Rishi Ahluwalia (00:07)
Welcome to the Convo AI World Podcast. In this podcast, we interview builders, developers, and teams that are building solutions for the conversational and voice AI space. Joining us today is someone from the astrology background who has shaped the digital age in astrology. And at some point, I think we've all asked an astrologer,

'When will things get better?'. Now imagine that astrologer is an AI who is available 24/7 speaking your language and understanding your emotions. So today's guest is the visionary that's making it real. Punit Pandey joins us for Agora's Convo AI World Podcast. He's the Co-founder and CIO at Astrosage and has transformed how millions

experience astrology, blending deep tech with ancient wisdom for over 100 million users worldwide. Welcome to the podcast, Punit.

Punit Pandey (01:04)
Thanks Rishi, thanks for having me and thanks Agora. So very nice joining here.

Rishi Ahluwalia (01:10)
Absolutely, Punit. It's a pleasure to have you here as well. I think, Punit, with the recent innovations like AI Astrologer on call, you are redefining what happens when empathy meets algorithms. So again, welcome to the show, and it's a pleasure to have you.

Punit Pandey (01:27)
Thanks Rishi. My pleasure.

Rishi Ahluwalia (01:29)
Perfect. All right, so Punit, let's dive into the contents of the podcast today. So in the first section, let's talk about your vision and the turning point, right? Obviously, you always straddled two worlds. One is astrology and one is technology. So can you give us that 'aha' moment when you realized that conversational AI

would transform how people experience astrology.

Punit Pandey (01:56)
Actually, there were two, three major moments, I would say. So first was related to AI itself. It was somewhere around 2015 or 2016, once I got into this neural network and deep neural network. So that time, I did programming for more than 30 years or so. So I started there

early and when I was 10 or 11 years I started programming.

So when I encountered artificial neural network for deep net, so I was kind of shocked that computer or a software or artificial brain can learn things. So I was always learnt and taught writing code step by step. So when I realized that machine can learn, that was a big 'aha' moment for me that will probably in this,

after coding for long time when you realize we were doing it very differently or maybe we were doing all wrong so that was kind of a shocking and surprising 'aha' for me. So that was I would say the first part and then the second part was, I saw this Google Home and Echo, Alexa Echo and all these voice-based device coming and then they are understanding our

voice so well, again, neural networks playing a role in that. So again, that was another 'aha' moment for me. And then that was the time actually, we decided that we will be building our own conversation. We released our first AI Astrology app in 2018. So name of the app was Bhrigoo.ai. So that was a conversational app.

I would say that the AI was not at the prime at that time. So that product was not that successful. But I would say these were the two 'aha' moments which brought us here.

Rishi Ahluwalia (03:54)
I'm sure I think a lot of aspiring developers and the kids today will definitely relate to that, Punit, right? Because it's always fascinating that you can create an artificial brain and train it over time. And in today's world where AI is moving at the speed of knots, I think all the aspiring students and kids and developers out there would definitely relate to what you just said. Thank you so much.

Punit Pandey (04:20)
Right. In fact, I would say that for us who are probably kind of a bit old in this in the AI game. So for us, unlearning was bigger challenge than the learning. So we were so much into the traditional way of coding and then just forgetting everything and starting from scratch. That was probably the most difficult thing for us.

Rishi Ahluwalia (04:45)
Absolutely right. I think that's something that will change for the future generations. AI keeps surprising us every few weeks. So it's definitely a lot to do these days, but especially not starting from scratch. Yeah, perfect.

Punit Pandey (04:50)
Thank you.

That's it.

Rishi Ahluwalia (05:02)
And I think, Punit, the next question I had in mind was that a lot of startups today continue chasing automation, right? But not conversation. So what made you deeply bet on conversation as the interface for something as emotional as astrology? Because the way we see astrologers in India or all over the world,

is very personalized and we are sharing our data which is again very confidential. So how have you bridged that gap of empathy, emotions and obviously changed the game with AI Astrologer?

Punit Pandey (05:40)
As far as picking up the conversational part of AI, so as I told this, this  came natural to us. So we started with Bhrigoo.ai which was voice-first platform. So we started from there and then now when generative AI became popular, that was also because it was backed by large language models, primarily language models so conversation

was, I would say that part of that whole process. So the nature of astrology in itself is conversational. So I think the choice for us was was easier. That is the reason when I saw the voice-based  platform earlier in 2016-17. So that made me clear that this is the future of astrology.

Rishi Ahluwalia (06:27)
No, I think that's again a great point because conversation is what drives astrology. And you build trust over time. And the remedies that come with astrological solutions, once you see that coming to life and performing at its best, that's where the conversation becomes into faith, right? So hopefully that's where AI astrologer unlocks the next level of growth as well.

Punit Pandey (06:52)
I hope so.

Rishi Ahluwalia (06:53)
Perfect, perfect. ⁓ Thank you, Punit, for those insights. Now I think let's deep dive into what is inside the AI astrologer tech itself. So say for those who haven't tried this yet, what actually happens when someone calls the AI astrologer? Could you walk us through the tech and the experience?

