and timing is gone. Right? Uh if you've. played cricket. So you have to be in the. flow and these mental you know memory. oriented constructs that we carry don't. help you take good decisions in the real. time. Um. >> it the cleaner you are the less cloudy. you are the less such baggages that. you're carrying. Um, and there's. something my mom, my late mother used to.

say, which is you're on a train, um, and you're carrying the luggage. >> and carrying it all through the journey. while standing on the train, right? That's exactly what we do with our life. We don't choose our birth, right? The. air that we breathe is not something we. have any hand in. The food that we're. eating isn't. Um and we keep thinking. that we're carrying the burden of. everything in our life and taking every. decision and you know we over complicate.

that out of that little bit of an. illusion that I am God which we aren't. we know. >> so I think if you stop doing that to. yourself within right there's no without. here but within if you can just re have. that little bit of a grasp of saying I. don't control everything I certainly. don't control outcomes what I can do is. use my time for something productive. towards that direction of an outcome. which is my karma. Right? And the moment. you do that some of what these. additional tags we carry.

my responsibility all of those are. egocentric things. Um but the moment you. build any real company. >> and you've been in the journey for. >> 6 months or let's say a year. >> Yeah. True. Can I hear you on your. ground as also. the moment your first big round gets. done right or we had a viral moment with.

warrior recently the next day morning. your ego will. >> kick up saying oh you know what I did. that. >> one week later you'll be down again. you'll know the reality will hit. >> so I think it's better to be in that. little and you equally don't want to go. down. >> what I'm realizing and I'm still. learning so you know take everything. that I'm saying with a grain of salt cuz. I'm still in the journey but. >> my exper experiences. Ride the highs. with humility. Ride the lows with. confidence. >> Mhm. >> Is the mantra. >> Yeah. >> Um don't get into the trap of the the.

noise. Um. >> be like Arjun, right? You got to be. focused and then Krishna is your si. So. >> you know the universe will take care of. it. And you know there'll be things that. are magical that will happen if you just. let go of this little bit of an illusion. of doership at some level. >> True. >> You said elevator pitch like you need to. have oneline elevator pitch. What's your. oneline elevator pitch? >> So we at some level believe that our. human eye and brain and the combination. of it will be replaced with technology. and we are there to do it.

>> How I is a camera right now. Uh brain is. AI and together I think physical world. and how we interact with the physical. world will redefine itself in many ways. >> Nice. And that's so that's how you get. in the VC office and then like okay not. I don't fake this Raj no I'm asking I. can't do that if I don't have conviction. >> obviously without doubting your. conviction I'm asking this is how you. get in this my line for first principles. decision making.

because that's when the pivot decisions. and all of those right for external. world it'll feel like oh he's changed. something for me it isn't because it's a. continuous journey of saying I'm here to. figure out the confluence of human eye. and brain. >> and I now see technology can do that um. started with Star Trek when I was a kid. right then Matrix then Avengers if you. see Hollywood's been predicting this for. a while so there was obviously it wasn't. my original idea I've seen Hollywood.

content which has been predicting things. like this for a while Star Trek I still. remember I I'll the buzzer which people. would uh communicate with first mobile. phone I saw in my life um and 20 years. later. we have we have it in our hands. So I. think at some level I would say that um. you need a first principle that you. don't violate. >> I think the rest of the vision mission. all of those things right are great to. tell the external world but within you.

to make the right decisions and say a. turbulent time or a peace time you've. got to have a first principle that you. don't violate. That is so true. >> And what it does is I I think people. underestimate this, but in my journey. over the last 11 years, what I've. realized is compounded learning is the. is the gift you get free if you. >> if you stick to those first principles. >> Yeah. And then you start connecting. >> dots. >> your entire like your lattis work. becomes. >> around this whole first principles.

thinking and that's how you start. thinking better. >> and that's and. I wasn't even passionate about it to. start with it was just a very left brain. but what I've seen is over the time and. I'm a very right I'm ambidextrous so I'm. half right half left in my inner world. >> and what happened was as I put effort. the belief was growing. and my right pain just joined the game. >> Nice. >> So it it didn't for me it wasn't about. some purpose or some passion. It was. very evident. I just as a background I.

was a Wall Street investment banker in. the early 2010s as a tech investment. M&A IPOs uh and similar offerings and at. in similar time was when Google was. chasing deep mind for an acquisition and. so it was a very leftbrain realization. that it's a matter of when this happens. not an if and that was because of. extrinsic stimuli it's not some original. thinking right uh but then I realized. that I I want to be early to this game I. want to do something which is a mix of.

