Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
How SquadStack is Winning the AI Services Race with Apurv Agrawal
0 Playsin 2 hours
How did Apurv Agrawal transform SquadStack from a data collection startup into India's leading AI-powered BPO generating ₹200+ crore ARR?
This episode reveals the 10-year journey to finding true product-market fit and disrupting India's massive outsourcing industry. Apurv Agrawal, Co-Founder and CEO of SquadStack, shares the remarkable evolution from Squadrun's gig economy roots to becoming India's most advanced AI-native contact center platform. Starting as a horizontal marketplace for data collection tasks, SquadStack pivoted multiple times before discovering their breakthrough in distributed calling services during COVID-19. What began as weekend experiments became enterprise solutions for major brands like Kotak, Axis Bank, and Bajaj. Apurv reveals how they built a billion-interaction dataset, achieved 95% AI implementation success rates while competitors fail, and created a unique hybrid workforce combining stay-at-home professionals with local agencies across tier-2 and tier-3 cities. He candidly discusses the emotional definition of PMF, why founders can't fight market forces, and his contrarian leadership philosophy of "leading with love" that attracted 19% ex-founders to his team. With voice cloning technology and hyper-personalized customer journeys, SquadStack is reshaping how enterprises handle sales, support, and collections through AI-human collaboration. He shared this transformative journey in this insightful conversation with host Akshay Datt, offering invaluable lessons on market timing, vertical focus, and building sustainable competitive moats in the AI era.
Key insights you'll discover:
👉How Apurv built SquadStack into a ₹200+ crore ARR leader through strategic pivots and market timing
👉Lessons from transitioning from horizontal gig marketplace to vertical AI-BPO specialist serving enterprise customers
👉SquadStack's breakthrough approach to distributed workforce management combining individual professionals with agency partnerships across India
👉Revolutionary AI implementation strategies achieving 95% success rates while industry reports 95% failure rates
👉Deep dive into voice cloning technology and hyper-personalized customer journey automation transforming contact center operations
👉Founder philosophy on high-agency hiring
👉Ubuntu leadership principles, and why love-based culture attracts exceptional entrepreneurial talent
#BPOstartups #AIautomation #IndianstartupsPMF #productmarketfit #distributedworkforce #contactcenterAI #voicecloning #enterpriseAI #gigeconomyIndia #BPOdisruption #remoteworkIndia #startupfunding #founderstory #entrepreneurshipIndia #AIimplementation #customerserviceAI #startuppivot
Disclaimer: The views expressed are those of the speaker, not necessarily the channel
Recommended
Transcript
00:00:00
Speaker
So we have actually cloned a lot of our callers. It's indistinguishable,
Introduction and AI Callers Discussion
00:00:04
Speaker
right? You just cannot say what is AI, what is human across languages, across emotions. Then no one cares if it's AI, if it's a monkey, if it's like someone, if it's Elon Musk calling. No one cares.
Founder Introduction and Company Overview
00:00:19
Speaker
Apoorv Agarwal is the founder of Squadstack. Squadstack is an AI-powered platform that helps brands like HDFC and Axis Bank to scale sales and customer service.
Can AI Replace Gig Workers?
00:00:29
Speaker
I'm wondering, does
00:00:33
Speaker
AI replace gig workers?
Finding Product-Market Fit (PMF)
00:00:44
Speaker
Apoor, welcome to the Founder Thesis podcast. My thesis is that ah PMF is a journey, and ah you are somebody who's going to help us learn why it's a journey.
00:00:57
Speaker
um ah First of all, you know just ah ah describe what is this PMF? What does this mean? and you know right That's a great question, man. I think that's like that's a great question. And it's actually, it's I think, one of the hardest concepts to understand as well.
00:01:15
Speaker
ah Someone really smart defined PMF as a name motion for the founders especially, where where you know you are your customers or folks you want to sell to, they are like really banging your door so hard. They're basically breaking your door, right? And you're just not able to handle the capacity of demand that you're getting.
Apoorv's Journey to PMF
00:01:37
Speaker
um So I think that is a fairly decent definition of PMF. I think that that feeling that you know you always have a product that the market loves so much ah that they are just like grabbing it more than what you can deliver. So you always feel like you know I'm under capacity.
00:01:55
Speaker
I think that i think is a fairly decent definition. Isn't that a very high bar? i If I think back to the last couple of years, I think only Chad GPT would have...
00:02:07
Speaker
seen that situation where there is more demand than they were signing up people ah you know there have been a few of these handful of these breakout products right right no i think uh i think uh no man i would say that at every scale uh that breakout happens right i mean we ourselves are feeling the same thing actually right now that we have a lot more enterprise customers that we can sign up and i am i'm actually running everywhere to recruit great
00:02:39
Speaker
folks who can help us you know implement ah our platform to these enterprises. right so So I think it happens at every level. Of course, I think what has happened with something like ChatGPT is ah its truly historical level of PMF, I would say.
00:02:55
Speaker
But yeah, I think PMF has that kind of level of bar. And if you think about it, like really, i mean, it is a high bar. right I think how many products truly fit in a market that it's amazing. And of course, i mean, you would probably say maybe a few hundred a year or something, right? Across the world, across categories.
00:03:14
Speaker
So it is, of course, a very high bar. Okay. So, you know, we've not ah told the audience yet who you are and why you are ah qualified to talk of PMF. Give us an overview. You've done multiple ventures, even your current venture, it's now on its 10th year. And I feel like you've really lived through that whole journey of finding PMF multiple times. ah Just give us an overview of who you are and why you are qualified to speak on this.
00:03:45
Speaker
Sure.
Early Ventures and Career Path
00:03:46
Speaker
ah Yeah, so my name is Apoor Vagrawal and I am the founder and CEO of Squatstack. ah ah my I originally was born in Ranchi and moved to Noida, grew up in Noida in NCR. and And when I was in my third grade, I went to ah and my my father's office and saw a computer for the first time. it was like 94, I think, right? And and and yeah And I think I probably understood the meaning of love at first sight that time. And I don't know i don't know what it was, man. like I was just like so ah ah drawn to it and I would just sit all day.
00:04:28
Speaker
And I was probably the only child who wanted to bunk school to go to my dad's office. you know and And I think and i think you know the... the Everyone in that office noticed who's this kid who's just like trying to figure out something ah on computers all day. So they they I think the office was sort of throwing away a your computer and then they ah the ah they gave that to us.
00:04:53
Speaker
and and you know And so this was like a pre Pentium processor and I ambitiously installed a Windows 95 in it. So I don't know how many people in the and your audience will still remember that blue screen loading screen, right? So so I used to literally sit for one hour, man, like for it to boot up and and all the dial up connection. and stuff i tasted for the first time in that time. and then And then went on to pursue computer science engineering. and ah And at that time, I was in a long distance relationship with my
00:05:31
Speaker
girlfriend at the time, now wife Vishnu. And so she was studying law, I was studying engineering and then and then we were you know we were figuring out how to meet. ah right and And we didn't have, of course, we didn't have a lot of money to travel, etc.
00:05:46
Speaker
So what I could do, I knew I could write software. So I started like with writing software for everyone in my law. Even now if you go to like a piza store there, the billing system is written by me.
00:05:59
Speaker
we created a business out of all the final year students who need to basic say implement these projects. Right. So we created a business out of that. But I think something that in this experience, which was more ah directed towards this, like, you know, making yeah enough money that i I was very grateful to be able to not ask for money from my parents at all. like I was financially independent ah from my first year.
00:06:27
Speaker
So I think that was great. But I think in this journey, I came across like a nonprofit, ah a very legitimate nonprofit that had, I think, 200 plus sewing centers, old age homes, like almost like 50 old age homes, like really like something that you would think, wow, this is impressive, right?
00:06:46
Speaker
And then they were literally like... getting people to ah ah shoot like videos on a camera and like make CDs out of it and ship those CDs all over the world.
00:07:00
Speaker
And I was like, come on, man, like, do you know, there's something on YouTube and there's something on PayPal. You just like put this online you getting a few clicks like what are you doing and then they told me that you know uh we paid like a lakh rupees to someone and who basically gave them a blog spot type thing basically so but they got fooled right so uh So that was very interesting and I realized, oh wow, okay, all these nonprofits are actually running large organizations ah and and it's just like one, two people who are doing, you know, the, let's say the management, right, and a lot of volunteers.
00:07:35
Speaker
So we realized that they actually need a lot of functions like that web tech, social media, bookkeeping, legal support. uh so uh and you know and i was a kid who was very happy to uh or i was as a first year engineering student who was really happy to contribute and uh because i felt that i'm creating impact and i'm actually learning a lot right so so i realized that a lot of students like me across the world who would probably be uh willing to sort of give back to society, but also can improve their skills. Also, most importantly, as you know, get that certificate which helps you with your masters, etc.
00:08:15
Speaker
So yeah, we created a whole platform out of this. So you're basically... Like a volunteering platform. Correct, tech volunteering platform though. so So you contribute your skills ah and you know, ah so you can, so let's say content writer can write content, a legal student can help with like legal work, etc.
00:08:34
Speaker
And we would help nonprofits on one end. And I was really like, I was 17 when I started this, right? So in my first year and I honestly could not even, did not even know the meaning of the word entrepreneurship probably that time.
00:08:48
Speaker
And it sounds so funny now when I'm saying it to you because, you know, the world of Shark Tank, etc. I think, I think, i mean, the great thing is probably every 17 year old now knows what an entrepreneur is.
00:09:00
Speaker
but But yeah, man, so I mean, that is how like I think I really tasted ah the next level of that addiction of building. and and And yeah, that was very exciting. And then in my um and my final year, again, I was working with a startup in Delhi.
00:09:19
Speaker
It didn't go anywhere, but it was still very valuable. And then finally, I joined Varun, who had basically sold his company to make my trip and was starting 91 Stringboard. Original idea was actually to be ah sort of a Y Combinator of India and like have like a huge space where you also fund startups. and ah so so did So we were were trying to do all that, but we ended up being being the largest co-working business, the formal co-working business, like a V-Work model for India, basically. And
00:09:54
Speaker
Yeah, so that was ah again like a really great experience. Build everything from scratch.
Creating Squadstack
00:09:59
Speaker
and And that's where I met Kanika and then Vikas was my batch mate from Willow.
00:10:06
Speaker
And three of us then joined. And i think I think when I was surrounded by Sophie, to be very honest, man, You know, sometimes I used to read like why entrepreneurs start companies and and there are so many like a lot of times entrepreneurs are just very obsessed with a certain problem. Right. And they just want to solve that problem. like ah But to be very candid with you, for me, it was that that I i had literally grown up reading blogs of the kind of cultures that we have now in Scott Stack. And I just wanted to be part of that culture. you know I was just like reading everything about the valley.
00:10:41
Speaker
And i was just like really fascinated. And in the 91st Stringworld, we created like a little bit of a bubble and there were so many entrepreneurs around. And i was in rooms where young startups were talking about their ideas. I really felt like, you know, I have nothing to advise these guys, man. I need to just, but I want to be this, right? I want to build something. So honestly, I came with this whole thing like a very purest, right? Like that I just want to be building, right?
00:11:08
Speaker
And And I think one thing that we were very passionate about for me was the same thing that I saw in NGO fuel. Decentralization of work is a very interesting, ah a very interesting theme that in some level has played around my whole life, ah you know, and because even then it was crazy that I had ah you know thousands of contributors many of them were like much more senior to me and and ah and yeah and they were just like it was beautiful write a tech platform can just enable so many people to do free work in a way and create so much value ah so I think that same theme I wanted to like i think at some level figure out how do we make a much larger business much more commercial business out of this and
00:11:58
Speaker
And yeah, I was very young when we started Squadstack.