Punit Pandey (07:16)
So as far as experience is concerned, this is pretty simple and probably better optimized due to the tech part. you provide in case you are consulting an astrologer. So before that, I would say our platform is more like AI avatars, right? So this is no single AI. So these are various different AI. So for example, there is a Mr. Krishnamurthi,

who is an expert, he is an AI astrologer but he is an expert in a system called KP system of astrology. Then there is Anita Jha and Tarot Vijay, they are expert in tarot reading. Then there is Vijay Vemani who is tarot expert in numerology and so on. So basically the point is that we do not have one single model, one single system. We do have a lots of

amalgamation of the AI models. So when you move to our platform, so generally when user comes first time, if we just decide on the two parameters, whether the birth details are available or birth details are not available. If birth detail is available, then you are automatically connected to one astrologer that we decide if birth details are not available, then we automatically connect you

to tarot reader, again, we then this is the first onboarding funnel and once that onboarding is done and you had a chat with or call with that particular astrologer then you can choose any other AI astrologer. So, there are bunch of different AI astrologers, but because we did not want to confuse people about all the technicalities. So, the initial onboarding flow is very simple and straight forward.

Now, I guess the second part of the question was that the technology behind it. So again, the technology is quite complex and very AI heavy as such. We do have different models for different astrologers or the avatars is probably the best word. So different models for different avatars. We do have the backend

selection-classifier model as well to with back end one has to go and then we also have a classifier or the kind of question user has asked so these are the technical part of technical starting part and then once we decide and decipher and understand everything about this then we have our own astrology engine which is based on this Bhrigoo.ai

which

we released earlier. Those engines understand it and then try to prepare a basic astrological requirement for that answer. And then the conversational AI of that, if it is a chat, then the model response will be the answer. Or if it is not chat, it is a call, then there will be some kind of a model that will be the voice or a real-time model, which will be answering this.

So again on a front, the last point, the real-time model, we generally use the OpenAI model and the Google models for real-time answers. And then of course, all that conversational call related things like Voice Activity Detection and the latency and other things, those part we would like to

optimize as much as possible because people want to feel like they are having conversation with the real astrologer. So it should not sound technical, that is the high level picture of architecture, I would say. Does that answer your question?

Rishi Ahluwalia (11:09)
God.

Absolutely, Punit. I think you've covered it across multiple spectrums here. The AI astrologer scope is defined, whether it's across the voice channel or the chat channel. And I think from KP part of astrology to Tarot part of astrology, everything is defined. And there are prerequisites if the birth details are available for the user or not available for the user. The AI adapts accordingly and gives

personalized and customized responses respect to what is being captured in the system. So that's pretty solid and you know what, spectrum-defined system, I would say. So that's a great response again. Thank you so much for that. And I think since you spoke about the scope, the definition, the various ways of how AI is responding to user queries.

So that brings me to the next question, is that astrological queries are often emotional. Sometimes it's vague as well, and sometimes very, very deeply personal. So when it comes to training the AI model, I think you've already answered this in parts in the previous responses. But how do you anticipate to train an AI model which will have such varied

intent and what it needs to respond with empathy as well. So how do you go about training such AI models so that the users adopt the AI responses whether it's through chat or voice in a personalized manner and consume it and in the form of faith and build faith over time.

Punit Pandey (12:45)
Right, first of all, as I mentioned that our AI is not one single AI. So we do have various different models. And generally, even in real life and because we are an astrology platform and we have a large number of data, probably the largest one. So what we have seen is that different people do have a different kind of expectation from the astrologer.

So some people like the astrologers who are very straightforward. They will say good thing is a good thing and bad thing is a bad thing. And other people like that it should be polished, sugarcoated and toned down version of what they say, so different people do have a different expectation from astrologers. So our job is to design and define

astrologers of various kind and let people choose whatever works best for them. So for example, we have an astrologer called Mom the astrologer. So as the name suggests Mom the astrologer is extremely empathetic astrologer. So generally she keeps and make sure that somebody coming from the sad or depressed background.

So that should be handled with the help of astrology of course, and then there are some astrologers like Dr. Raman. These are very straightforward astrologers. They will say whatever is correct according to astrology. So generally our idea is to make it more like a human astrologer platform. You have availability of all different ways unless

until these astrologers do not cross certain boundaries, certain guardrails, then I think we want to give as much human experience as possible with the benefit of AI. So of course, there are lots of benefits of the AI in Indian astrology, but yeah, that we want to give, but idea is to have this similar to human experience.

Rishi Ahluwalia (14:51)
I think that's another fascinating take, right? Because there are people who prefer very direct responses, no nonsense responses. And some people do not really like that. And then empathy is how you handle that. I think Mom the Astrologer might become a hit because that's like your mother giving you advice on astrological predictions. and what you should be doing. So people might find that more fascinating, but I think it all depends on maybe Gen Z versus

Millennials versus the senior citizens who are using the platform.

