And that's why I like the title of your podcast also. And you'll find this is not Zara, it's a lot. So the name Humanronic actually signifies Human-tronic, which is AI on its own, but also not just AI as software, but AI in the physical.
Hi, I'm Arti Marathi. Welcome to Zahhatk People Podcast from the CXO's Desk. I sat down and had an amazing conversation with Monish Darda, co-founder of Humanronic, an enterprise AI company. Monish has talked about his vision of Humanronic, AI, his life story, and his entrepreneurship. If you are passionate about technology, cloud computing, AI, enterprise software, and would like to become an entrepreneur who would make a global impact, then you are in for an amazing conversation.
Hi Monish.
I'm good. I'm good. How are you?
I am good. Welcome to the brand new series of Zahhatk People, which is from CXO's Desk.
Yeah. And when I say CXO, the first name that popped up is of course you, because I look up to you.
And who else would start this wonderful series other than you, because you inspire me and so many others.
So I wanted to sit down and have a conversation, you know, to know about your life journey from CH...
Which you already know a lot of.
Yeah. But it would be really good if we start with the latest venture.
Yeah. Which is Humanronic. So I would like to know about it. And when we say Humanronic, I think somewhere Humanronic and AI connect. And I think a lot of times you have talked about human-centric AI. So what tasks should always remain human-led? And is AI making us smarter or is it creating a dependency that would weaken our critical thinking? Maybe from a Humanronic angle, or you can start with the latest venture and eventually we can go to AI.
Absolutely, and that's why I like the title of your podcast also, Zahhatk People. And you'll find this is not Zara, it's a lot. I think I have a very different view of how AI and humans work together. So the name Humanronic actually signifies Human-tronic, which is AI on its own, but also not just AI as software, but AI in the physical world, like robotics for example, right? So humans and AI come together on many different terms and in many different ways.
So the way to think about this is, if you today, if you're looking at AI, people are first worried about, hey, is it going to take up my job? Is it going to stall creativity? Is it going to make people redundant? All of these, there's a lot of discussion. I don't have to go in there. What many people maybe do not realize yet is that AI is inevitable. It is like the internet right now. You can't stop the internet. It is like electricity. You can't stop electricity, right? It's like the wheel. You can't say I want a world with no wheels in it, right? So it'll be an integral part of us. But beyond that, AI is now getting so evolved that I don't think there is any doubt in my mind that AI and humans will meld together.
What do I mean by meld? You are a painter. You're supposed to be creative, right? Or you are a poet. You're supposed to be creative. Obviously your creativity is a very human trait. And we try to distinguish today, AI is not creative. It only knows what it learns and it cannot be creative. It can be trained is what we say. It can be trained. But if you think about the human mind, and you know, my daughter being a neuroscientist helps in this thinking, we are also trained from the beginning when we grow, when we are infants, in terms of how to react with the world. Right in the womb, not just outside. So you start our training with that, and we are trained, and AI is trained very similarly, but it's trained on a much wider database, right, where the world's knowledge, books that we would never read in our lifetimes, you know, tweets that we would never see in our lifetimes, and videos we would never watch in our lifetimes, all go in training a model.
So the thesis that we started with at Humanronic, because the approach that if you think about AI, the approach that humans should only be involved in case of escalation and everything AI should do, or humans should use AI as assistance, is probably the wrong model. The model should be that how do humans and AI work together absolutely together every step of the way. So to do that, the first step, you know, my thinking was that if I had to make AI one with my life, what would I need to do? And the obvious answer was, oh, I would need a model that understood me and reflected me.
So we started with what we called hyper-personalized human language models. Hyper-personalized because they're not just personalized, they are hyper-personalized to you, the way you think. Right, so we took six, and because we are obviously focused on the enterprise and the use cases in the enterprise, we said, as a leader or as someone who works in an enterprise, how do you do your work? And we came up with six things.
That everything that you do, you wake up in the morning, you respond to an email, you go on a call, you do something in the day, it falls into one of six buckets. Either you're deciding, you're taking a decision saying, hey, should I hire this person or that person? Should I put five thousand dollars here or ten thousand here? Which tool should I use? I'm taking decisions. I'm delegating. Oh, this I don't want to deal with. Arti, can you please take a look at it? Right? Or I say, hey, my salesperson is the best person to handle this thread. Can I delegate it to him or her? All that is delegating.