Vertical Specialization and Market Focus
00:12:03
Speaker
which earlier was called squadron when you started it was called squadron right it was called squadron and correct because the first idea literally was that squad is with a group of people right so we knew we always knew it would be decentralization of ah work and then the run was because we wanted people to essentially we wanted to create an army of people who can essentially be data collectors right so so we realized that you need to digitize a lot of data that is offline right so we were working with like zomato and
00:12:34
Speaker
uh and uh let's say housing.com folks like that would basically say that you know can you just go to a restaurant click the photo of the menu like collect some of that information and uh yeah so mean that's how it started and then uh so how did you ah how did you get your first few clients like you had this theme of uh decentralization of work which is also called the gig economy ah So you you had this theme in your mind that you want to build a platform. You had previously built that tech volunteering platform where people could come volunteer skills to NGOs. So you had some exposure to that ah space. ah
00:13:14
Speaker
But, you know, the you could have done something for tech. You could have done something. I mean, there were so many options. How did you land on this specific ah problem statement of helping ah companies like Zomato to go collect data from restaurants through a yeah through an army of gig workers right right right no man i think uh great question and i think uh a couple of thoughts right here i think first thing that we did uh uh which probably was a good thing in the beginning uh but definitely uh an advice to startups is that we were uh we were horizontal in the beginning so what i meant was i would shoot literally tell people
00:13:54
Speaker
that you know we have like this army of people and they can do anything what do you have in mind right and then then then and then so data collection came uh as a big thing in the beginning but uh but then there was also uh what we later in evolved to a lot of data tagging work for the ai world which we are uh living in now right so like for example for a snap deal like you know you have to tag what kind of caller is this etc right so so that you are basically training ai models so that actually became like a very big thing for us uh in the beginning but but yeah just to wrap that point i think
00:14:31
Speaker
When we started as a horizontal, ah first thing that we learned is that being a horizontal is actually not very valuable. right I think like, for example, maybe if you're the first one ever, like for example, Craigslist made sense.
00:14:45
Speaker
But I think if Craigslist is already there, and and also if you think, of of course, it's a great different kind of venture, but obviously Airbnb is a much larger business than Craigslist. i so i think that's the i think that's the beauty of vertical which i got to really appreciate uh that i think if you focus on a vertical then uh then i think you can go really deep right and like you can create really superior outcome for demand and for supply that in a marketplace like that. And I think that's where the real value exists also, because what I realized is in a world of 10 billion people, they're saying that, you know I'll give you thousand people and no value ultimately. Right. I think value is that how can you actually create a very deep platform ah that actually creates like a 10x outcome overall for everyone involved.
00:15:34
Speaker
ah Yeah, so I think in this in this, then I think that the two use cases that came up was this offline thing and the online, like this kind of data tagging stuff. And online was a little bit more interesting because we could get demand all across the world and not just India in the beginning.
00:15:52
Speaker
and And this was mostly driven by e-commerce. Like as retailing is moving online, so ah ah you need the catalog,
00:16:03
Speaker
to have more information so it is searchable. well Like if I want to if i want a 14-inch monitor laptop, then all laptops on that e-commerce site need to have that tag that this is 13 inches, 14 inches, whatever.
00:16:17
Speaker
so that it is easier to find, recommend, et cetera. Yeah, yeah, yeah. And a lot of unstructured stuff, right? See, I think a 14-inch example, ah that is kind of easier to do because it's structured data, right? So when when the catalog comes, you will automatically structure it.
00:16:33
Speaker
ah There was lot of unstructured data, which was not automatically, like the kind of caller type, et cetera. It was like the whole fashion category that was not structured data. Or like even like very interesting problems we solved. I remember, like for example, um you know what is it that someone would search as? For example, AirPods you were talking about. So air so so people so let's say even if ah e-commerce... web So a lot of people might actually call earphones as pods, right?
00:17:02
Speaker
So how the platform... Yeah, yeah, yeah. E-port is a friendly generic term. Yeah. Exactly. So so so the idea is that you actually then tag data that, okay, this kind of... Like if someone's searching for port, that actually means that they're trying to get to your phone, right? so So, yeah. So, I mean, everything, man. Like there's so much structuring of data from like from e-commerce to like driverless cars.
00:17:23
Speaker
You have to figure out like label the roads that where is it okay to go, not go.
Expansion into BPO and Calling Services
00:17:29
Speaker
ah Which we... ah So yeah, I think that that market also, we were able to, ah you're absolutely correct, so we did work with lot e-commerce marketplaces and and started growing very fast, but I think when the whole e-commerce bust happened globally, so we scaled up globally as well, and ah you know in some level, we had a snap deal of almost every country, like we had in Japan, we had a bunch of companies in the US, we still have many of them actually as customers, and
00:18:02
Speaker
Europe, everywhere, right? But then I think the e-commerce, like what we saw in India, I think that similar kind of impact happened everywhere. ah And in this period, interestingly, what happened is Mindra, which was one of our customers,
00:18:18
Speaker
ah they They told us, ka if you guys are doing all of this data stuff, do you know, this whole market, what is it called? you know This market is called BPO market. And you know, what's the much bigger market within BPO is the calling thing, right?
00:18:31
Speaker
That people need to actually call a lot. And there's lot of this operational calling we need to do in Mitra. Like if someone is not accepting a product and, you know, there's an RTO type call that goes. Right, right, right. Customer service, basically.
00:18:46
Speaker
Exactly, exactly. Customer service, operational issues, right? And so they actually told us that, you know, why, why i mean, you guys have this decentralized talent and then why don't you do that, right? And I think, ah and and and then we honestly just started that as a, so early day, ah you know, we were also very inspired by that Google's 20% weekend thing, right? That in weekend you can hack something,
00:19:11
Speaker
So this just came up. So we just said, OK, let's just hack it for fun. And then we ah we did it. And then some people ah you know started using it. And then we put a PM on it. And it was literally just like one person running. And it was growing really nicely. Which year was this when you launched Calling?
00:19:31
Speaker
ah So we launched that around 2018. And and and by 2018, you had already raised two rounds. I think your angel round, you had Dipinder Goyal also there, I guess.
00:19:46
Speaker
Once Zomatu became a customer, he must have been impressed enough. Correct. kind like come on so they' been there So we were very grateful, Akshay, that we had Dipinder. We have, I should say, sorry.
00:19:59
Speaker
ah We have Amit Ranjan, who is the founder of SlideShare, which I deeply respect. I mean, imagine, man, that type. the I mean, like no one, I mean, we still are debating can Indians build a product globally.
00:20:11
Speaker
Right. And I don't know how you will still know that SlideShare was created by a group of some of the smartest talent in the world. Right. yeah LinkedIn acquired.
00:20:21
Speaker
Yeah. Correct, correct. LinkedIn acquired SlideShare. um and ah And then... Anyone who's done an MBA in a certain era would have come across SlideShare. ah Everyone would. I think I'm sure everyone.
00:20:37
Speaker
they They kind of invented a term of content marketing and... UGC, user generated content. Exactly, exactly. So ah ah yeah, and then we had Rohit Kunal from Snapdeal. And they also must have been customers.
00:20:53
Speaker
Yeah, yeah, yeah. Snapdeal was actually our largest customer uh at that period and uh and yeah and then even infosys founders uh we were grateful to get so yeah so lots of founders supported us uh in this journey and to be very honest when i i think it was so early you know i sometimes think that that whole period of startups and this period of startups is very interesting and shows the growth of india as well right but now i think thankfully in the startup uh thing a lot of respect and anyway so we also evolved a lot in this phase and I would say so I mean that is all i would still consider everything pre-2020 actually if you see my LinkedIn I think that's like squadron journey we did that journey we are of course grateful to have ah an amazing team and
00:21:41
Speaker
and amazing group of investors that continue to support us. But for me, that was ah really like the past.
Balancing Supply and Demand
00:21:47
Speaker
ah What we've been doing since turn 20 is... ah Let's start from the the calling. So you, Myntra, told you that the bigger... What you're doing is BPO work and the bigger pie there is calling. And so you started a like a feature in it in which... One question, though, before we come to this calling feature.
00:22:08
Speaker
Yeah. You know, you're building a, like you're aggregating supply and selling it to corporates in a way. um So ah was it a bigger challenge for you to get demand, which is like corporates who want to outsource work? Or was it a bigger challenge for you to build supply, which is ah people who want to ah do these gigs?
00:22:29
Speaker
Right. It's a really great question. I think... um ah I think it changes, right? I think it it keeps changing, ah right? Especially in this kind of PMF journey or or where your market only is changing so rapidly.
00:22:48
Speaker
ah so So I think in the in the very beginning beginning, it was a supply, ah was a little bit hard just to figure out how to even engage supply. like And I think some a very good, I would still say,
00:23:02
Speaker
a principle of um within PMF, this concept of MVP, minimum viable product. Right. And I think, I think we have one of the best examples of that, that, ah ah ah ah that you typically say that for launching a product, you should actually not launch a product basically. Right. So I think what we did was we basically got a lot of, uh,
00:23:25
Speaker
at that time college students to use the platform and the platform was basically just a landing page where you sign up and then once you sign up and this is again back in the day there was not even whatsapp groups man so we basically did a gmail group a google group right so you would essentially give you and we but we did gamify that well we did what do you want to do in an app basically so you have to but in a terrible experience like you're basically using your standard google tool so you go somewhere then you have to add that your pin in your maps and then upload it et etc so we we ran the whole workflow without writing a single line of code right and i think and for like two three weeks and we and i think what we learned was really a gold mine right like i i give you one simple learning so one of the biggest thing we learned is that people were very happy like uh very enthusiastic to accept a job
00:24:17
Speaker
But ah only 1% is enthusiastic to do the job, right? Because there's really no... ah In some level, I think people procrastinate, etc. Because there's nothing to lose.
00:24:28
Speaker
ah And that's a huge problem, right? Because I cannot manage capacity like that. I cannot assume 100 people will do it and 1% does it. So... So this led to us giving birth to a feature of life in our app, right? Because we said that if there's nothing to lose, then how do I solve this? So so when you come to a platform, we'll basically give you three lives.
00:24:48
Speaker
And then, so if you take a job and you don't do it, then you lose a life, right? so so so So, so, so things like that, I think that would have been a great lesson in, in, in, in building an MVP and not, not really going, ah writing any code and actually learning.
00:25:07
Speaker
But they these college kids, how were they discovering? Was it pure viral word of mouth that, okay, you can make money, you can make pocket money here?
Homemakers as a Workforce
00:25:15
Speaker
Or were you also like Facebook ads and all?
00:25:18
Speaker
Yeah, no, no, no. I mean, it was all just hustle on ground. Like, yeah, you can make college money. We would, I think we were in Delhi. So we went to Delhi University and, you know, did some of that.
00:25:29
Speaker
But I should tell you there also. So we started with college students, but somehow the homemaker segment found us. right And then we also learned that gig is not very valuable in my opinion. What I mean is gi gig is where there's a lot of volume of people and they're just trying to do something to pass time. right and And that's why and ah ah probably this is my own definition that decentralization, I believe, is a lot more...
00:25:52
Speaker
valuable because you actually figure out um supply that and you're very clear that this very serious work, right? This is not something you just like do to pass time, which again depends on the category of demand. But in my opinion, again, with the verticalization thought process, I would say the that it's much better. Like it is the value can only be created when you have supply that is extremely serious about the platform.
00:26:20
Speaker
ah that they are doing and it it really matters to them, right? So so we actually in this journey... The platform becomes their source of livelihood, basically. Exactly. Livelihood or anything, but it has to be very significant, right? like So I'll tell you like our story, right? So we basically... um Sorry, I would say the homemaker persona, they found us, right? Through all this. and we And we used to see this college kid kind of like really like a garbage...
00:26:49
Speaker
I would say retention but retention we had. They would just come and in our five is probably the back end they did a party and then they're out. But then we saw this segment of, you know, homemakers who are like just coming and then I'll tell you a story.