Punit Pandey (15:21)
Absolutely, as a tech company we want to make sure and we must make sure that people do have choices and with certain guardrails it should be accessible whatever they would like they should be able to choose. again I say that for Astrosage AI is not a single model, single platform kind of a thing it is more like astrology for everyone.

Rishi Ahluwalia (15:45)
Exactly. No, absolutely. I think that's where ⁓ all the founders need to aim as well. It's addressing everyone across the globe. Perfect, perfect.

I think that brings me to my next question here, Punit. Now, I think during the development of the AI astrologer, there must have been moment where AI might have surprised you, since you're working across so many facets of AI. So the surprise might have come either by working better than expected or getting something hilariously wrong. So do you have any examples of that?

Punit Pandey (16:22)
⁓

tons of examples and sometimes AI surprises you, sometimes AI scares you. So, but tons of examples. And these are very interesting examples. So I'll share a couple of them. So first, when we were building Astrosage AI, so that time, our company CEO, so generally I take care of

I write my designation as CIO, Chief Innovation Officer. So to CEO, I gave product to him for testing and he is little bit interested in fitness part. So he asked question, he asked for my birth chart how to reduce weight or something like that. The AI The AI astrologer analyzed his chart and then said that

you should you should practice planetary combination. You should practice martial art. So Pratik, the CEO, he then followed up that question with which martial art should I practice according to my birth chart. So the AI astrologer again went forward and analyzed his birth chart and predicted that for him the best martial

is Krav Maga, which is Israeli martial art. And actually, I was not aware of that point of time that Krav Maga is a martial art. I started astrology pretty early. So I am doing astrology for 30 years or so. But that kind of shocked me that as an astrologer, we do not understand everything in the world that exists. So for us, it is really difficult to tailor

and customize the answer based on what user is looking for. So to answer this particular question how to use astrology in weight loss you need to understand all the various possibilities of the weight loss and then only you can choose astrologically which one works best for the user. So that was kind of a surprising, shocking, interesting for me and good for us that my CEO also, I got the backing for him that okay this is the good

product you have built. So that was a good part. Another time we were building Instagram integration. So Instagram integration, I use Android phone. My son, my son is a Table-Tennis player. He's 14 years old. So he's a national player. So, but he uses iPhone. So I wanted to see the difference this Instagram integration was making with iPhone versus Android.

Rishi Ahluwalia (18:39)
Indeed.

Punit Pandey (19:01)
I asked Pravar, can you just check this, is it working properly? So Pravar, Pravar being 14 years old Table-Tennis player. So he asked the astrologer, as you know, in Table-Tennis, generally you have a blade, you buy blade separately, you buy forehand and backhand rubber separately, you stick them together. Then based on your playing preference, you create your 

So he asked AI astrologer which forehand rubber should I use. Then the AI astrologer said that your Mars is in 9th house or something in exalted in 9th house. So you use medium to hard rubber on your forehand. So that in Table Tennis, the rubber are classified as soft rubber, hard rubber, medium hard rubber and so on. So then he followed up with

Rishi Ahluwalia (19:51)
it.

Punit Pandey (20:03)
which one should I use on forehand. So again, AI AI Astrologer analyzed and then based on combination plan, it said that because Dignics O9C is a tacky rubber, so you should use tacky rubber on your forehand and non-tacky rubber on your backhand. So this is this is kind

of again shocked and probably brings down astrologer, my astrologer's ego a little bit that okay  I am entering into a tough territory. So AI astrologers are going to give a pretty interesting perspective to the whole thing. So they bring a lot of new and different ideas to the

field of astrology everywhere but I think because we know more it from the astrology perspective so fresh idea. So when we launched AI astrologer in this in Astrosage AI so that time our AI astrologers were far behind human astrologers. Now, at this point of time, AI astrologers are rated higher than the human astrologers

on our platform. We do have 25,000 to 30,000 human astrologers, 20-30 AI astrologers. So average rating what AI astrologers receive on any day is higher than average rating human astrologers receive. on any day. So these are all surprising to us. So when we launched it, we had all doubts and all questions how people will be perceiving AI and is it a good decision to launch AI

in on our platform? Especially, in fact, even our competitors ask this question to us they call and then they chat and then they eagerly want to know why you did the exactly opposite thing of changing your brand name from Astrosage to Astrosage AI. Earlier our brand name was Astrosage and now it is Astrosage AI. So they find it very difficult to understand that they say that we try to hide AI even if we

use AI, we try to hide AI and you are doing it opposite. So, but fortunately for us, it worked really well and I would say, astrology never stops surprising us.

Rishi Ahluwalia (22:24)
No, absolutely. think this perspective is going to be very well received by the viewers as well, Punit. The kind of questions that a 14-year-old, your son is asking versus your counterpart in the company. And correlating that with martial arts, which is practiced in Israel for weight loss, is quite fascinating. And I think this is where AI obviously will work better because it has access to abundance of information

not restricted to a region or country or to a specific subject. It kind of looks at everything very holistically and since it has access to global information, it can give very curated responses which is something outside the scope of a

single or multiple human astrologers out there. So I think unlocking that level of growth and with the experimentation that you're doing Punit is gonna be aspiring for the current and future set of innovators that the country and the world would see. So definitely fascinating and I think it gives me as a

individual user, it gives me more hope that I can ask anything to any AI astrologer and I will get some sort of response which is in line with what my planets are speaking as well. So great. Yeah. Perfect. Perfect. I think that that section went really, really well, Punit. I think these fascinating stories are what the viewers were anticipating and I'm glad they're getting the same.