creativity and hardcore left brain. >> combination. And that's when I started realizing you. need some specific thing that you are. excited about. Um and back then Iron Man. and all of that was happening. Um it was. just starting the Avengers. >> And if you look at say um if you really. want to create an Iron Man, you'd need. to understand the world in its entirety. which can't be just text, >> right? Just understanding internet. doesn't get you there. So you'd probably. have to have multimodel AI which means.

that you need to have somehow image, video, 3D, all these modalities so that. when you're seeing the world you can. recognize, you can reason and then act. on it, right? Um and understand, reason. and act in other words um and at some. level I started realizing that right now. everything's happening on text. >> but the real future and the frontier. would be what we now call as world. models which is what Dr. Philly is doing. with word labs. She's a Stanford. professor who's exactly on that vision.

>> It's funny how um things came to play. I. think there's a lot of uncontrollables. here. So it's easy to look back and say, "Oh, I was visionary. I was I had no. clue. I was taking a big risk leaving a. very cushy job investing my money and. taking zero salary for at least me and. my co-founder took no salary for the. first 3 four years because it was seat. funded and we wanted to just make it an. R&D. learning exercise. So looking back it. looks good. Um back then I think it was. just purely saying I want to do.

something which breaks the frontier. Uh. a deeper purpose I came back from I came. back to India because of my parents as. much but equally I my. stint in US was Microsoft then bought an. MBA then Wall Street for 3 years. Um all. through that feeling I used to feel like. Indians are viewed as IT folks. Uh. unless you wear your dimple tie on Wall. Street people like a laptop taker. So. there was a little bit of a dissonance.

inside saying that hey we and if I look. at say I versus Watton I had crazy smart. people around me right so the talent was. visible. I've been in both sides. educationally to realize that hey we we. can compete head on right. um but it. wasn't showing in terms of how people. were perceiving the talent. So it there. was somewhere inside a deep purpose. saying can we be a product nation and. tech creating an innovation out of India. that vibrates on the opposite side of.

the planet right can we create something. like that coming out of India. >> so we started with a product mindset and. evolved evolved evolved to where we are. today. >> you know here's what I want to. understand. which will help a lot of people right. now who are probably in their journey. right so. >> you at Microsoft. >> y. >> then Warton. Yeah. So top IT company, top business school, then top salary and. top money in Wall Street, right? You had. your career set. If you come from a. middle- class family, this is a dream.

life that you're and it's a stable. career. You don't come from family of. entrepreneurs where you have you're your. innate is to take risk, right? So you have a set career. You're making. in dollars. You're minting money. What sign did you see like what made you. take a risk and leave all of that in. 2014 and say like I want to start a. company? What was so convincing that you. leave your entire or you bet your entire. stable career? >> See AI is 60 years.

>> People think it's Chad Gupt moment but. there's a long history to AI. >> and. >> I I would say it's generations that have. got us to this point not one company or. one person. I was fortunate to know a bit more about. it back sitting in Wall Street and. Google and Deep Mind was happening right. in front of me. It was one of the hot. deals in the mix. >> Um I I would say that at some level it. was just a left brain clarity of thought. saying it's a matter of when rather than.

if and that was purely mathematical. >> What was when? Give me one or two signs. make it simpler for me. >> Yeah. So world so far was deterministic, right? Everything was based on rules. um. you could create great software but the. software is a bunch of rules codified um. that's what effectively digital. transformation meant. where science started showing is. intelligence could be fuzzy that's what. was called as fuzzy AI which became the. nent non-generative.

>> AI used to AI was called fuzzy. >> yeah it was effectively being uh done by. mostly manufacturing folks back then um. primly around figuring out how do. predict. not so straightforward rules based. predictions for example how do you do. certain aspects. >> so it's like if in 3 hours I want 100. pieces to be made. >> how many workers at what time will I. need that is what AI would help you. >> yeah as simple as that.

>> as simple like backtrack it. >> is very basic or I'll give you another. sense shipping fishing industry right. these ships would have 30 40% of the. inventory get rotten by the time they. reach. So AI needed to predict where the. temperatures could be a problem. If. there's a problem then you have to take. it out because it spreads very quickly. So what is the mitigation if the event. happens and things like these now that's. not a straightforward equation. >> You can't predict every variable in the.

so you will have sensors which are. detecting and giving realtime signals. and an intelligence behind which is. trying to. >> do a good job. >> So this was happening in 2003. >> Yeah. I think at early days of fuzzy AI. this was effectively where it started. Fuzzy then evolved to neuro fuzzy neuro. then evolved with neural and I think the. the paper that matters is attention is. all you need um from Google labs and ent. GPD is based on that everyone uh it was. a tectonic shift that happened 2019 or.