The third one I'm doing is analyzing. I get an email, I start thinking, oh, what happens if I do this? And should I do some research on pricing, or should I do some research on this particular technology? Will it fit properly with what I'm doing as my business, or should I, you know, research my partner? So this analysis falls in that third bucket, right, that we analyze constantly. The fourth one, which is very interesting, is creativity.
I looked at my entire iGATE journey which spanned 17 years and I identified about 50 emails where I started the email. You know, think about how many times you do a new email, and you start a new idea and a new thread, and what happens to that thread over time. And I followed these 50 emails over many years, and I found that one small idea can actually make a big change in the future of a company.
Right? So now you think about creativity like that. Right, you come up with an idea, what happens with the idea? Some ideas just disappear. But some ideas actually go through that execution and they actually produce results. And you want to make sure that most of your good ideas actually produce results. But the reality is 99% of them just vanish because execution is far behind creation. So we handle that creativity aspect of the human brain.
And then you respond, right? That's the fifth thing you do. You say, I will meet you at 5:00 PM. No, no, no, not 5:00 PM, I'll meet you next week. Or I send you a hi and say, hey, will you do a podcast with me? That is all responding, right? And then the last one is coaching. I either get coached, somebody tells me, Monish, you were wrong here, or maybe you can think of doing this, or I tell someone, hey, can you figure this out? Maybe you do a training in this area, or maybe, you know, today in today's sales call you should have said this. Those are coaching moments. So if you look at what a leader does in their whole workday, it falls in one of these six buckets.
And we hyper-personalize through a model all these six things so that you can be a better leader and, more importantly, you can scale that skill across the organization.
Okay. So I got the idea of these six buckets and how a leader should be. But if I consider, so what are you actually creating?
We're creating a hyper-personalized language model. And we call it a human language model because it's supposed to work with a human. You cannot separate the human and say the model will work on its own. So it'll work with you in your inbox, in your teams, in the applications that you use daily, and it will, first of all, you know, try to make decisions the way you do. And then when it is successful, you can leave some of the decisions to it. So they are automated. And then finally, the vision, if you think about what is the vision of Humanronic, finally you get to a stage where most of this you can automate. And that goes to our vision, which says we are building the foundation for the autonomous enterprise.
So currently when people are creating these AI agents, right, they are on LLM, large language model. So how this model, or I won't say agent, as you said, automated work, because that's what AI agents do right now, how would this be different than the current AI models?
Yeah. So that's a great question. So if you think about how you take a decision, right? You remember some things that you've done in the past. You use that knowledge, tribal knowledge, you use some training, you use something called intuition. Intuition is nurture plus some nature, but also nurture. So your personality comes in. It also, you know, if you really analyze it, also depends on, you know, how your day started. Were you angry? Were you happy? You know, your decisions sometimes reflect that. So your emotional content. But if you look overall as a pattern, some of these things directly come together in terms of how you make decisions.
When a frontier model like a Claude or an OpenAI is trained, it is trained on the world's knowledge. But while making decisions, it is as important to forget as it is important to remember. So you have to forget some of your biases. You have to forget some of the things that you did wrong. You have to forget some of the things that now don't work in today's world. And frontier models are not very good at that because they're trained on the world knowledge, right? So they will give you an overall average response unless you are very good at fine-tuning, which most of us are not, right? Or prompt engineering, which most of us are not.
Right. So now what human language models like Humanronic do is they go in there and they will reflect your thinking. Not everybody else's thinking in the world, but your thinking. So this whole thing about will humans be redundant? My answer is very simple. They cannot be redundant, but you cannot separate the humans from the AI going forward. They have to be part of you. Right? And that's the theory and the hypothesis behind Humanronic.
Yeah. So that's where my next question goes. So you already answered, but still, if you can add anything more. What does the ideal collaboration between humans and AI look like, maybe let's say in 2030, like four years from now?
Four years from now, that's a great question, looking into the crystal ball if I can look at it. As you know, I'm a big fan of Asimov, right? One of the worlds that he predicted was obviously there will be robots, and they will do everything for you. So for example, I don't like this office, I deploy 100 robots because energy is cheap, compute is cheap, AI is built into the robots. The robots have a swarm intelligence, so they can work together, and in 15 minutes this whole office can be transformed into a completely different thing. Right? Because now I have infinite power, infinite compute, infinite AI, which is where we are moving.