Punit Pandey (23:42)
Yeah.

Rishi Ahluwalia (23:55)
So I think moving on to next section, and this is mostly like stories from the trenches. Right now, you've been building and associated with astrology for over 30 years now. Right? So I think if you would want to take us behind the scenes, a story, another story or another moment, maybe even a bug that captures what it was really like building an AI that talks like an astrologer.

So I think we have discussed how people are using it and what kind of responses they are getting. But any stories that might define how the country enacts or any stories from the technology side. ⁓ But taking us behind the scenes would be more fascinating here, Punit. So over to you.

Punit Pandey (24:44)
Right, lots of stories again because as I mentioned that especially AI never stops in surprising us. One recent story, I think this is the last week or the last-to-last week only when we launched AI on call. So now you can have a telephonic call, telephonic conversation with AI astrologers. So there, and interesting

one thing we have noticed that the generally, the five star ratings we receive that is higher from the non-Hindi, non-English languages then in Hindi and English languages. One thing which surprised me that there was an event when we were doing audit of the this also one person who was Bengali and

he was speaking Hindi in a Bengali style. So in India we have a different way of speaking Hindi. Maharashtrian people will be speaking Hindi in a different tone and Bengali people will be speaking in a different tone. So that Bengali person was speaking with AI in Hindi in Bengali tone or dialect whatever is the appropriate word.

And then AI was also responding in the same way. That kind of conversation was so amazing like two brothers talking to each other kind of thing. So it was feeling like that AI is so close to the person which was otherwise for again for any human astrologer, it is very difficult. So I can speak one language, two languages or maybe three languages, five languages, but talking

every language in the style of the way user is looking for, that is absolutely amazing. So this was probably last week's story and kind of a shocked and surprised once again.

Rishi Ahluwalia (26:39)
No, I think that that's another part of personalization, right? Because I think this challenge has existed for most of the part. I think even in astrology, there is a North Indian style of Kundli versus the South Indian style of Kundli. And people generally prefer North Indian astrologers because of the

language adaptability instead of South. I think, which brings me to another question, Punit, and I think I'm probably jumping the gun here. Now, I think since you brought up the language challenge, right? In addressing a proper human–human conversation before, but with AI, can probably solve that now, not just at the language layer.

But like you said at the accent at the dialect layer as well and not just Indian languages, right? I think I also saw that the AI astrologer speaks global languages like Japanese, German, Chinese and French and I think you mentioned that you are getting maximum five-star ratings from non-Hindi or non-native speakers. So I think since you are building from Bharat, but not just for Bharat now building for

global audiences as well. So how did you overcome those local hurdles first while also scaling Astrosage conversational AI for the truly global audience? Because this challenge is something that I think everyone will be able to relate who has tuned into the podcast.

Punit Pandey (28:09)
Right, so one thing probably good for us, first of all, we all in our company top leadership is all from small town.

So we were part of that thing. When we launched, in fact, our first mobile app, we launched was in year 2000 or 2001. So the name of the app was Mobile Kundli. So Mobile Kundli, that time before Android and iOS, the operating system used to be the Palm OS. So the Palm PDAs or Palm OS, Palmtops,

was at that time. Palm itself did not support the regional fonts. So, what we had to do is to build regional fonts for our own. So, that the app in 2001, 2002 mobile-only. So, that was multilingual app. Then at that time the web also did not support Unicode. So, that time there was a technology called

dynamic fonts so you found the fonts were downloaded in background kind of a thing but that also came later. Unicode was not properly supported anywhere so in fact, in Hindi, we wrote the second Hindi blog so when blogging came so first blog was another one so and so name the domain also I had the Hindi blog, Hindiblog.com and because

this was the second blog, so Hindi blog is the default domain you will have. The point I want to make is that because of the background and the need of building called Bharat is very different from the rest of the world. We were forced to create something which is very Indian. So for example, Android supported the Hindi fonts or the Unicode fonts in 4.0.

Android itself was not supporting and we were very disappointed by Android that why they are not supporting Hindi. Even the Arabic and other languages were supported but not the Indian ones. Samsung, they did something in 2.3.5, just gave a little bit of support. But Android

itself came very late in the language game. So we had to do all the hardware to create app which was multilingual in Indian. Now, we currently operating 11 or 12 different regional languages, Indian languages. Our app is in 9 or 10 languages. And then we are coming up with all the international languages that you mentioned, the German, French, Japanese and so on. But I think because of the basic idea

of building and seeing everything from that perspective that helped us building that kind of a product. And that is the reason I think Astrosage, AI is among the very few products which are organically grown in India. In India, generally, most of the apps that we see are grown using advertisement or cash backs and all those kind of a thing. But Astrosage is among very few apps which are grown organically.