00:27:07
Speaker
We have now 10,000 stories like that today. but But like say, you know, someone who's working in a IT job and and ah and very good like education, etc. And then And then let's say she gets married and moves to Ludhiana because, you know, the the the partner now has a business in Ludhiana, etc.
00:27:27
Speaker
So what do you do? Right. And there's literally nothing to do. um And in outside like the four cities in India, I think it is impossible to find a job that pays more than 10,000 rupees.
00:27:38
Speaker
Right. Half of India makes five thousand rupees a month. Right. So so it's really and I keep thinking more and more that how can AI and tech always be an enabler to create more and more opportunities for people.
00:27:53
Speaker
ah and solve for equity in a way but make the world more equitable. ah but But yeah, and and and and and then of course, there were single moms, there were women. And so we have a girl ah ah from Kashmir who was affected you know when there were a lot of challenges there. And and since then,
00:28:11
Speaker
At that time, especially like she was supporting like a huge as like a family of almost 20 people just by working in the platform. And she's been financially independent since then. so yeah, so so amazingly like beautiful stories. And in fact, I think especially in supply, and my biggest learning is that ah To be honest, we almost had zero people in our company for a really long time managing supply,
Enterprise Solutions and Market Needs
00:28:35
Speaker
right? It was such a self-managed platform, so community driven. No one would do fraud because people are really like, you know, trying to create value. We would create groups, people would help each other.
00:28:46
Speaker
ah So I think if you find a right kind of segment of supply, I think i think they'll build a platform because the value prop resonates with them very hard. ah In fact, we've been grateful that and we are 250 people as a team now and I think almost 20 people in our team, including some really senior product people, you won't believe are these ex-homemakers, right? like in fact one our most In fact, the lady who leads our supply now is also like one of but of our ex-platform people, right? so ah So anyway, so that's how the whole supply journey happened.
00:29:21
Speaker
Okay. And typically, what was your take rate? Like if a company paid 100 rupees for a job, how much would you keep and how much would go to the gig worker? Right. So we were we were not actually such an open platform. What I mean is that we were, ah ah because we were deeply verticalized a solution, ah ah it was not just that someone would say that, you know, I want 1% to be hired to start doing my calling and do 100 calls and I'll pay 300 rupees and then, you know, we take 20% or whatever, right?
00:29:55
Speaker
It was a little different. And that's again my learning on the demand side, which I'll quickly tell you. So, see, I think one of my favorite things that one of the leaders of, uh, A bank told me, karraur if you know, I love your technology. I love your mission.
00:30:11
Speaker
I love your AI demos. But, you know, I'll tell you some a secret, right? That no one cares if it's AI, if it's a monkey, if it's like someone, if it's Elon Musk calling.
00:30:23
Speaker
No one cares. What we care is if you're doing sales, there's a metric called cost of acquisition. right CAC how much are you optimizing that if you're doing support there's a vector call CSAT right and cost per resolution what are you creating a denx delta for me right and and that I think I love that how enterprises have to be ruthless around it right and I think we will come much later but I would love to talk about the 95% failure rate etc as well
00:30:54
Speaker
ah but But enterprises need outcomes, right? They are ultimately, I mean, the public companies, the V.S. shareholders are asking them that, okay, whatever you're doing, show me the metrics of growth, right?
00:31:06
Speaker
so so so ah So we were very clear that our our demand is enterprises, so they need services, right? And they need solutions. They're not going to figure it out. ah And so we built a manage managed platform. So in platform, marketplaces can be very various kinds, right? Super thin marketplace like a Craigslist.
00:31:26
Speaker
is just a directory or you can be like a much thicker marketplace where you go very deep on that vertical and solve the problem so as a simple example when we finally understood that you know the biggest use case is this communication problem that an enterprise has with their customers it could be around sales it could be collections it could be support service ah Now, one big part, of course, of that is the supply or people actually making those phone calls and talking correctly, etc., which is fine.
00:32:01
Speaker
But a very big stakeholder we understood here is that the enterprise whose end customer, let's say, is Akshay, who is now trying to buy an insurance from Tata AIG is Akshay, right?
00:32:15
Speaker
so So, and then when we started, saw this problem then then we realized that it's actually not a lot of people are thinking about how do we improve the lead in all of this is supply demand and we always have our roadmap as like what about the endper empire right how what about the lead ah how do we improve their experience significantly, right? So right from the beginning, and in fact, that was, a I would say, one of the first, ah ah let's say, 10x things that we did in the beginning that everyone, like the whole traditional BPO is very simple that, you know, like bunch of people sitting in a a huge ad office will just like dial, dial, dial and like just repeat
Enhancing Customer Interactions
00:32:55
Speaker
the scripts, right? So,
00:32:56
Speaker
ah ah but one of the most important things we realized that, okay, Apoor's Akshay, Apoor's mom, very different people. All three of us, by the way, get completely different ads of the same product on Facebook.
00:33:12
Speaker
Marketing became programmatic, right? ah Like maybe two decades back. ah Now, but all this communication, like sales service, It is still very much same as like I think what it was 100 years back. It's the same thing, same thing. Why? Like if my mom is seeing a different ad than Apoor, then why are we not doing a different level of communication channel, right? Maybe as an example, if someone calls me right now and I really want to buy that product or service, I'm not going to pick it up, right?
00:33:42
Speaker
But maybe if you set up, like if you reach out to me on WhatsApp and then, you know, and set up a time with me in the evening, then I would be a very good candidate for that. so lead So how do we drive much better end buyer experience, lead personalization? So we built so much tech on that. In fact, we are the leaders in that side of the tech globally, because a lot of this thinking has happened on the enterprise side that as an enterprise, if I want to sell to ah a Tata, you know then how do I write email in a beautiful way? how does
00:34:17
Speaker
like They call it the whole term called account based marketing there. right But no one has thought about in the consumer world. right That is if an end consumer is looking to buy, there's a lot of intelligence and personalization that we should try and drive through tech.
00:34:32
Speaker
um And again, we'll come back on this in the world of AI. But I think ah so. So that's what ive I wanted to tell you that the the platform basically became very thick, right? Where the platform's outcome or the job was to drive those superior outcomes for an enterprise partners.
00:34:49
Speaker
Of course, a huge part of that was, ah was the, was or is our colors and their quality and, and, and all the tech that we have built to also manage their life cycle.
00:35:03
Speaker
Interesting. ah yeah So essentially you're not charging per task, but you're charging per outcome, like per sale made rather than per call done. ah No. So our business model was per minute as per minute of interaction plus ah plus some kicker on per sale also.
00:35:20
Speaker
ah but But that's also... this So I'll tell you why that is more disruptive because traditionally the model was per seat, right? Which means you're just paying for...
00:35:30
Speaker
ah for I don't know what right just like you're just basically paying for unproductivity as well right like so then you are then so you basically buy these seats and then you you're not sleeping at night thinking oh what are they doing and ah well what is happening is it productive or not so we said you know what main main outcome for you main output for you sorry I should say is is if I'm driving high quality conversations uh and uh so that's what you pay for and then and then there's also a cut that we get on the final outcome
00:36:01
Speaker
ah So that it works for both service and sales type use cases. Interesting. I remember there was time when telcos had migrated from per minute pricing to per second pricing, something like that had happened. And ah yeah, that was when the,
00:36:18
Speaker
the price of a minute mattered. Yeah, yeah, yeah, yeah, man. I remember that pre-WhatsApp, we used to all, in SMS, we used to ah be super creative with acronyms. Yeah, yeah, yeah. Because every letter mattered.
00:36:36
Speaker
Yeah, yeah. So, ah you know, how did Myntra saying, do customer service migrate or evolve into you selling insurance policies?
Impact of COVID-19 on Business
00:36:49
Speaker
quite Right, right, right, right. Yeah, I think ah ah so. So yeah, so very interesting thing that happened in all of this is also being COVID. ah Right. I think so. What happened was that what was telling you. So yeah, this was like a this was to be honest, this was a side team that we did not even get valuation for right when we raced our round.
00:37:14
Speaker
Which round are you talking about here? The series A round also. So even after the seed round, you know, like this was... The series A in 2020. Sorry, the seed round, so the bloom round. Okay, in 2017. Yeah, yeah, yeah. So this was considered a side thing. 2017, we were still doing data categorization or data categorization. Yeah, yeah. And we were fairly large. And we still, by the way, do that. That's a huge part of our platform.
00:37:39
Speaker
What was your revenue then ah in 2017? So 2017, would be like, think 2017 would be 5 crore annual revenue. And then 2018, we would be like 10 crore, 10, 12 crores type revenue. eighteen we would be like ten guro ten twelve kros type i mean Yeah, yeah, yeah. So fairly decent. That's what I'm saying. Like it was fairly decent and we were ah and we were and we had also cracked the US, right? The holy grail for all the all the ah founders building B2B.
00:38:10
Speaker
And yeah, and I remember I was in the US when, you know, Trump won, etc. So I had remember like all of that period like, ah you know, when the US selling very hard and we were becoming hot also. But But yeah, I mean, a lot of that revenue crashed also, man, because ah because I think whatever happened with Snapdeal, think a lot of these marketplaces were honestly, maybe they were overfunded, I don't know, all over the world, right? So, but but yeah, I think we were just... Plus, technology would have also automated a lot of what you needed humans for, right?
00:38:47
Speaker
No, so I think that way that was not a problem because we have continuously, we were actually, so we were automating it ourselves also. So we basically started providing a service that we will basically do the automation. So you are basically at reducing your cost significantly. So you can give us more scale and we would start doing more complex things also.
00:39:08
Speaker
So I think that was not a problem. i think the I think this is the lesson, I'll tell you, the so this was the biggest lesson I learned where I think as a much more younger person,
00:39:20
Speaker
founder with lesser experience, I would say wisdom, I would say probably, right? I think is that the r key I used to always, I used to say, you know, great founders and teams build markets, right?
00:39:31
Speaker
To be honest, I think it's foolish. I think anyone who says that, mark and I mean, nothing is greater than market, right? I'll tell you why, because market is nature. ah And so I'll use an example to explain this. So so ah So, I remember with a team member, we were in the, ah we were basically, he was in my home and we were discussing this and he essentially was smoking a cigarette. Okay.
00:40:02
Speaker
and And when he was smoking the cigarette, to ah so basically it was in the homes. we went up So we were in the ah window cell essentially. right so So and I think the air was essentially coming in this direction.
00:40:18
Speaker
And then he was essentially not wanting, obviously, the smoke to come in the house. So he was essentially smoking really hard. He was pushing it very hard that it goes out.
00:40:29
Speaker
But it just wouldn't, right? Because the air is like coming with all force. It was coming inside. And I told him, dude, this is market. I still, so I'll never i forget this example. think the best example for me to understand. And I think anyone who says, you know, I would say, agree okay, jump in the ocean and change the tides, right? Like, come on, man. Like you cannot fight with nature.
00:40:49
Speaker
And market is exactly nature, right? And you have to respect nature as humans. i think you have to figure out how to integrate with nature. Nature doesn't have to integrate with us. right We keep saying world is going to die. World is not going to die. We will die. World will be still here and nature heals very quickly. right So it's the very same respect you have to honestly give the market. And that is when i understood that you know not understanding the market really well is a very ah big mistake. And I think you should really follow the market um correctly. and And that's where it comes back to your product market fit. And that not only the market...
00:41:26
Speaker
ah I think having a very large market is just so much more easier. And I truly now understand why even VCs, you know, would always be very bullish on you on a market that is at least a billion dollars or or more, et etc. Right. I think I truly understand that because what I've seen is that.
00:41:45
Speaker
ah And my my golden rule is simple. I think have a market, ideally $10 billion, bare minimum one billion dollars, and then have a 10x offering, right?
00:41:56
Speaker
If you have these two things, nothing can stop you. And that is where the PMS thing happens, what I was telling you, right? I think that then the then people are banging on your door. Because you have a 10x offering and because the market is so large, there are so many people. right So I think either and that is why the PMF keeps changing, right? Because your market is also evolving. And what is your offering truly 10x today?