What I'm hoping will happen, also a little bit excited about what will happen, is if you think, and now I'm talking about humanity and philosophy as well, right, as you know I love doing that. If I have to think about the three problems that are facing humanity, one with AI is compute, right? Everybody's now hungry for compute. Your GPUs, Nvidia is trillions of dollars in valuation. Never thought a company like that would go to that height. No company in the world has done that before. The second is power. Everybody's hungry for power. They were putting data centers but don't know where the power is going to come from. Right? So they're taking time to do that.
So you put that power together. And the third, so once you have compute and then you have power, the third one is the limit of human life, right? So our average is about 70 years, but if humans can live to 200, 300 years, it can change the game. So these are the three big problems in my mind. AI is now focused, and the best talent in the world is actually focused on solving these problems.
You look at the recent IPO at SpaceX, Elon, whatever his political allegations or whatever he thinks, he's probably one of the defining personalities in human history. He's thinking that I will get AI, I'll get something like Starlink, and I'll get the ability to send rockets in space at a commodity cost. And I will solve the power problem, and I will solve the communications problem, and I'll obviously solve the AI problem, right? Which is the compute problem.
So that vision will get pushed to reality very quickly. So you look at it, SpaceX becomes the largest IPO in history.
Yes. There's a reason for that. You recently put it on LinkedIn also.
Exactly. Exactly. So you're seeing all of this coming together. Compute, power, AI, all coming together. The other thing that is happening, first time in history, there is a drug that is now approved by the FDA for human testing, still test phase, very early, we don't know if it's going to succeed, but first time in history which claims to reverse aging. Never happened before.
There's one guy, right, who...
He's a billionaire, correct.
Who is into this whole health thing, but now there are also language models, you know, superbly trained, personalized language models that will monitor your health and tell you what you have to do every day, what you're eating, what you're exercising, what's your body mass, you know, all kinds of 500 things, right?
Think about how it will accelerate where AI becomes, you know, really powerful. So now if I look at four years, five years down the line and all these three converge, we are going to see a world that we have never seen before, not even imagined before for most people, right? And what is going to happen is we'll live longer.
There is a very interesting theory called the escape velocity of human aging, right? Or longevity as it is called, longevity escape velocity. So if you manage to stay alive for five years from today, you'll stay alive for ten more years. If you manage to stay alive for ten more years, you'll stay alive for another ten years, right? And so on, because this is going to go up, this whole longevity business.
Because AI is fueling it. And now because people who do not have coding experience, or drug experience, or deep medical research experience, but are very good at looking at patterns and putting things together, can also contribute to the development of these technologies.
So in four or five years, I think the whole world is going to change. We're going to live longer, we're going to get AI embedded in everything that we do, and we are going to get super compute and super power. And that's why, you know, one of the key problems to solve in the world is to convert human intent to outcome. Right? That's why people are spending trillions of dollars. That I have this intent, can I get it to become real? Right? Can I change this office in 15 minutes, that's my intent, how do I make it real?
Is that a driving force behind Humanronic?
Okay, so since we're already on this topic, let's talk about your latest book.
So, okay. Yeah. Would love to know. I of course, I'm waiting for my signed copy, but...
You'll get it right after this.
Actually, we should have put it here.
Exactly. Yeah. So it's interesting. I was going through a fairly bad phase last year because Sami was my co-founder of 17 years and he was fighting cancer, and I had wanted a distraction. I had had this thought about DeepSeek, which is one of the Chinese models that came out a few years ago, and when they announced DeepSeek and they launched it, the stock market saw the single largest fall in value of a company. Nvidia stock went down, and because it was already at a very high price, market cap, that fall was the largest in history, never happened before, because of that one announcement.
Okay. So I'd always been fascinated. I'd been following DeepSeek, you know, obviously, and figuring out, you know, what language models like that are actually doing and how they're progressing. And I had this idea of writing a book about what I called the DeepSeek moment. Where people at that time when DeepSeek announced were completely convinced that OpenAI was going to win this AI war, right? Anthropic was trading quite a bit. Claude, which was Anthropic's kind of flagship, and then there was Mistral, and there were, you know, two or three others, but OpenAI was here, everybody else was here.
Right, and everybody thought that unless you have hundreds of millions, if not billions of dollars, and the technology to do it, you can't actually build a foundation model. What DeepSeek proved was you can build a foundation model, you can make it open source. So one of the things that is secret sauce for all of these models is the model weights. How is the model weighted internally so that it can produce the output that it does? DeepSeek said we will open source the weights.