And I guess the reason this, the localization and the support for  Bharat from Bharat is probably the leading cause for that.

Rishi Ahluwalia (31:35)
That's interesting as well, right? So you take us back another two decades where obviously the Hindi dialects or the Indian languages were not really available and not supported across multiple platforms. So I think the experience of building something from scratch across the diversity of languages in India made it much easier for you to scale in other regions of the world and adopt new languages. And I think the best part I liked about this

response, Punit, was the organic growth. So hold on to that thought. The next section is based on that how you can share that how you are scaling from Bharat but not just for Bharat anymore but to the rest of the world as well. So as Astrosage charters globally,

Punit, because you have now 100 million users on the platform. So how do you envision preserving its Vedic soul, the soul of astrology, while adapting to audiences shaped by Western or other astrological traditions? So I think everyone on the platform is aware and mostly relates astrology to Vedic sciences defined

in rural India and now urban India as well, but mostly related to then called Hindustan, now called India and Bharat. But there are various other astrological methods and traditions across APAC, across maybe Europe as well, across the US as well, where many of your audiences there. So how do you envision preserving the Vedic soul when adapting to various audiences

who have multiple religions and follow very different traditions.

Punit Pandey (33:23)
All right, so currently our idea is more to take the Vedic approach to the world.

So right now, this adaptation part is not high on our priority list and we would like to see these AI avatars which brings the Vedic wisdom to the Western world how they are perceiving it. And then probably there may be a time where we will be modifying and blending it but not currently. So we believe that the wisdom provided by the Vedas is universal and

So majority of the part should be applicable to everybody. So I think that is one part, having said that, as you mentioned correctly, the astrological methods and systems are very different or at least different from every country has a different system or method. But fortunately for us, we already built lots of systems earlier. For example, the Western world

and the Indian astrology is a Sidereal astrology and their astrology as a Tropical astrology. So, this Tropical astrology, we built it is their part of Astrosage from the long time. So, we already have a AI of that, astrologer Anna. Anna is mainly a Western astrologer. So, there is one or more astrologers

I think the Sidereal astrology, Western astrology, the Arabic part is having a Tajik astrology. Tajik astrology is also part of our system for a long time. In fact, the annual horoscope that we made is based on the Tajik system of astrology. It is called Varshaphal.

Majority of the world is already covered by the algorithms what we have. So, I think as far as the algorithms are concerned, nothing new to build only thing is we will be seeing that current approach is how closely it is working. probably rather than a change we would be needing some kind of tweaking for the rest of the world. This is my assumption, but we will.

have to wait and see.

Rishi Ahluwalia (35:46)
No, absolutely. I think you already gave us a very quotable quote here that Vedas is for the universe and it is universal. So and definitely, thanks for touching upon the various astrology methods followed in different religions as well. I was not aware about the Arabic astrology that you mentioned. So that's another fascinating way of looking into the different aspects.

And since we are touching on astrology a lot here, Punit. So astrology is also considered a way of life in many different definitions that we have out there. So if you think beyond astrology, do you see that conversational AI will become a more holistic life guidance companion?

Like something that can blend astrology, wellness, and emotional support in real-time. And wellness can be in both ways, right? Physical and mental. Now the kind of lives that we are living at the moment, it's become very

focused on sitting a lot during our working hours. So mental health is something that is of grave importance these days. So how can astrology blend into all of that and offer a more holistic approach to life?

Punit Pandey (37:04)
Yeah, so this holistic world itself, I would say is more of a Western concept. So in India, it's not like that. So in India, you have a single person working as an astrologer, as a panditji, as a pujari, as a guruji, as a coach, as a guide and everything. So this is the one and the single same person working as a guiding

force in everything. It is a wellness or it is a psychology, astrology, puja, everything. So this is very Indian concept. This is anyway that gels well with the AI. So AI is very good in doing multiple things and act as various different personalities of avatars and I think this gels

perfectly well with Indian concept of astrologer. So I think there is no difficulty in that.

Rishi Ahluwalia (38:06)
Absolutely. I think multitasking is something that is inbuilt into our DNAs as well. And yeah, I think when we approach someone who's from the same background, right, they truly become our life guide as well. Giving us solutions, performing the rituals, and offering advice, which defines our life and shapes our lives forward as well. So that's another great take on that, Punit. Thank you so much.

The next thing I wanted to touch upon is one thing is building technology, right? And one thing is building trust within these spiritual products that is being offered. So how do you build a monetization model that respects the user's vulnerability while keeping the experience authentic and accessible? Because this is the time

where the user is completely transparent in front of the system, right? That this is what is going on in my personal life, professional life, and how do we tackle that in a very human manner, although the AI astrologer is giving advice. And the most difficult part is building a monetization model out of that. So I think that if you can give your take on that, Punit, that would be really good.