00:42:20
Speaker
Right? Like if you again, let's Think of actually Google, right? Like Google search. I mean, who would have thought in our lifetime that Google search will not be a 10X product? Suddenly, like Gen AI came in and then everyone said, oh, wow, why will I use search now? Right? I don't want information. I want knowledge now.
00:42:36
Speaker
So everyone then went to went. And then, of course, Google responded so strongly. Right. but uh so it' certainly a 10x product overnight was not a 10x product so i think that is a beautiful example of uh again pmf and i think we also learned this the hard way that uh i think the mistake we made we were 10x product in a in a very tiny market which is also shrinking uh in general and uh there was a lot of consolidation that happened in these marketplaces and and i think that's a bad place to be uh
00:43:09
Speaker
But customer service also is a large market, right? ah Though you are more sales focused than customer service focused. So what made that change happen? yeah so the change that happened essentially was uh uh i think so we were a little bit clear that you know this this business is not becoming very large uh and we are actually seeing like our customers get out of but run out of business basically and uh so i think we ourselves at that time were also profitable i remember and it was really nice to be like that but
00:43:44
Speaker
But all our customers were massively not profitable and you know and they were they were running out of basically this VC money. right so So we ah ah so we we we said, okay, we have to figure out something. and then And then we knew this calling thing is there on the side. And I was looking at very deeply that, okay, what's happening? What can we do here?
00:44:04
Speaker
and And then, time I mean, ah Sorry, I was going to say thankfully COVID happened, but of course, when um I mean, i like a lot of founders feel similarly that thankfully COVID happened. if Find a market or PMF. yeah Yeah, man, I think it was just like ah ah change happens, I think, right? And the only constant is change. And and and yeah, and I think and that was a huge acceleration because before this, um like we couldn't crack the large enterprise of India. I think the larger enterprises... By 2020, what were you doing? You were doing that e-commerce data business and you were doing customer service.
00:44:47
Speaker
Correct. And we were still doing our calling based stuff and the calling based stuff had also come to a decently large scale, mostly working with large startups. And I think rather than one, we would have probably had 15 people team there and and they were super profitable, just like, ah you know, but no one, well, it was not very serious in some level for the whole company, like the board would never care about it. It was just like one of the things we were doing.
00:45:13
Speaker
Okay, okay, okay. Like for Myntra, this whole thing of shipping and returns and all that is just a thing they have to do. to It's like a hygiene factor for them.
00:45:26
Speaker
Correct, correct. You would be like a vendor who's good to have, not a must-have vendor for them. Yeah, yeah, yeah, yeah. And I think our we were also not very, I think were just, ah ah we never got that.
00:45:41
Speaker
Yeah, and I think what you said is a great point, right? I think that was my biggest issue. That when I, ah ah Because then i was seeing this, right? I think as these companies were running out of money, the first thing they wanted to also cut was basically cost, right?
00:45:56
Speaker
um And so I was very curious that, okay, ah that what can make more money for enterprises? And I think that's when coincidentally, so do these two, three things happened. I think one is that the platform started getting discovered by some of these fintech companies and they started calling me.
00:46:15
Speaker
K.R. kiya you know, Aapur, we love this decentralization thought process because there's this aspect that definitely scale up and down with volume you know I can't hire 100 people as a fixed asset etc and I'm and and I asked him okay how much are you how much are you paying and how much are you willing to pay you know then then very interestingly they said that we don't care how much we are paying uh here what we care more about is the value that we can get because it's more about sales right so I don't want to lose one sale also
00:46:48
Speaker
and And that was very, very interesting that, okay, wow. And the second insight we got is suddenly enterprises who were, so all the enterprises started calling us. or that you know i mean And they used to look at us that we are like, I don't know, we are nuts or something. We are saying that, you know, suddenly like tech will be there and people will be remote. and they were like what are you saying man and now decentralized uh workforce became exactly normal exactly exactly exactly and and and honestly for good reason right no one wants that all the banking data is sitting in some house in uh in Vijaybada or whatever right so but then the thing is key honestly tech had solved all of that problem and we had proven to people that you know it's actually more secure than any of your contact center that People are just taking out pen drive, etc. right so
00:47:42
Speaker
ah So I think ah people really understood that and they they I think this whole thing was a more of a mindset issue, right i think more than logical issue. I think when people started seeing that remote is the only option that time, so they had to figure it out. And then we were the only ones who were actually ready with large scale tech, etc.
Challenges in Scaling Globally
00:48:05
Speaker
Compliances.
00:48:07
Speaker
so So, yeah, we we got a lot of pull for both our first sales use case service, etc. and the remote also.
00:48:18
Speaker
And that's when we knew that this is actually much more interesting than everything else we are doing. And in fact, ah At that time, I had actually moved to the Bay Area as well and and then I came back. so and and you know And the golden rule is that if an Indian startup can somehow ever figure out more than a million ARR from US, then that's like the best thing possible and then you move there and then you know you build.
00:48:48
Speaker
So we were doing all that, you're very clear. But again, like ah no one thought if no one planned for COVID. right and And this happened and suddenly like but this became very interesting. And so I moved back immediately. and Actually, i was already kind of stuck in India at that time.
00:49:07
Speaker
This calling thing wouldn't have worked for US, right? Because of accent. No, so we were actually, so we did, ah so we we were at that time trying the US and we were in fact very big in real estate US there.
00:49:20
Speaker
But then the problem is that US completely paused, at least our US real estate paused for six months and the cost waste was so high that we couldn't figure it out and to be very candid with you i think we did struggle a little bit on the supply side us for sure because i think it was just not uh it's just very expensive also and i think we did not have like we just had sales team there right and So, like the quality that we were able to build, but I think we could have figured it out with enough time.
00:49:50
Speaker
I think we were possibly U.S. would have only happened if you also built supply in U.S. I doubt. Yeah, yeah. No, we did. So that's one thing. Oh, you built supply in U.S. We did have like probably 30, 40 callers and we had a very decent business in real estate U.S. like ah Millionaire Plus, which I'm telling you.
00:50:06
Speaker
But I think that also paused basically. And yeah, so anyway, so so... What was your error when
00:50:18
Speaker
yeah I will have to... Single digit million... Yeah, yeah, yeah, yeah, that also got a lot of it basically disappeared as well. I remember that time as soon as COVID happened, I remember all of us founders were talking to each other and and the boards sent us scenarios, okay, and then they would send us, ski okay, plan a scenario in which COVID will end in three months and plan a scenario in which COVID will end in three years.
00:50:45
Speaker
And like, dude, what are you saying, man? Like we're startups. We don't even know. what to do and we can probably of course you can plan if covid ends in three months i will go all aggressive and not touch anyone if three years i'll fire everyone other than me right so so it was really really a funny time but uh but anyway and like our whole world collapsed but we uh but i think gave us so much opportunity suddenly like large enterprise in india like Kotak, Kota, Axis, Bajaj, those type of groups started working with us immediately. So I think that was a great ah thing. And then we just basically like went all in. And that's when I would mark ah that. I would say that is ah when I said everything is that before that is just like gone pre.
00:51:36
Speaker
that That was like your first PMF in a way. Yeah, no, but I think, see, that's the thing. Actually, that's what I'm saying. I think PMF, we it's a journey, right? because Because, see, because all of these different things we did, we had, see, all of this. Yeah, you hit certain scale, you had.
00:51:53
Speaker
Exactly, exactly, exactly, exactly. But that, I think that kind of door banging was not happening as much, right? I think continuously. I think that's the problem. You need that continuous door banging, it will keep happening. And I think then you know, okay, okay you continuously are doing the right thing.
00:52:08
Speaker
I think first we had a 10X product, but there were not enough people in the door or they were they were and they had other problems a ah or whatever reason. But I think this is the first time we had both happening. right I think we were customers who are loving what we are doing. We were a 10X offering.
00:52:28
Speaker
um and yeah and then we And then we had a large market. So we said, okay, you know what, like ah why should everyone go to the US? Because in India, I understand no one wants, I mean, it's hard to build a pure software business selling just in India, right? But I think the world of AI and the marketplace type stuff, I think there you can really do that because the AC, the ah like for example, we make anywhere from $300,000 to $1 million to even $2 million dollars with some customers.
00:53:00
Speaker
So it's like a lot of money that you can get from one customer, right? And if you can, ah let's say... is Now your TAM is their sales team spent, like what they would be paying to a BPO to run their outbound sales process. Exactly. It's not a or a SaaS Apps, exactly,
Decentralized Workforce and Agency Collaboration
00:53:21
Speaker
man, exactly. So like I used to tell our, you know, the sales people that we would interview that you make 200 bucks a user, ra I make 40,000 rupees a user, right? so so ah So yeah, so it was... But was was your distributed workforce of stay-at-home moms good enough for, you know, for for a Kotak Mahindra?
00:53:46
Speaker
Yeah, yeah, yeah. So that was great. So it was absolutely good enough and we were able to, we scaled up more. And again, I should tell you that these are people who were working minimum half day. So they are either working minimum half day or full and some people are working average even 10, 10 hours plus a day, right?
00:54:04
Speaker
So, so, so the minimum was that. And these are very serious people. They were completely, their livelihood was linked to us. Or, I mean, some places, i mean, there was something very, like they would have basically been extremely upset and there was no alternative if we would not be there. Right. I think that there was a very significant kind of push there. And again, we, we built a lot of tech to solve. ah ah So this was a great period where, in fact, when we get to our, so I would say, so in this,
00:54:34
Speaker
So in this like big PM first, like great PMF where we are like seeing this door banging and, and in that time, uh, uh, the first act of that, I would say is the Uber act in a way, right? Like you have decentralized workforce and you're working with enterprises doing very well.
00:54:52
Speaker
a part two of that act to also answer your question is that when we, uh, started working with some banks and, uh, certain processes, then, then there was an understanding that, uh,
00:55:03
Speaker
you know, the compliances for which you need people or there's some part of the process that we're just not comfortable putting it outside a building. And that time we realized that, you know, the decentralization, we have just individuals, but we can actually add agencies to it as well.
00:55:18
Speaker
So like, for example, in a Bhopal or Indore or in Vijaywada, there are like people running small BPO's with 100, 500 people, 1000 people. And I realized BPO is a cotton industry of India. Every street you go to, there's someone who's doing it. and and And then we realized that, you know, they are all like good people in a way. and And they have actually, they are really resourceful,
00:55:45
Speaker
in how do they get the right kind of talent and they make sure they work every day, e etc. So we said, you know what, why should we build that capability or how to find like a good office space and run the office space in a city that none of us have ever visited, right? so So we said, let's leverage them and add that to the network also. We added agencies to the network as well.
00:56:08
Speaker
And that again was a huge, I would say that was a very big game changer that opened up a lot of the market that we couldn't also do. ah and And yeah, and then the third the third act that has happened really, ah I'll tell you before that is for this to happen, we needed to,
00:56:28
Speaker
ah like we did Like I couldn't tap Akshain Se Ki Bhai Ye Kar De, right? Because this is someone sitting in their own home and all over. So we had to build a lot of ah tech to understand, you know, how do I screen the right people? how do I match them to right jobs? And most importantly, every interaction that you're doing.
00:56:52
Speaker
ah We were analyzing 100 things out of that, right? Like, okay, is this through this interaction, is this person like great and should they continue to do or which or let's say, let's say for Kotak, right? Like if someone said on the call that why are you, why are you better than Axis, right?
00:57:12
Speaker
and And then I have all of these callers and of course they have probably even more callers. No one even knows what's the best answer, but there's a best answer, right? Someone is saying it and we, of course, the data was seeing it.