Okay. What that means is you can take DeepSeek's model and actually tweak it and train it for your use. Which you can't do with a foundation model because the weights are not known. You can fine-tune, but you can't train. So that was a very big step in the world of AI where people started thinking, oh wow, you know, if a small company in China can actually do this, we can do it too. And that changed the whole equation forever. That's why I wrote a book about it.
I think June 14th was when it was released on Amazon. I still need to do a formal launch. I was still thinking about it. So as it happened, my mom also wrote a book.
Kind of somebody ghostwrote it for her, but she kind of put her life story in, and it's called Danger Dadi.
And then, so as it happened, you know, we have this, and Quor is doing the English translation of that book. So Marathi, she's doing the English translation. So we wanted to put all of this together into a launch. So hopefully in July...
You are thinking of doing it.
But it's available on Amazon.
Yeah, definitely, I'll buy it. Okay. So...
You're going to gift it.
So we talked about the latest venture. Now let's go a little bit back. Servosity, StorageTek, SunPlate, BladeLogic, BMC, of course, WebSim before that, and the 40,000 crore rupees profitable unicorn. I've seen so many different domains, so many different companies, right? And so since you have worked on so many different domains, was the domain shift easy, and how was the overall learning curve? Because I joined BladeLogic, but before that I knew about you, so...
No, I think I'm grateful first of all that I got to, you know, experience so many startups, so many domains, worked with great people like you, you know, learned a lot from the people I worked with always, and I grew as a person, and in the right direction. So I'm very grateful for that. I think the domain change, the shift, happens in your mind in the sense that anything that is new is always exciting for me, right?
So if it is exciting and then you put the effort in to learn what is going on, then it becomes, you know, much easier. Then you don't see it as, hey, is this a new domain? You see it as a, hey, there's a new story in my life, and I would love to start, you know, picking up a new book and starting to read. And that makes it really interesting. But it is also hard work.
Right. Like for example, if I pick up a book because I like the cover and it happens to be in Russian, then I'll have to either learn Russian myself or figure out a translation, right? Which by itself today at least are harder, right? So now you have to figure that out. For me it was always, you know, hard work. Because I think if you don't put the hours and you don't put the mental focus required in what you do, it is very hard for you to succeed.
Right, and success of course is defined very differently by different people. It's a very personal thing, right? For me, success always was, hey, am I actually getting the outcome that I started with? What was my intent and where am I, right? And then if I'm not, is it because I'm doing something wrong or is it because I need to change my intent? So, one of the things I learned was change management was the biggest piece here for yourself also. How do you handle this change?
Every company that I was part of, every company that I was part of as a startup, as a group of five people, one or two people, you know, growing to 200, 2,000, 5,000, was always, can I change? Because it needs me to change. And I'm grateful that I could in many places. In some cases, I had to really train myself and force myself to change. In some cases it was natural. And once that happens, domain switching doesn't even enter the picture.
Interesting. Okay. So through your companies and work, I definitely felt or feel that you have taken India and the IT work at a very different level, and of course there are many people like you. So...
Not too many. I'm kidding.
I'm kidding. I'm kidding.
So, and we have talked about Humanronic, we talked about AI, we talked about how the world would be in 2030, and since you worked with Indian people in India, like lived here and worked here, do you see that India can become a global AI leader? And if that has to happen, what must be the reality?
That's a great question. So when I was growing up as a professional, when I started WebSim, at that time in the '90s, body shopping, bad word, body shopping, essentially sending people on H1 visas to the US, was the activity to do, right? People, obviously a lot of our current services companies also built on that foundation and it really worked. People thought it was bad. In the long term, it turned out to be a huge blessing.
It was like, you know, when I was in school and college, people were talking about the brain drain, that, you know, the government invested in IITs, you know, one crore or two crore rupees per student, and everybody went to the US. And people were talking about the brain drain and it is so bad and people should be bonded and not allowed to leave the country. And look what happened after these people who went on, you know, after their IITs, or because of this body shopping, consulting kind of revolution, so many people went to the US, became people who could actually make decisions and sent a lot of work back to India.
The services revolution, where I was, you know, part of that services revolution in the beginning, but the aspiration was always, hey, why doesn't India succeed in product development, software development here, for building products for the world, you know, came later, but even for India. That took a long time.