Punit Pandey (39:22)
Yeah, so actually people, what I have observed in all my experience, pay using two different methods. One, they are manipulated, they'll pay. Okay, and then when they trust, they'll pay so in astrology these are the two methods so for example, if I am an astrologer and you came to me and I I just scare you with all

dangerous-sounding  names like Vish Dosh and Kalsar Dosh, Sade Sati and so on. So this is easy to make money. So making money becomes much more easier. So I can make you spend say a lakh or two lakhs, three lakhs on gemstones and poojas. So this is one method. Other method is also there when you trust somebody, then you are willing to pay. So For Astrosage AI, I think we chose that trust

that was slightly hard, slight hard work or maybe a lot of hard work. But the benefit of this is that once people trust, they themselves are willing to pay and offer. And I'll give you one or two examples. we went with the Astrosage when we started Astrosage.com in 2003 or 2004, that was completely free platform.

You can do everything on cloud. You make horoscope matching for marriage and everything. Everything was free. So I think people use it, people trusted it and initial years, we never intended to monetize it. And then people themselves said I want to buy gemstone. Why do you not sell gemstone? You need to sell gemstone. Then they said I have to get Kundli, why don't you sell

Kundli. So these monetization came as a result of trust, I would say, then actual intention of monetizing the platform. I, as individual always wanted to be a tech kind of a software engineer. So my thought was that this platform I will be building for as a hobby project. And then I was involved with the JSR Java expert group and I was building Java

and other things. So I wanted to focus more on that part. But people themselves made us monetize and then they started paying us. And I reached to a point where Astrosage was making more money than my US salary. I was in US and I had a green card and I was working pretty well there. But Astrosage started making similar or more

money than my US salary. Then at one point of time I thought that probably this is what people wanted so I think these are the two models, a quick model and a long-term model but generally both work well for monetization

Rishi Ahluwalia (42:04)
Thanks.

No, perfect. I think the summary here would be Punit that the intent matters here. I think what started as a hobby project and what was mostly tech-driven and I think something that brought you back to India from the US. So faith, intent, all of these things matter more when building products and then monetization falls in place if you've already built that faith and your intentions are good.

So I think that's something for the next breed of founders who might be listening to this podcast. I think they will definitely take inspiration from this. Thank you so much for sharing this, Punit. But I think I'll ask you to hold on to that thought, because the next section that I wanted to discuss was the ethics and user trust. So like you mentioned, it's very easy sometimes for astrologers to make money.

You know, pound the user with many things like Sade Sati, Kalp Dosh, Sarv Dosh, Manglik Dosh, all of those things. So it needs to be a balance and the conversation should not start as fear in the mind of the user. Right? So it should come as empathetic. It should come with shaping major life decisions. So when it comes to astrology, most users tend to fall on astrology

to shape major life decisions like something related to health, right? Now, something related to marriage, finance. So how do you make sure that the AI that you are using and building and scaling does not over promise or does not mislead?

Punit Pandey (43:55)
So first and the foremost thing is the model you train.

Right. So this is the AI at the end of the day, these are just models. So they'll act as you train them. So we want to make sure we follow the approach we are following for since this 20-25 years. That you build trust and then rest of the things will follow. So do not build for the monetization, build for the trust. So this is our idea of model,

I think, is one of the fundamental part. Of course, you need to have a different guardrails and filtering in place and auditing in place but I think those are all secondary if the fundamentally thought of what kind of astrologer you want to create is itself is a crazy thing and I am sure more manipulative and cheat AI astrologer will also follow

because as I mentioned there are only two major ways of making money, one using manipulation and one using trust. So I can expect more such astrologers, AI astrologers to follow but I think our job is to do what we do best.

Rishi Ahluwalia (45:08)
No, absolutely. That's correct, Punit. And I think everyone who's tuned into the podcast will definitely relate to the approach that you're talking about. There's another interesting thing that with the adoption of AI, there's something called ethical and responsible AI that mostly the security audits

are part of. So while you have the experience of building over two decades and three decades, but AI being AI, can hallucinate. It can sometimes mislead even on trained models. So have you deployed any additional internal guardrails or audits that you are building

for supporting the concept of responsible AI, especially in the sensitive space of astrology.

Punit Pandey (46:00)
Right, so I touched upon some of the things in the last answer.

There are quite a few guardrails which are there. So most important thing is this quality and probably emotional audit, probably maybe the right word. So that is there in place and we do have a pretty strong content filters as well for keeping these things in mind. And as I mentioned, but at the end of the day,

  then then they are intelligent enough to pass all audits and content that AI is quite intelligent in that sense. So yeah, I think you need to build that kind of a soul. That is the most important part. And then of course, these are important

aspects the continuous and audit and monitoring plus the very strong content filters.

Rishi Ahluwalia (47:04)
Correct. I think content moderation is another area, especially in this sensitive space that is definitely driven by AI and countered by AI at the same time. Great, Punit. So Punit, as we approach the last bit of the other sections, right, now one thing everyone would want to understand is the future roadmap. Now, everyone's very excited about AI astrologer. I think the adoption is increasing

Punit Pandey (47:16)
Good

Rishi Ahluwalia (47:32)
day by day. But what's next for Astrosage AI? So will you be bringing more voices, more avatars, or maybe using the power of the large language model these days to come up with visual graphs or visual charts while the conversation is happening in real time so that

the audience or the users who are on your platform can relate better to what the AI is speaking about.