00:57:24
Speaker
So now then we are also using that to pull up that best answer, spreading it to the network that this is the right one. ah you should like practice this etc so so I think using like what we call as co-pilot now right so we're the first AI co-pilot of this industry where we built a lot of tech to manage and that of course in luckily gave us kind of a lot of data I think we have the best of best data globally I don't know if anyone else would have because we are the only full stack company globally so there were have been tech companies that are trying to do it
00:57:55
Speaker
but adoption has been generally very less, right? So even the CRM adoption is like literally like 12% or something. so So what will you do? And the first lesson of AI is that AI goes through this journey where there's data, then it has to be hundred called close to 100% captured. And then there has to be labeling, tagging,
00:58:15
Speaker
take insights out of it. And then you can build, of course, machine learning and agentic workflows. Right. So I think I think by having like all of these people, we built a so we built like our own mobile app, like not using any CRM in which the way they use the app, it was designed that you will do 100% data capture.
00:58:32
Speaker
So again, to be honest, I would love to say that you know we were so visionary at that time that we said five years back you know that Sam Altman is going to change the world and we'll be the best data company and we'll build the best application. But we were just trying to do our job, man, and convince like all of these enterprises that you know even with type, we can manage these resources and no fraud, et cetera, can ever happen and all of those things.
AI Advancements in BPO
00:59:00
Speaker
And every call would be recorded, transcribed, analyzed. And you were probably using some cloud telephony instead of Like people calling from their phones, they it would be a call through their app with some cloud telephony service behind
00:59:38
Speaker
first of all, one language. We speak in English, right? And so so I still struggle with that. And this podcast is getting transcribed real time.
00:59:48
Speaker
But when I look at the transcription, whatever you say in Hindi will be like in Roman, you know, like. Exactly. so So you should really use our speech to text tech for that. I'm happy to give you an API. But basically we built ah all of these things like natively then, right? Because then we're like, okay, how do I do like i mean people are speaking hindi english and tamil in the same conversation so so so if you if you like for example let's say have a ah someone who's in chennai and uh and uh and talking to a caller then you'll hear these three languages right and it's very natural so so so we had to basically build all of this uh speech to text layer et etc as well and uh and yeah and i think that really opened up the uh uh i think like
01:00:34
Speaker
Obviously it's incredible what NVIDIA and some of these open AI, Google, et cetera, have been doing. And of course the, I think the most exciting thing that's probably happening in our life, lifetime till now is AI. and And then we were just naturally placed for it, right? I think we were, I would, gent I am intentionally using the word lucky. I think that we definitely are. And and yeah so we were sitting on like this beautiful,
01:01:02
Speaker
data sets we have distribution with enterprises and our positioning in the enterprises that you know they are the tech bpo ai bpo or something we used to call ourselves and uh and uh so once you started this uh uh like the the hybrid model where you had ah gig workers but also agencies by that which which year was that 21 22 which year was that yeah Yeah, that no that was 22 to 24 basically.
01:01:31
Speaker
Okay, and what kind and you also raised a pretty big round in 22, I think 17.5 million. Correct, so roughly 150 crores, correct. um And Burdelsman led that round along with Chirate and Bloom. What was the pitch then?
01:01:51
Speaker
So the pitch was this only, right? I think what I gave you, so we, we, so. Like sales outcome as a service. Yeah, yeah, yeah. So we were, so we, so we basically were positioning ourselves as talent marketplace for inside sales, right? Because, because if I would use the word BPO,
01:02:09
Speaker
then I think all the VCs would have run away. actually But thankfully, budgetsmen did not. As it's like BPO. Correct, correct, correct. But the BPO had such a bad connotation, you know, like people used and and And I love, honestly, I'll tell you now over time, what I told you about the nature of market, right? I love boring industries because Boring industries that are especially so large, right? I mean, BPO market is so large, man. I mean, we talk about e-commerce market, etc. They are nothing. Like if you literally, India has something like almost hundreds of thousands of people like just domestically serving and talking and doing this work right and globally uh it's quarter trillion dollar market right i mean mean it's ridiculous ridiculously large market and uh and yeah and there was really no innovation i mean five years back when we came in ah there was zero innovation right i mean uh so so i think we we just saw the whole problem and truly now in the world of ai
01:03:09
Speaker
Apart from the AI stuff, which we'll get back to later, but I think that, by the way, and my biggest feature still is all the and lead personalization, lead management, which is, of course, becoming 10x with AI. but But even like basic things like how do you connect to only channel, etc. are still like really, really critical features, right, that helps me versus anyone else who's who can build uh voice port or something like that right so so how does this personalization work suppose um you onboard a let's say a fintech which is selling uh let's say uh let's say loans okay uh say loans to people who are salaried and uh
01:03:51
Speaker
So how would the person, they would give you data of leads, right? That these are people who have filled out some online form, have somewhere shown interest or ah have downloaded our app or something like that and call them and pitch the ah loan to them, right? so How would you personalize this?
01:04:07
Speaker
Right. let me Let me tell you that actually in a different way, right? I think first, like if I'm sure all of us had that experience that... That you will talk to a brand.
01:04:18
Speaker
You sign up in a brand. And let's say and then you buy it. Then also they will keep calling you. Right? And you will get so pissed off sometimes. And you'll say, man, like what the hell are you
Customer Journey Optimization
01:04:29
Speaker
doing? Why are you killing me?
01:04:31
Speaker
Right? It's not that they are bad people. Or they are you know really trying to attack you personally. It's just bad systems. Their systems are not all people. Yeah, the the outside call center doesn't know...
01:04:44
Speaker
who's our customer exactly exactly in fact we used to tell our enterprise customers so there was this term called digital transformation which became a very uh critical thing for enterprises to ah become much more digital friendly right and i used to say that your digital transformation goes to death at the contact center right uh because there's really no data flows at all and uh and yeah exactly so i think uh the ah So the problem is that these systems don't talk to each other. So the first step of that personalization journey was first, okay, let's map your customer journey, right? Because these are, ah imagine like a decision tree that is ah like almost like 1000 square feet, right? Like it's like a huge decision tree.
01:05:29
Speaker
So you will say that that, okay, this lead who has, let's say, ah as you said, like it's a salaried employee looking for this kind of loan. and Okay, now in which step of the app, ah were they a mobile app? Are they in your web platform? Are they your in your like store where you give it?
01:05:48
Speaker
So where are they coming from? Which step are they in? Based on that step, what should you communicate to them? Which channel should you be using? Right? So I think there is still just like mapping out the whole journey properly itself.
01:06:01
Speaker
is ah something that we do as a step one, right? That when we work in enterprise, hill let's map and design our journey and actually put all of that in our platform.
01:06:12
Speaker
And like we have like this basically a journey builder where you can essentially create through logic trees like okay if this happens then we do this etc right so so that uh that is that still is like i would say one of our biggest like 10x value props where no one i don't know why no one else has thought of this uh and you also like are pulling real-time data from your client Correct, correct. So so there is deep API integration. And now, so first step, as you can understand, so personalization, I would try to say is journey, right?
01:06:48
Speaker
So first step was just like actually stop spamming people, right? Like actually have your systems all talk to each other and have a very seamless journey. ah oh Second step was...
01:07:01
Speaker
How do we make it intelligent, intelligent right? ah Okay, then based on certain behavior that this person has done, can we say that this person will prefer more of a WhatsApp versus call, etc.?
01:07:13
Speaker
And then based on the interaction I had, like each interaction gives me a data point, right? So based on that, I know more about the lead and i can I can essentially keep like then now personalizing it as well. like So that's step two.
01:07:26
Speaker
Step three, which we are entering right now and we've been doing little bit is... is hyper-personalization, right? Where the whole, basically, it's a slightly complicated concept, but like think this whole tree is actually an LLM, right?
01:07:41
Speaker
so the so but So rather than a static tree, your models are actually determining that, okay, you know, this is Akshay, this is where he is. Now when I'm talking to him, what should I tell him that he needs to hear so that so that you know it works?
01:07:56
Speaker
So I can actually, like i can what I mean is that if I do for access, if I'm doing 100,000 interactions a day, all of those interactions should be completely or highly different from each other right and that obviously can only happen when machine is doing the talking which we are in the phase now right where we are literally uh i'll give you a simple example like credit card also right like if i'm trying to sell you credit card or faxes And again, like in a legit way, you want to buy it, etc. Then but I'm not going to waste your time as an example, selling telling you, they you know, I'll give you 100 rupees cash back, right? And I'll probably me tell you... Accidental insurance may not matter to me. Maybe a lounge access matters to me.
01:08:39
Speaker
Exactly, exactly. Again, international lounge access is may be something that you care more about, that I need more, like, I need actually, like... 30 times a year or whatever right so so uh so so personalizing everything like the the script itself the voice itself uh the language of course uh like even things like what we've done is that even if i'm speaking english if someone is from the north let's say delhi so the caller will be delhi or the voice you'll hear will be delhi based ah so we used to do this matching earlier right and then versus you'll lead from bangalore let's say someone who is in karnatica will talk to you even the same process See, like just it's just basic human comfort, right? Like you feel better when you hear a voice which is more recogniz recognizable, who can pronounce Baner Gatta Road properly. Right. so so ah So now we've replicated the same thing in the world of AI.
01:09:31
Speaker
And we used to try some personalization, but it was still harder because see humans are a little bit harder for them to say, okay, every call start doing new new things. Yeah, you can't give a script for every call. They they need a script which they can repeat.
01:09:45
Speaker
Yeah. Correct, correct, correct, correct. Exactly.
Evolving Company Models
01:09:48
Speaker
and and ah And then, of course, not with AI, we are able to like change things in real time and then the whole interaction is completely new.
01:09:58
Speaker
you ah Okay, so 22 to 24, you had this period of agency plus gig workers. I'm just curious, ah did you start leaning more towards agencies? Because as you gave the example that from Uber model, you went to OYO model. Actually, the truth is even the Uber model today is ah fleet operators, like very few cars on their network are driver-owned cars. It's mostly fleets who are providing them bulk of cars.
01:10:28
Speaker
ah so So did you also see that happening, that more and more it's organized players you're working with instead of moms? Right, right, right. No, yeah. So I think, I mean, yes and no, I should say, right? I think as an org, we became, definitely we became little lazier where we were like, oh, this is kind of very comfortable, actually.
01:10:47
Speaker
You don't have to do... Yeah, there's somebody who has skin in the game to ensure quality, right? Exactly, exactly. and And at some level, you can, you know... ah I mean, a lot of problems. And the control seems very seductive, to be honest, for sure, right? I think that, oh, actually, someone like someone is sitting there and whatever. But why is that a bad thing?
01:11:10
Speaker
It seems like you're not very excited about that. No, no, I think it's a good thing and I think it works really well. I think why remote is still very, very exciting, I'll tell you, where we saw that, see, ultimately, the problem with contact centers especially like And that's why I really, um you know, i think I am very passionately against this thought process where every BFSI company wants a center in Bombay just because it's easier for them to travel. Right.
01:11:43
Speaker
But I'm like, here who cares in some level about me or you? Right. What we should care about is the end customer. And for them, the second stakeholder should should care about is that agent in some level right so see someone who's so i think it's a terrible idea in general i feel so right if you think about it's a broken model like someone who's in delhi or or ah bombay and this kind of salaries that they get and they they have of course these are like you know hungry uh ambitious people who are educated who've moved cities and and and then then they're working in this kind of job where unfortunately it's still
01:12:19
Speaker
not treated really well and and culturally it's not the most exciting thing and then and then you're also like traveling on average on ah a bombay like we used to have a center there also i remember like i spoke to people and they were traveling like five six hours a day man like dude this is crazy like this is just insane right so so i am a much bigger see end of the day it's like what people should be fundamentally happy right so i think So, I mean, my goal has been that how do we move a lot of this to tier two, tier three, tier cities, because we see so much more success with, I think this Bhopal, Indoor type places. At least people are just happy. Fundamentally, they're happy, right?