>> what is tectonic shift explain. So you. know the way every AI lab operated was. completely different. That's a pivot I'm. also talk referring to because what what. Google did was effectively introduce. something called transformers underneath. which there was an attention based. mechanism but the idea was now you're. able to actually predict the next token. in a way a language word or in other in. simpler terms um that it almost feels.

like intelligence. right um it almost feels like this is. thinking. >> it it started giving that flavor of oh. my god this almost feels like a human. kind of intelligence. >> until then no one ever felt that. >> okay. >> Raj so. >> before that nobody. were attempting um the and I think that. was a moment when Turing's test became. obsolete because you can't apply Turing. test for this kind of a tectonic shift. in how determinism has shifted into.

non-deterministic intelligence u and. what nondeterministic effectively means. is you won't come with the same answer. every time because you're not applying a. set rule-based approach. You're applying. >> multiple huristics and you know the you. could be getting into predictions, you. could get into estimations, multiple. real world use cases but. >> intelligence is now not straight. >> simple equation. It's not algebra, it's. calculus. >> That's how I explain it. Um because if.

calculus has that element if you if you. gone deeper into math so it's algebra is. the deterministic approach. calculus is. maybe the closest I can see. mathematically to express the. non-deterministic intelligence part. That I think was the moment where the. world shifted or we also shifted and. that was our moment of hard decisions. cuz things were going well. >> We working with like we had gone deep. into 3D and computer vision. >> Um we had launched we were working with.

Amazon for example which was our first. million dollar contract. I remember. >> which year was this? >> 2017. M. >> uh that was the reason SEO. gave us a term sheet to start with but. that was our first real validation and. Amazon is a very credible buyer on the. other side. Um they have the best tech. >> teams they have. So it was a credibility. on our team credibility on our vision. >> and we we started helping them do these. you could drop your couch in your living. space and you know actually look at a.

life-size product evaluation sitting at. your home. >> and also look at how does it look in my. room and things like that. So we helped. him launch that that went fairly um. large scale um globally. but what I'm. it wasn't easy to then look at this. paper and shift and that's the pivot I. was referring to uh but that I think. that the evolution that I saw was more. happening to me Raj and I I tend to find. it very hard to believe anyone who says.

I did it. >> cuz just imagine the uncontrollables. here right. >> there the Silicon Valley academic. inflection uh which is happening in a UC. Berkeley or a Stanford or MIT we we. today collaborate with all of them. they're doing academic level. state-of-the-art some of them are. qualified for the industry levels which. is where we kind connect the dots right. so the journey starting with some PhD. sitting in some university for all you. know who's just randomly trying an.

experiment and what you need to do is be. able to have the connections to connect. multiple such dots that are evolving day. by day. So it isn't really original. Uh. it is a illusion if I say Avda creates. everything original. What Avda does or. even an open AAI or any any AI lab in. the world, right? What we're effectively. doing is connecting the best dots, >> right? And those dots are individual. moments of excellence by some great. human being somewhere. I may or may not.

even know them but might be playing a. role. So the journey is honestly that of. seeking knowledge um and then applying. it in the most integrity and truthful. way you can for every problem statement. that you're identifying ahead. Right. >> I've had great time with you. I loved. having conversation but we miss Vary. >> Oh yeah. >> Yeah. We went on the flow. >> Yeah. So what's Vary? Explain me. So yeah, this is the other side of Avdar.

where um you know we also wanted to give. back. So we're part of the India mission. and a lot of credit to the way they've. gone about it. Um this set of different. companies that have come in to try and. create the inflection for India in the. journey of AI. Um there is a clear. sovereign AI aspect to this which given. the geopolitical situation right now has. its own standing and right of its own. need. Um and then there is an applied AI. aspect which is how do we democratize AI.

to citizens to MSMES to you know. effectively the 1.4 billion plus Indians. at scale. Um. >> and we kind of playing both roles. Vary. is our first uh release and it's India's. first video model um which is part of. the India mission. We went viral I think. about 6 n 8 months back. Um it was a. press release with Matei. Um and the. idea there was Serv's done a great job. on text. Um Bharaj Jan is another great.

company that's done a good work there. Gani and others have done great on. voice. How do we now expand the. modalities? So Avda's done now video. We're about to also release image. models. Eventually all of this is towards an AI. that can compete um with the world model. that valley is talking about. So all of. these will be 111 pieces and eventually. I hope that India emission gets to the. world model level focus still early.

days. um what we achieved um honestly. Raj is see I I do want to be practical. here I don't think India has the capital. to compete square on with some of the. labs in valley at least and equally in. China I would I would say right um and. therefore we do have constraints. so what we kind of took um and along. with um you know minister Rashini Vashno. G and his u ministry the the team as.