Punit Pandey (48:02)
I think, see, in any

technological innovator or, I would say, linear. Objectives have not changed in last 200 years also. From the days of Charles Babbage, so everybody wanted to build AGI and ASI. So if you want to reach to a certain level where the technology is at par with the

general average human intelligence and then you want to surpass it and build something which is really phenomenal and which is impossible for any human to achieve and to do. So that is probably the vision of AI itself and that is also vision of us and probably any technology person per se. So I think we are building in that direction. So we would like to see, in say,

one to two years timeline that for anybody it is very difficult to distinguish who is the human and who is the AI. They should be consulted just like intelligence not the human intelligence or artificial intelligence. The point should be of intelligence. We are launching say one big product every two three months.

So this is the AI call on committee where you can talk to AI astrologer over phone call. This we released recently and there are lot many follow up but the overall, the guiding principle is the AGI and ASI but for us, this is related more to astro-limited more to astrology than the general so may not be the correct definition but the closest definition, I would say.

Rishi Ahluwalia (49:54)
No, absolutely. I think that's where the industry also sees it, right? That AI is mostly an AI assistant rather than, you know, changing the dynamics of the game. Like you mentioned that the fundamentals have remained the same for the past century or so. And it's all about exploring more facets that can unlock more opportunities. But maybe AI astrologer is more of a life companion assistant

rather than changing the entire rules of the game. All right.

So Punit, now with all the innovation that's happening all around with new AI models being available and companies like OpenAI, Google launching their LLM models and bringing new parameters, new dialogue boxes and everything. So how do you see everything changing or how do you reimagine astrology five years down the line?

Obviously complemented through AI. How do you envision the experience changing or completely redefining the rules of the game?

Punit Pandey (51:00)
So one thing the way we believe that astrology is a very useful science. Generally, there are various ways to look at that and generally I see that, for example, in India astrology is not utilized as much as it should be.

And to give you an example that there are certain point or period in our life, for example, marriage. So when marriage is happening, we generally do horoscope matching or marriage matching done. We also choose a good mahurat (occasion) for marriage. And generally, there is a period of time where in India no marriage happens. There is a point of time where marriages are happening.

What I have noticed or what world has noticed, I would say, that the success percentage of marriages are pretty high for India. So even if we see our neighboring countries, so our success rate is pretty high. The example is just a point that I want to make is that astrology if used more frequently, it will be more useful.

For example, people are, let us say, investing. There are people who are doing trading in share market. So I think astrology can help there as well. So astrology can help what the way it happened with our CEO, weight loss. Right? So these day-to-day things, I believe in astrology will be part of day-to-day things. Then right now, because of the unavailability

of the good astrologers you cannot do day-to-day discussion so you have a minor question so you generally go to astrologer only for marriage or career or something very important but if the companion or the astrologer will be available to you 24/7 so you will be asking all sort of minor questions. So if you are not getting sleep at 2 o’clock in night so you can ask astrologer what should be the astrological way of

getting better sleep, right? So, which is currently is not there. This is how I see the astrology getting more personal, getting involved into day-to-day and minor things. This is how I believe astrology is going to be.

Rishi Ahluwalia (53:22)
Correct, I think the availability, the affordability, and the accessibility in the near future is going to be defining the next wave of assistance offered through astrological ways. So that's another great point, Punit. Thank you so much.

Punit Pandey (53:39)
Here, I would just add that we would have read all those stories that kings used to have their astrologer. They are used to call

Royal Astrologers. So that but not everybody can afford Royal Astrologer. Even today, I know lots of big politicians, some of the people who are very top in India. So they consult and they have their own panel of Astrologers, advising them constantly. But that facility is not available to the common man. So the benefit of AI is that and technology in general is AI, in particular,

that it democratizes everything including astrology.

Rishi Ahluwalia (54:22)
And thanks for leading the charge on that, Punit. And I think five years down the line with the penetration of internet in India and the globe and the affordability of internet as well, this will probably reach more than a billion or two billion people.

And that's how the next wave of growth will be defined through AI. And I think one thing I wanted to mention also, now, we are at 5G. Maybe 6G unlocks the next level of growth that we have not even imagined. So again, in combination of AI, it will make a big breakthrough in the world. So great. Perfect. Perfect. Thank you so much for that,

Punit, these are fascinating answers. Now I think let's probably increase the pace of this podcast and dive into a rapid fire round. All right. So I'll probably ask you a set of questions here, and you can feel free to respond in any manner you want. It just needs to be rapid enough, all right, so that the audience understands your preferences and the way you are thinking.

So the first question is, what's the most misunderstood thing about astrology?

Punit Pandey (55:33)
So yeah, this is an age-old debate that astrology is science or not. I think astrology is science and people, not everybody understands this. So people do expect magic from astrology, is not really, it works in the same perimeter as any other applied science like healthcare or this medical science work.

Rishi Ahluwalia (55:53)
Absolutely. And the next question. So what's one AI model that you admire outside of Astrosage?