01:13:01
Speaker
And I think the second thing that I think why remote has been incredible is because I give you, so some of my partners are very passionate about this. I love it. Now when the market is changing, right, where and and So in fact, one of our enterprise partners said this a beautiful thing. He said, keep dude, I have a 21 year old guy basically selling average 40 to 50 year old that buy life insurance.
01:13:25
Speaker
You should buy life insurance because you're to die. What's going happen to your family? Yeah, I mean, they, i mean, what are the 21 year old boy with a lot of life? yeah Yeah, or like, how can they explain? Like, how can you, I mean, you're basically like abusing them, keep lucky you have to sell like this much every day. I mean, i mean what will they know, man?
01:13:43
Speaker
So, and I think, and I think this is a great broader change. We are also seeing that even enterprises, really, really i have started caring about customer experience more and more. Also because you have to understand an average Indian today, i think is as probably 28 or something, right?
01:13:58
Speaker
So so So it's crazy, man. I mean, i mean every Indian is sort of now, an average Indian is basically born in the world. Even if they were not, let's say, super financially affluent, they still are born in the world of ah internet and using Google, et cetera, from the beginning, right? so So like it's very unlike ah you know like our parents who would who felt happy to wait all day in a queue in a bank ah to to just put their money right so so this is not at all I mean like today people I mean I say to the to my customers that you know it takes a click to get a date they are not going to wait for five days to get your refund right so and this is your customer right so I think people are also understanding that so coming by so I think remote in my opinion
01:14:54
Speaker
Because how does our world change, I think, with AI doing more and more? And I think i think the way it changes is by actually being able to pay the humans who are doing the job much more money so you can actually ah make it a truly valuable career for them long term. Because is this is a transitionary career otherwise, right? Like people just want to do it two, three years to figure out what they want to do.
01:15:18
Speaker
ah Versus in remote, the beauty, I think, is that we are able to get segment that wants to just do this, right? And then you are actually getting, ah you know, really, really high quality people. a lot of these ah people are much more empathetic, e etc. So I think, so that, I think, just creation of that advisor type supply versus just like pushy people, reps, right?
01:15:42
Speaker
I think that is truly like something that is happening. So so I genuinely believe that the best solution, right? is going to be the combination of this Uber, OYO, and AI together. right I think if you can have a platform where all of these three exist together and they use the strengths of each other, I think that's where the magic really will happen.
01:16:03
Speaker
OK, interesting. What's your error now? Yeah, so I think that would be hard to disclose. Yeah, but like fairly. Last year's error. I mean, just for context, like in 2020, when COVID hit you were... Right, right. So, yeah, yeah. So, I mean, okay. So, little bit older, I can tell you. We were basically at roughly 200 crore annual revenue run rate.
01:16:27
Speaker
and But yeah, we've grown very well from there. Okay. and it's but We are a private company, man. So not... Double digit under 50 million era kind of. Yeah, yeah, yeah, yeah, yeah. Okay.
01:16:39
Speaker
Okay. Amazing. And this is all India revenue. It's all India revenue, man. And we, as I told you, we have like, you know, anywhere from when people are paying. ah So our average, if i have to say average ACV would be like $500,000. And even in the world of AI, of course, the ACVs are ah very decent.
01:17:01
Speaker
And I think we are seeing this crazy phase where... um um should we just get to now the AI side of things? like what have one One last question I want to ask. sure ah yeah So what's your margin profile like? You know, how much of your revenue is shared with the, like the gig workers or
Efficiency and AI Integration
01:17:22
Speaker
the agencies?
01:17:22
Speaker
Yeah. Yeah, so again, like it would be hard to share. Like just like double digit teens or ah like... ah No, it's that's very healthy. So what I can tell you is that even sort traditional BPO's have like really terrible, unfortunately, they have like, you know, like not even single digit actually. I mean, everything single digit were like really the early single digit type margin, right? and and that is uh but i think the great thing about us and i actually let me cancel this question a very ah good way see if you look at the unit economics of a bpo if at 100 right if they're making 100 rupees and you think that margin is so low you would assume that all of this is going and paying the people but that's not true actually so in that hundred thirty percent would go on paying the people who are doing the job right then literally almost fifty percent type
01:18:13
Speaker
would be just management players because every every 20 callers would need a trainer a qa a manager ah i think five roles basically for every 20 and then that organization of course becomes very bloated So for a teleperformance concentrics, like, for example, even when they are charging like a very good, like much more 50% to 100% higher than market, then also the margins are very, very thin because it's very hard to in that kind of ops heavy structure, it's very hard to make any margin.
01:18:49
Speaker
The superpower for us has been that we uh because of from the beginning we of a remote dna tech dna we have automated all the supervision from the beginning so that is really enabling us to uh to have almost uh uh like basically we've eliminated that whole layer right so we actually pay our people more um so than what they would get anywhere else and uh But we still have like very good margins compared to BPO because there's no supervision layer. Basically, there's no loss of margin, right?
01:19:24
Speaker
There's no ah rental. and like All of that is your margin opportunity. The the the rent, the supervision cost, the the management layer, all that is your margin opportunity.
01:19:35
Speaker
Fascinating. Okay. So now, you know ah I'm wondering, does... Does AI replace gig workers?
01:19:49
Speaker
Yeah, i mean I mean, the honest thing is that, I mean, if you look at... Actually, you know, I'll take a step back. i heard me tell you a little bit about what the AI thing is doing right now and what's happening, right? I think...
01:20:02
Speaker
I think what is very interesting, it's actually very, very interesting. And I've been, of course, building for a long time as you, as we spoke about, and and I have actually definitely never seen anything this exciting truly. in every way possible what i mean is i think the kind of excitement that we see in enterprises to adopt it the velocity with which things are moving ah you know like you can we are literally seeing I mean sales cycles shrink from 12 months to to two months ah one month also in certain places same organizations because I'll tell you why right I think you do a demo the product they are buying is a different product like you have a separate product of human telecallers and a separate product of AI callers
01:20:51
Speaker
No, so what we've done is we have to yeah, we do have two offerings that said the first offering is basically where we are doing a combination of AI and human together in one workflow.
01:21:03
Speaker
So what that means is the that insurance example we said, right? So the actual insurance sale today because of IRDA guidelines plus um plus the complex complexity of their interaction that needs that advisor, right? Who's sitting and highly trained on that company.
01:21:19
Speaker
Not just trained, they're actually certified by IIDA for that company to sell insurance. so ah so so So they, of course, still very much stay, but anything before pre-sales and post-sales, right? Like, which means qualifying the leads or after, say, like collecting your, ah you know, medical information type stuff, scheduling a medical test, getting payments, getting you onboarded, all of that AI can do really well.
01:21:44
Speaker
and And that, again, is a huge superpower as an offering because, again, we're globally unique in this way to be able to do it because what's also happening is a lot of enterprises are trying to adopt this and they have you know some AI vendor, would love and joke saying Bangalore have never seen a call center, right? and And then they have some call center who of course hates like tech because their billing is going down and the enterprise in the middle, right? so they So it's like they're all hating each other and the end customer basically getting abused because it's all one motivation. We are aligned on one CAC metric. So that becomes a very powerful. Offering that human and AI is in one place.
01:22:28
Speaker
The second thing we are seeing is AI native workflows as well, where AI is able to completely do the work. at very And that is where I think the... I'll tell you, I'll step up actually overall, on right? I think what ah what's happening with ai ah I feel that there is this tipping point or inflection point, right? Where when it truly blows up.
01:22:50
Speaker
And what I think the tipping point is when AI, for whatever category we think about, from first, it starts with co-pilot, which is all very nice, but you know it's not really...
01:23:02
Speaker
gonna change the world right then it starts with that okay it is it is doing now suddenly like few few percent like and you can see it will at some point do something where it can start doing a big chunk of the work as humans are doing and it's very critical that it has to deliver the same outcome See, no work, in my opinion, will get automated, which is going to be inferiorly done by AI in an outcome, right? Like, like for example, we are not ever going to write content from chat GPT if it is like terribly written, him right?
01:23:36
Speaker
so ah So I think that is where the inflection point for every category is that as soon as you cross the point where AI suddenly starts delivering same or better outcomes than human,
01:23:49
Speaker
Obviously, it happens at 5x, 10x lower cost because of the cost structure, efficiency structures and all of that. But I think even if it's way cheaper, etc.
AI's Role in Workforce Transformation
01:24:00
Speaker
But if it's going to do one lesser sale or it's going to be great just that like 5% of our customers are going to be so pissed off that you know They're going to abuse you all day on Twitter. right Those things don't work. These are enterprises especially.
01:24:13
Speaker
So I think that especially like how the crossing the chasm is, I think that inflection point in AI is getting hit ah and i And I feel very ah very, very proud and that I think we have really hit that inflection point in India for our category.
01:24:32
Speaker
We are anyway, from whatever I understand, and I think I'm fairly informed here that we are the largest scale company in this category and in terms of volume or brands or scale or everything, right?
01:24:45
Speaker
So, but I think we've been again sitting on this, experimenting with our clients and seeing the excitement of AI. But I think what has truly changed is i think a lot of the engineering stuff that we've done is also really great. but ah But yeah, but basically that's the inflection point, right? And I think there are a lot of, I'll give you examples.
01:25:03
Speaker
So we have actually cloned our a lot of our callers, right? And I think i think in the mark when you spoke to our team, I think you heard some of those recordings also, right? so and and and And so it's is's indistinguishable, right?
01:25:18
Speaker
You just cannot say what is AI, what is human across languages, across emotions. but And that is when it truly, I think, becomes very powerful. So I think that that has happened first.
01:25:31
Speaker
hu Second, now coming back to our discussion. So yeah, the offering is one is hybrid. The other one is AI native. And now... ah What is interesting is that because we've hit this inflection point, we are definitely seeing that AI is replacing a lot of the workflows fully. Right. So like and yeah and and it's happening industry by industry. Right.
01:25:56
Speaker
or Or complexity by complexity, you can say. So we like the first as an example, um e-commerce, marketplace, logistics is a very big industry for us. So we work with like a delivery to Zepto, so a bunch of these companies who are a lot of marketplace companies, e-commerce companies, etc. Right. So ah in all of them, because I think their use cases are this way or their DNA is also like low margin, e etc.
01:26:29
Speaker
We have seen that we've been able to actually automate all of the workflows fully, right? Because these are not very complicated interactions. But the extreme end of this is education and insurance, right? Where, as I told you, it you are looking at advisor. In education, we actually have our advisors do video calls, right? And demos and stuff. so so So I think it's that spectrum where I believe that, you know, and we've been seeing literally that every quarter, 10% of market has shifted to full AI native.
01:26:58
Speaker
Right. And we are actually sitting right now in the center where we are more of a 50-50, where 50% of the market or workflows can be fully native and the rest 50% may to be honest, I can say the next 30% there, I believe will happen. i mean, I of course have to be more optimistic and audacious. So I believe it will happen in the next three to six months.
01:27:22
Speaker
But let's say if we have a more pessimistic view, then it could happen in one to two years also. But yeah, so I believe that overall market, like forget squad stack, I think you will only need 20% humans for the workflows we are doing today.
01:27:38
Speaker
And so basically that's how I see the involvement. Low-end calling, like logistics and shipping and All of this complaint resolution kind of calling will be 80% replaced. Support. All this soft collection buckets.
01:27:54
Speaker
Like I'll give you very interesting. Collection interesting problem for contact centers because it's very basic calling, right? But even it's so complicated because, you know why? Because human emotions come in.
01:28:05
Speaker
Right. So so because see collection is sensitive thing, right? You're calling someone, boss, you have to pay me. What is your promise to pay? Which date will you pay? Now, why it is hard, you know, because you have to follow the ah RBI guideline, right? you You can't start abusing someone. Like you can't get emotional. kind of your Your job is not to, it's not your money anyway. You have to just get a promise to pay.