Right. But we did it. You know, we have companies that are respected all over the world. You go to the Silicon Valley and you'll see Zoho's ads everywhere. You feel proud that an Indian company built in India, made for India, can now transform itself to make for the world. And that is, to my mind, that is one of the greatest achievements as a country, that you can actually transform the thinking to be global thinking. And we have done it in many sectors. Software becomes obviously the flagship, but we have done it in so many vectors.
Now you think about AI. Look at the challenges. AI models need skills, but they also need data. Our data is very fractured as a country. You know, the India Stack has produced a lot of data with UPI and UID and all of that, that is still building, but it is not the data that can be easily turned into model training data.
Our, you know, people talk about the Vedas, people talk about their ancient history, and we have so much richness and so on and so forth, but there is no structured knowledge base across the country about all of these. In different parts of India, we are so varied, so rich, but the basic foundation that is needed to actually put that data together, like a Facebook, like a Twitter, even though a lot of it is nonsensical data, it gives you a fabric. We don't have that fabric.
We're still dependent on the western world, right? We still end up using Facebook and X and LinkedIn and so on and so forth to build the structure that is needed for large language models to be built.
Companies like Sarvam are trying, the government is trying, the government is trying to put together this, but it's a long journey. So the short answer is, in the short term it'll be hard for us to become a leader. In the long term, we have the capacity, we have the largest English-speaking population in the world. We are a very thriving and a very live economy. We don't have the constraints of other countries like China, which is viewed to be a little bit more autocratic and constrained. So we have the freedom to express ourselves and to build that structure, that human structure that is needed to get that data in a way that can actually build and power some of our models.
And the same thing with Humanronic, you know, we want to build models for India. We want to build models that will actually show, you know, how a human being in India can actually interact and take these decisions and delegate and analyze and create and respond and coach. And we want to build it for the Indian ethos, right? So something to think about. I think there is a lot to be done here, but we have the potential.
So you want to make sure that the potential comes out, and again, it's intent to outcome transformation.
Yeah, yeah. So as you said, data is...
Fractured. So I guess that is a sector where people should work if we want to become global leader in AI.
That is one of the foundational things that is needed. For sure. Correct. Correct. Yes.
Okay. So since we were on body shopping, I had that question from a very long time that you founded Websim in 1998 when people were literally calling it body shopping. So seven, eight people every single day were going there, right? So when so many people were going there and the main intention maybe for companies was body shopping, but people were going there to earn money, right? Dollar was like always, still it is there. So what was your thought process when you came back here? Was there any prediction you thought of that time that you know, you never know, that India will become the biggest population working in IT industry or something like that? And the second question is, while doing this you actually are a mechanical engineer but then you shifted to IT or software industry, so how all these things have happened?
Yeah, yeah, so very interesting. So first of all I was not a visionary. I didn't come back because I thought India would be a much bigger opportunity for me or my career. I think I came back because I felt more comfortable in India. Okay, that was the only reason. I think I just felt more comfortable in India. I wanted to be with my family. We are very close. You know, we are extremely close actually. If I look at most of the people around me, my family probably feels like it is, you know, one of the closer families that I've known, right? So we are very close and I always wanted to come back. As you probably remember I was already married when I went to the US. So my wife was also a very deciding factor on hey Monish let's go back, right. So it was an easy decision for me. It was not a visionary decision.
So mechanical engineering essentially at that time, even today I love working with my hands and putting things together. You know I'm a gadget freak as well. Yes. So I love that aspect of it. But even in college I was doing a lot of programming. Okay. Right. So I remember I went to Maharashtra Institute of Technology, MIT in Pune, and this was the first batch, that was the first batch where you had computer science as a degree, right? So the first CS degree that started was actually by MIT in Pune, and I used to help, you know, all my friends in CS to actually do their assignments. So yeah, because I loved it. So, and I was good at it. So you know I always thought mechanical engineering and software were joined to the hip, that you could not be a better mechanical engineer till you knew this new thing that was called computer science, and I was fascinated by it.
And so it was not surprising that when I came back I was passionate and still am about cars and motorcycles. So when I came back after my masters I spent three years in the Automotive Research Association of India. I walked into it, you know, there's a guy there, a director level guy, very senior, very nice guy who was a PhD from, ah I forget, but one of the Ivy League schools. And he said what are you doing? Why don't you join me? I said wow you know this is great. I can work with cars and motorcycles and engines and I can do a lot of software. So I did a lot of software in ARI. We actually developed one of the first electronic data acquisition systems at a million samples per second inside the engine cylinder, real time. So I spent three years and then I started Websim.