Punit Pandey (56:00)
I am a big fan of Anthropic models especially the agentic and tool use capabilities. Again, when I first encountered those agentic capabilities, I was kind of shocked or surprised whatever is the right word. Yeah, anthropic doing really great. They are popular among software engineers but even for those non-engineers that agentic models are extremely useful.

Rishi Ahluwalia (56:27)
Great, great. Next question quickly, Punit. So code or Kundli, which one came first for you? Code. I think, yeah, you mentioned that in the beginning of the podcast, right? All right, the next question is your favorite AI feature in Astrosage right now.

Punit Pandey (56:44)
Right now the latest one is AI Call. I would say AI Call. I am loving it and hoping that people love it.

Rishi Ahluwalia (56:53)
I'm sure they would. All right. So one myth about AI in spirituality, you want to completely bust.

Punit Pandey (57:00)
One myth of AI in spirituality. That is

probably tough question, need to think. I think, yeah, as I mentioned, generally people do think AI is something different. But as I mentioned that AI can act as a guru and astrologer and teacher and everything together. So right now, I think this is not well understood.

Rishi Ahluwalia (57:24)
True, I agree with that. Next question, if you could have a conversation with any historical thinker, who would it be?

Punit Pandey (57:33)
So yeah, the Aryabhata and even Brahmagupta, I am always fascinated by math and even the computer, the decimal system, the zero and everything that is amazing on which we have based everything. So all those algorithms that we see, so the foundation came from there.

Rishi Ahluwalia (57:53)
Yes.

Perfect. I was expecting you to say Aryabhata. You did that perfect. right. The one prediction AI can't make yet.

Punit Pandey (58:04)
So right now I think astrology is very vast and the AI models are not trained on all branches of astrology.

So there is only one branch of astrology and so sorry for taking a long time in the short. Yeah, only branch where this thing is the model training is done in the Natal astrology, Birth Chart astrology. But for the muhurta astrology, the most important is Mundane astrology where you predict about elections and earthquakes and the political and international scenarios right now, lot many things going on internationally.

Those the mundane astrology is, I think, currently AI cannot do that well and I am sure it will be doing in coming time.

Rishi Ahluwalia (58:52)
Thanks for bringing up the international point as well. I think everyone's kind of worried about that. Maybe AI can help us anticipate that better in the future. All right. So I think looking at the humility coming from you on this podcast, and I think that's how a person you are. So what mantra you live by while building from Bharat?

Punit Pandey (58:58)
Bye.

So mantra that we have is called innovative Indian. So if you open Astrosage.com you will find innovative Indian written in bold letters. So the idea is that we talked about Aryabhata in the innovation and development and contribution made by Aryabhata. So and I believe that Indians can make much bigger contribution than what we are doing currently. So

we must be awakened and we should start contributing more.

Rishi Ahluwalia (59:44)
Perfect. Yeah, I think this is the beginning for India and the creators and innovators in India. So I think that will have a lasting impression. Thank you for that quick responses there, Punit. I think the viewers are going to love that. Now, I think as we approach the last section of the podcast, it's more about giving back to the community. Now, with all your years of experience building different

Punit Pandey (59:51)
soon.

Rishi Ahluwalia (1:00:10)
products and being a coder at heart. So what's one lesson that you have learned while building conversational AI that every Bharat-based or any global founder should know about?

Punit Pandey (1:00:24)
No, yeah, one less lesson is we should look problems from our own perspective. So suppose if we are coming from the small town, so we should see problems there and see from their perspective. In tech space, the fundraising and the valuation, everybody starts thinking about it much more earlier than they should be. So building for Bharat,

having perspective of Bharat or whatever place you are there. That is I think the most important one.

Rishi Ahluwalia (1:00:59)
I think since you have come, the entire team of Astrosage has come from small towns as you mentioned before, right? So this is definitely inspiring for, let's say, an 18-year-old or someone who is at 14 years old planning what to do next. I think these are inspiring stories. And a bigger, wider perspective is much more needed in today's world rather than restricting ourselves to regions or countries

or stuff like that. So perfect, Punit. I think that is indeed what the audiences might be anticipating from a humble leader like you. So thank you so much, Punit, for your time today. Now, just to close this out, now from the heart of Bharat to the heart of AI, your journey, I think, will remind everyone who has tuned on this podcast,

that innovation is not just about technology.

It's about empathy, persistence, and understanding human emotions, which I think you have done exceptionally well while building Astrosage. And you had a great global perspective while working in the US. And then we are glad that you returned to India and are innovating from India. So Punit, thank you again for joining us and for showing the world how ancient wisdom and modern AI can speak the same language.

Punit Pandey (1:02:27)
Thanks Rishi and thanks for having me on your show. This is absolutely my pleasure and hope this kind of a conversation keep happening.

Rishi Ahluwalia (1:02:38)
No, absolutely. Thank you so much, Punit. And thanks for everyone who tuned in. Thank you very much.

Punit Pandey (1:02:45)
Thanks. Thanks everyone.