Krishna and sir um Sudep shastava and. others who are running the India air. mission all of us collectively took a. call that let's also solve for. efficiency because there are two. objectives India needs to use one is the. frontier and being able to have. something in case geopolitical. situations and create a need for it. >> the second is to say okay fine great but. >> even if we create that all Indians can't. access it because our per capita income. won't allow for it a farmer sitting. there can certainly benefit from knowing.

a planned disease that's predicted. through just mobile camera open and boom. it's giving you you know Bengali. what you need to do right mitigations so. there's clear impact but can they ever. effort it can they ever pay for it. answer is no but there is a mobile phone. in most of their hands today so how do. we leverage this digital inclusion. that's gone right combine it with. somehow democratizing AI and the answer. there was efficiency so we we said let's. figure out if we can keep the frontier.

quality. um and then reduce the cost. significantly so that the untapped. unserved market is also being it's. almost like financial inclusion, digital. inclusion or AI inclusion. >> as the third wave of maybe what we're. doing. Um was just to say that we need. video um video is there's a large. creator audience here. They're not. served today. Um most global models are. very expensive. So we we went in and. created a model that's 27 times cheaper.

to the best open-source model that comes. from China band 2.2 um for comparable. quality um used a lot of tricks in the. table to get it to that point and we've. just put it out as an open weights model. on AIOS so that developers in India can. also start. >> adding stuff on it. there is a live. interface. We're seeing a lot of. traction. Um I'd say there is an API. side of the business B2B angle that's. really going wild. Um this is large.

retailers, large brands, a lot of media. creative agencies that are doing ads and. stuff coming on. Um and then there's a. almost a 10,000 plus last I checked. proumers paying who are just coming. purchasing and creating videos. cuz now. there's some amazing videos. The ones. that satisfy me the most are educational. content for like K12. Um, so we're. seeing some crazy anime videos being.

created for a thirdyear school student. by a teacher who's in some Rayur. >> who's just trying to say video is a. better way to communicate and teach. We're seeing um devotional stuff like on. steroids. I didn't expect that. There's. obvious storytelling and film making. action happening in the space. Uh I. didn't expect the proumers. I think. credit to maybe the India mission. announcement. We ended up having a lot. Um but I can certainly say that APIs. were a expectation because I think what.

we're solving for is if you think of a. large brand in India. and their personalization needs the. number of videos one needs to create for. a product so that they can be. personalized to different cohorts for. example itself is. >> a crazy number. If you think of their. long tale of low SP products the ROI. doesn't make sense to create videos. So. you would if you go to Amazon and if you. go to the cheaper products you won't. find a video there. Not because they.

don't know video sells more. The simple. point is even if I sell three four more. it won't make the ROI for the video. investment that we've done. By reducing. the price of video down we effectively. unlocking that audience. So we did. expect the B2B uh that's not surprising. me and now it's global. So we've also. seen a lot of attention. Techrunch. covered us next web covered us. So. suddenly that triggered a lot of. international inbounds coming saying how. are you 27 times cheaper. I still think. we have a distance to go on quality Raj.

Um so there's a version two that's. cooking with some very very interesting. stuff happening. Lip sings you can start. put your photo and then ask create. whatever video you want uh kind of. things. There's audio native status. coming in. Um we should be in the miniax. zone on quality by the time we release. Varia 2.0. quite excited with what's. happening. A lot of luck as you would uh. ascribe it. >> I'm excited. >> Good work. Good work by the team. >> Here's the last question I have for you.

>> What is one advice that nobody should. follow. >> that you can't do it. Don't let anyone else tell you that. >> But thank you so much. This is fun. >> Pleasant. It didn't feel like a podcast. >> Thank you. >> So that's your skill. Thank you. I'm. glad it didn't. It certainly didn't. It. felt like a conversation. So, >> perfect. Thank you for watching this. episode till the end. We would love to. know what you liked or disliked about. this episode and which guests you would.

like to see on the show. Let us know in. the comments. Your feedback help us. improve and make every episode a little. better. I'll see you next time. Until. then, keep figuring out.