01:28:28
Speaker
So I'm sure you would have seen some controversy in the past. It has happened, right? But so as soon as AI does it, it's actually great because AI, of course, is not getting emotional here especially. And they are just like doing very objective. So I think something like collection happened the first. Then we saw in terms use cases, support again, like happened second. But I think now, especially with the tipping point, we saw sales and we were again more sales oriented. So we attacked the market. That's also a very interesting strategy, I tell you.
01:28:57
Speaker
Lot of people attacked and today, and it's it's really, again, I'm grateful to my team completely, I would say honestly, but but you know we have had players in India competing with us on this whole AI or even before Gen AI, etc.
01:29:13
Speaker
ah Doing it for four or five years and we have, you know oh we are we are much better, like 3x better than them now in in our POC outcomes, etc.
01:29:25
Speaker
I think it ah is very, very interesting also that I think it is again kudos to our team. But I think the strategy that worked for us is from a tech point of view, we never obsessed with cost. Right. What I mean is that that we first obsessed with solving the hardest problem. Like we were very clear that if you can cross this tipping point.
01:29:48
Speaker
then the market that will open up from niche of trying AI to actually like real applications where you're saying okay how can I completely like transform and automate and then so we solved a much harder problem and did not care too much about cost optimization in some level and I think obviously cost over time hopefully it should reduce and margins will improve but at least that opened up a much bigger market and now we are of course going downstream also like will not leave collections etc so that some startup can come kill us later on um but i think that's been a good good idea as well
01:30:24
Speaker
Versus I think companies who were more like, they couldn't come up, right? I think it was, it's still very hard. And now the market in fact is now saying for collection also, I want squad quality. Because I want it to be hyper realistic. Like I don't, why should i because if if if I can have a call in which you can't make out versus a slightly robotic call, I'd rather have the call where you don't make out, right?
01:30:44
Speaker
ah Why did you rebrand from Squadron to Squadstack? So, I think that actually happened when we went to the US. And I think everyone told us that squadron, people will pronounce it as R-O-N and it will have army connotations and some of
Building Effective AI Products
01:31:01
Speaker
those things. And then by the way, we did a lot. If someone searches for you, then they'll type the wrong spelling and...
01:31:06
Speaker
Exactly, exactly. And maybe there's negative conditions. And then then, so we did squad run, then we did squad platform and then, and then we were doing a few things. So we just said, okay, you know, squad stack is best because at our very fundamental core, we are more about decentralization and tech, right? So, ah and by the way, now are squad stack.ai. Okay. Okay.
01:31:28
Speaker
but so So yeah, yeah, because we, We realized that AI is not a feature for us, right? We realized that big first we thought it is, right? Because we said we are all the channels. So AI is just one of the channels.
01:31:40
Speaker
And we obviously are new tech, but then we realized that it's not something that is, you know, a feature. And I told you it's going to be 80% of this industry. ah So yeah, it's very interesting for sure.
01:31:54
Speaker
ah Is your margin profile different? so So you said currently 50% of the workflows are human, 50% of the workflows are agentic. ah Do you earn different margins on these two workflows?
01:32:08
Speaker
Yeah, we actually make, well, not today. So we are today, margin profile is similar, but this is just a minor optimization that, I mean, basically because either the elasticity of margin profile on the AI side is very high, right? So I know that I would know. What does that mean, the elasticity of margin profile?
01:32:26
Speaker
ah Right, right, right. So what I mean is that on the AI side, see, what is your cost? Your cost is actually the cost of your models, the cost of your servers, right? the cost of your inference etc right on the other side on the human side my cost is the cost that i'm paying this person the human right so i've optimized everything outside that in the human side anyway so i'm not giving out any money outside what i should pay meaningfully to that person so which we are very clear that we never want to ah so there's no in some way i can't reduce it right because in fact i want to actually figure out a way to pay them even more money right
01:33:03
Speaker
So, but on the AI side, ah thanks to you know Moore's law and and plus plus plus that's happening, right? I think the cost of inference, cost of servers, everything is going to, intelligence, everything is going to reduce significantly.
01:33:18
Speaker
Plus, even at our end, we will do a lot of optimizations with engineering. So it's easier to really. Yeah, got it. So, you know, there are a lot of, you LinkedIn influencers talking of how this is the time to build, you can build one-person business selling agentic workforce to replace the salary time. you Everyone talks of this salary time and how you can replace it by selling agentic workforce.
01:33:47
Speaker
ah Is it as easy as that, that you just get an API and ah create an agent and start selling it and saying that, ah why do you need a recruitment agency? i run ah I have agent tech recruiters who will manage recruiting for you.
01:34:03
Speaker
ah So is it as easy as that? or like what's the hard problem in building AI products and especially AI agent tech products? Right, right, right.
01:34:14
Speaker
No, it's a great question. See, I think the allure is very... um ah very good of that one 10 people like you know building an IPO business etc but I think it's like anything else ah I think ah I think we tend to over do some of these concepts I feel right I think ah And maybe I'll blame Twitter for that.
01:34:45
Speaker
But I think we over discuss these things. I think it's a great point, right? I think that what the point is simple that that your, anyone, any human actually is now 10x, right? And it absolutely, by the way, obviously, if we go back in history and 100 years back, go to someone and, you know, we show a smartphone also. So of course, we already 10x with a smartphone, right? Versus like who they were.
01:35:11
Speaker
And now with AI and then maybe Neuralink, like, et cetera, like, so we'll continuously become 10x, 10x, 10x, right? So I think, so I think that point is very well taken, right? And I think any,
01:35:22
Speaker
I would say any human who is not agentic today, but having a bunch of AI agents themselves, right, like is really missing out and is going to get disrupted. It's just like obvious. ah It almost means that, you know someone just doesn't adapt with time types, right?
01:35:37
Speaker
So I think that theme I am very excited by. But I think why do need to overthink this stuff? Like, who cares? See, if you are a you if you can build a billion dollar valuable company and IPO. Now there are 10 people or 100 people who really care. I mean, maybe 500 people in LinkedIn will be at excited, but no one really cares. I think you should really do what really matters for your market, for your customers, because who has the luxury? I would be very happy to talk to a CEO who has the luxury who is saying, yeah you know what?
01:36:07
Speaker
I'm going to solve this problem. Man, you never have it. It's not even in your top 100 list, right? You will solve all your customers' problems, or your market problems, your team's problems, or you will like say, okay, what's your team? I'll tell you where I'm coming from.
01:36:20
Speaker
um What's your moat? Can someone just go and build... what you have built using open. Right, right, right. No, no, so we'll discuss that separately. That's a very interesting question. But I think I i did understand. Sorry, but I was just like, oh, yeah, so over commenting on that point. yeah i think I think that's OK. Right. I think agent take, of course, makes things 10x. But I think i'll I'll tell you a very interesting thing.
01:36:48
Speaker
ah And actually, we can talk about this whole MIT 95% report, right? That 95% of the POCs are failing. ah And very interestingly, I can share this data that we actually are seeing 95% success rate ah in our POCs.
01:37:04
Speaker
and And what are we doing like differently in a way, right? And I'll answer this question, which you asked me, how are you implementing these agentic workflows, et cetera, right? I think the the first reason why this is really happening ah for us is first is that the tech fundamentally or the application actually I should say, right? That your core application that you're driving has to pass that inflection point, right? and And in a way, it was, i mean, we failed our internal POCs because we had a lot of human traffic and we were automating it, right? So we've been trying that for now one and a half years. So obviously, a bunch of things we did in the first six months that kept failing. So tech wasn't there, right? So that's totally fine. I think the tech needs to be there.
01:37:50
Speaker
The second thing is... and That I think it's very critical also to be very ah smart with your enterprise partners and and and design um these transformations to succeed, right? you cannot ah you can ah you cannot obviously ah implement something this large and not think about it properly. Like what would happen? How would the workflows change?
01:38:17
Speaker
What would success look like? What would failure look like? What would... um Like if something works in this metric, then would we scale it up? Do we have the budget? What will happen to the people doing this? Right. So so I think people need to think through it deeply. And I think and we also so this is my biggest struggle with my sales team as well all the time that That, you know, I think everyone's over enthusiastic at some level as well, which is obviously great.
01:38:43
Speaker
It's I mean, I would always have this. But I think it is the job of our team that we say, okay, okay, this sounds great. But let's answer these slightly harder questions to understand where we are trying to go and think about it a little bit deeply and plan this for for six months, 12 months across scenarios.
01:39:00
Speaker
yeah So I think the enterprise or the partnership needs to be very ready on that end. The third and the most important thing, which has been um our DNA is also driving the service and outcome for the enterprise. Right. Again, this going back to my same example. Nobody cares if it's AI or a monkey or an Elon Musk or some human.
01:39:20
Speaker
But if it's not impacting that metric, then you cannot do that. So you have to build around that. So I'll tell you like my biggest choke point in the company actually, Akshay, is I need more and more AI implementation and product managers who can actually write, like beautifully understand, work with an enterprise, keatchaa this is the problem. Okay, this is how I write the prompts to make my platform work for for one of our partners, right?
01:39:46
Speaker
So... So I think that is, again, very critical. And in fact, like ah the biggest skill set probably that the market should have. and And to be very honest, I struggled a little bit with my own team. Like people are were looking down at this role, right? My product engineering team was first like thinking, you know, what is this kind of role? And I love that, you know, our leaders were like, dude,
01:40:09
Speaker
all of us i are to writing prompts. So if you're not, if you don't, if you look down on this role, then, ah you know, this is ah being very short-sighted, right? So I think, so I definitely believe that I think to make these implementations mega successful, you need to have very good discovery and very good solutioning and implementation ah on top of having tech and great platformer.
01:40:36
Speaker
Is ah your ah proprietary data remote? Like you've been, you have access to these call recordings of like a couple of years worth of call recordings. Right.
01:40:50
Speaker
No, no, not couple. I mean, I can share that also. think we've done billion plus interactions, one billion plus interactions that we basically have the permission from our clients to use also, right?
01:41:06
Speaker
and uh and uh and yeah so it's it i mean it is absolutely a huge moat i would say right because i think it gives us uh it has it i mean it is i would say if you ask me the single or that i think one reason that our platform is the best right now is for sure we have an amazing team and they've written an incredible optimized code and thought a lot about english indian problems right like uh or even conversational problems like i'll tell you like we've done optimizations on like when you're saying then should i stop or not and when you speak in english then i respond in english if you speak in hindi i respond in hindi or if i don't so let's say if i if you speak something which was not very clear right
01:41:53
Speaker
Now there are multiple layers to this text. So let's say the speech to text will say, I understood this. It will send to our brain and the brain LLM will say, kar it will determine in real time, key was this 70% data enough for me to understand the context to actually respond meaningfully or should or not. If it says no, then it die I will say, so sorry, sorry, can you say again?
01:42:16
Speaker
Right. So, so it's literally like your own brain. so so So what I'm trying to say is lot of engineering, which, of course, i mean, indian and Indian teams will definitely, there will be some team that will, of course, have that level of engineering. But I think the second thing is all those billions of data we are sitting on. and And an actual, I think DNA of driving outcomes and, and the rest of the tech stack, right? Like the lead management is still a very critical thing for us.
01:42:48
Speaker
um And that is actually getting new 10X because now I can use that not just to do ah like that group level personalization, but actually like user level personalization, et cetera.
01:42:59
Speaker
So I would say these three things are kind of making us the moat. So yeah, and then long term, of course, it will get data and the application continues to be good. So it's literally like your own brain.
01:43:10
Speaker
so so So what I'm trying to say is lot of it. Then you finish it. Wherever there is a huge tectonic shift, i think I think, I mean, I do i do think about Moat all the time, especially at nights, but I think it is not very useful.
01:43:25
Speaker
From what I understand, I think it's very useful in this time to distribute like crazy, right? Like get get plugged in as many places as possible.
AI Talent and Company Culture
01:43:36
Speaker
and And at that time also don't over optimize. Don't think, here or why is my implementation also not, like why am I not one?
01:43:43
Speaker
one person, IPO or whatever, right? At that time, actually invest in really smart people who are doing these implementations, spending time with customers, sitting in their offices. In fact, I would tell you, ah um I think probably 30 of our team members stood as of today are sitting across like six cities in our customer offices right now, right? so So, why? Because when you do all of this, then you will truly understand what mode will look like, right? I think in this time of big change, you shouldn't overthink it.
01:44:14
Speaker
So you think there's a land grab phase. Every customer you get embedded into is unlikely to switch you out. Like once you're in, you're unlikely to get switched out. Yeah, exactly. Exactly. I think if you're a good team, then you'll figure out and you'll bid so many wedges where, and even the platform, right? Like with data, it keeps improving for our clients. Like if a client is using it today in a month, that is much more optimized. So,
01:44:40
Speaker
uh yeah so it's absolutely a land grab uh and i'll tell you on the other side right customers are looking to adapt like when we had a capacity issue our customers like dude we are gonna go ahead it's up to you are you are gonna be ready in a week or not ah so you you spoke of the struggle to hire uh what is the new talent that will succeed in this world but what you know what's the kind of so you need people who will um Like make your customer go live.
01:45:11
Speaker
What is that profile like? what do you What do you look for? How do you screen and decide, yes, this is the right guy? Right, right, right. No, great question. So at least for this specific ah role of AI, so we're calling it AI agent product manager, and we've gone through so many names actually, because we're trying to think what is the persona really needed, right? And and I think it's very similar to like that product, ah ah ah kind of a great product person, right? Which basically means we look for fundamentally three traits, right?
01:45:45
Speaker
ah First trait is high agency. I think it's very, very critical to be really high agency, ah willing to hustle, work hard with the clients and and people are also a lot of pressure to quickly move fast, et etc. right So I think that is that entrepreneurial high agency is very critical.
01:46:04
Speaker
And that trait, by the way, on a side note, what you said, in today's world is more relevant than ever. Right. I think you, I mean, just being very open to change and, and, and, and having that fundamental self-confidence here, you know, what, whatever happens, I will learn new tech and I'll figure it out and I'll keep moving on.
01:46:22
Speaker
uh so yeah so first is high agency second is uh customer communication i think it is it is uh you just have to be good in today's world to be able to talk to different stakeholders understand their challenges to detail discovery solutioning uh so uh so that's again a very critical kind of skill and the third is actually a technical skill of like how good are you in uh in actually writing the solution is similar to coding like prompting like how good are your prompt skills uh to be able to uh like i mean all of us prompt actually when we use chat gpt or anything then we write those prompts right so you can write it like way better almost like how you code so yeah so these are the three fundamental skills we look for uh traits we look for and is there a pattern on uh
01:47:15
Speaker
such people, like people who are working in such roles have good prompt skills? ah That's exactly what I'm trying to figure out, I think, Akshay. And we are actually, interestingly, you know, and I am not being restrictive, right? So we have like folks. So see, I think the the first is people who've actually been which is very small pool is people who've been actually prompting and designing, let's say, chatbots things like that.
01:47:41
Speaker
ah The second is... And these are typically computer science graduates who... Yeah, yeah, yeah, yeah. But that's in a very small percentage. But I'll tell you, the second... so the so ah I think a very big segment is engineering, right? So people who've been just engineers, they're getting very excited by it because they are...
01:48:03
Speaker
ah they are So we always had this product engineering concept that you are a product engineer, which means it doesn't, well it's not about just writing code, it's actually creating impact, right? I think that DNA ah has led to engineers also love this role because they are like, wow, I'm writing, actually working and meeting and working with the CXOs of some other companies that, you know, see billboards of every day. So i think I think driving that large scale implementation, they are also very kicked in the kind of optimization they do directly.
01:48:35
Speaker
Third is of course product. I think people who have been great product talent, I think it's a great role for them. So this is, I think these two are very obvious ones, but what I'm also very keen ah to see how a lot of the non-obvious roles will get into this. So like i'll give you an example, like we have lot of these people are trainers, right? So people who were training humans on how to like do better conversation and they come with linguistics backgrounds and ah you know, it's very, very interesting persona, right? And we actually are seeing many of them become really good.
01:49:12
Speaker
because they they are very creative and they also understand ah they understand like what needs to be done and they have probably have the best context. They have heard millions of recordings. They coached real people. So people who have been coaching people on this skill, they can probably coach AI really well.
01:49:31
Speaker
ah I think that marketing also we are seeing very interesting that we realized that marketing folks are very creative by definition and they... They actually were really good at this also because ah ah because it ultimately requires creativity to write these prompts.
01:49:47
Speaker
so So, yeah, so I think very interesting and it's completely open, right? I think companies like us are absolutely, we don't care about the background, nothing actually, ah because these are very simple, fundamental. And in fact, no one actually had a lot of experience on this already. So we are anyway looking for smart talent. Yeah.
01:50:06
Speaker
How do you screen for this? like Like you give them a scenario and ask them to write prompts? Yeah. So first, the first round is just technical round where they have to write prompts for, which means see anyone who has, ah to be honest, I would say and anyone who's high agency and alive has, I'm sure, written enough prompts, right? Like because they've tried out least all of these GenAI tools. And the way it's like communication to GenAI is by communicating through prompts, right?
01:50:34
Speaker
So, ah so, so that's the first thing then we do. ah And before that, of course, we look at their profiles to understand like evidence of agency. Right. So I'll give you an example.
01:50:47
Speaker
If someone, any callist, like, so we try to draw, I try to teach, I try to teach when we draw a career. ah growth curve versus time, right? and When you see a CV.
01:50:59
Speaker
So curve is a very interesting thing, right? and And look for like something which is exciting. So like, for example, we don't care if you were a topper or if you were a five pointer. But if you were a five pointer, I hope you were at least like managing your fests, right?
01:51:16
Speaker
If you're not doing these things, then maybe you ran a startup, right? But you have to do something, man, right? Or you could have been just a great top. But if you don't, if you did not do anything, then that's not very exciting for us, right? I think, so I think we look for agency in these things, right? Okay, then you're working in a company.
01:51:34
Speaker
okay maybe it's not the most exciting company in the world but on the side you are actually like running a great podcast right so that's amazing so so the point is to figure out that uh and andll and it's a hard lesson man uh but i'll tell you in talent hiring what i've learned is like every other truth in the world, is that history is truly the study of future, right?
01:51:56
Speaker
It is... ah so so so So if someone has not had all the evidence, and maybe they will be, but but but I don't think it makes sense to bet on them to be high agency and see now if they've never done it. How do you retain high agency people?
01:52:13
Speaker
Because a lot of high agency folks are somewhat allergic to regular corporate jobs. Right, right, right. But not having a regular corporate job, man, the short answer. But, but no, I think we actually uniquely, we're probably the most unique in this way, where I said we are roughly 250 people team right now. and And, and 19% of our team has been full time founders, right? So after college, so people have started up and ran companies full time. And
01:52:45
Speaker
And these people have touched wood as something that in our culture that I think they love. And then they have stayed with us ah for years. And in fact, you know, many of them are leaders now in the company.
01:53:00
Speaker
So... so what What is it in your culture? Like, that that's what I want to get out from you. Like a couple of pointers. Yeah, I think... ah I think ah one is...
01:53:12
Speaker
ah um Let me take a minute to think about it because it's very interesting question.
01:53:26
Speaker
Basically, if you were to interview them and ask them, why did you stick on? Would there be a pattern? Right, right, right. Yeah, there would be a pattern. I think one piece is ah that I think we ah ah we expect and we allow for a lot of ownership and i do
01:53:54
Speaker
yeah hu
01:53:58
Speaker
um So I was saying, a shakiah so I think people staying in the company, right? So I think one thing is that um we expect a lot of ownership from folks, like in whatever they do. And we don't, we don't like to micromanage them.
01:54:14
Speaker
And I think, I think that, so I think giving them a lot of accountability da is something I think that works really well. I think the second thing, which is,
01:54:28
Speaker
Really something that I have gone on to accept over time is that also I think leading with love. I'll just take a minute to explain that. I think if you think about it, there are various emotions that you can leverage to build a large org, right? So
01:54:46
Speaker
I think ah and think love is a very underrated emotion. I have never heard that ah from a founder before about leading with love. ah What's a...
01:54:58
Speaker
How does it play out? Like, give me an example to help me understand. I'll give you an example, right? Like for your mom, you would jump out of the building right now, right? That is, that, I mean, what, tell me what other emotion for your, if you're afraid of your boss, are going to jump out building? No. You're to probably push him from the building. but So, so that's what I mean. I think, I think, I think love is such a strong emotion that I think if people could figure out a way to channelize in the right way, ah like deeply caring about people.
01:55:32
Speaker
Sorry, just give me one minute. Yeah, yeah,
01:55:51
Speaker
Yeah. So I think,
01:55:55
Speaker
ah Yeah, so should we do the whole thing properly again or? Yeah, yeah, yeah. So just start from the answer like what you know to help me understand what that means.
01:56:08
Speaker
ah Right. So... So, yeah, I would say that I think what ah the high sorry sorry the entrepreneurial thing and all that is okay. Now, leading with love, that we can continue. Just leading with love. That mom example you gave, you can start with that.
01:56:22
Speaker
Sure, sure, sure. Yeah, yeah. So, I think leading ah with love. I think love is a very underrated emotion, I feel. And I think, you know, you think if, I mean, it makes you soft probably is what first that people will think about, right? But I think...
01:56:40
Speaker
ah you know, like, for example, you will jump out of the building right now, if you know you have, ah if it's for your mom, right? So I think that is what, and no other emotion can make you do that ever, right? So I think love is such a strong emotion, which is, which is not really leveraged well, I think, right? ah So, and and it cannot, and the beauty of love is it has to be pure, right? It cannot be fake like I cannot it can exactly like people see through it right so I think when you truly deeply care about everyone around you and and I think they start caring about ah everyone else and I think like
01:57:23
Speaker
when you build a culture around that, uh, then you are, and you're, of course you're, then there's a lot of passion. There's a lot of love towards your mission. There's a lot of love towards the work you do. There's love towards your the impact you're creating for your clients.
01:57:37
Speaker
Uh, you know, and you're absolutely like, still you're very audacious. You're very hardworking, very, uh, ambitious. We have, we strongly always say stolen from Amazon that, you know, we work hard, we work wrong and we work smart.
01:57:51
Speaker
Like there's no, one out of the three that you can do right so so all of that man but but i think if you encapsulate all of that in a culture which is not afraid to put a lot of love smileys on each other's messages and and and truly deeply care right and not afraid to show that i think i think it's a real superpower and i think that is what uh definitely makes a lot of folks with high agency really ambitious. i mean, most of the team is much, much more smarter than Vikas, Kanika and I, who are the founders. Right. And I think I think the reason they continue to build with us and dream with us is is because of that fundamental emotion.
01:58:40
Speaker
And it's like a sports team, right? I'm actually very inspired by sports, ah especially sports coaching, even more than watching sports, right? ah To understand how to be a great leader.
01:58:52
Speaker
And if you see, ah you know, like how coaches have... ah So there's this term called Ubuntu, which is an African philosophy, actually, which Nelson Multela used when the...
01:59:07
Speaker
when the South African independence happened, right? And I know if you go if you know a little bit, but basically what happened was that the independence that happened, obviously, the the the the half the pop i mean the majority of the population hated the the whites and then you know and if they would have all fought everyone would have died so so so so that way mandela of course is one of the strongest leaders in the world uh forever right and i think he had the foresight to say you know what this cannot happen and we can't like kill each other we'll kill the country and they used this philosophy of ubuntu which is a which is a
01:59:49
Speaker
um which is an African philosophy where they say that my valuation, my value depends on the value of the people around me, right? So this means that I don't have to, you know, be like, oh, you know, Akshay is like doing this like really great podcast, he'll get better than me. Like, I never feel like that, right? Because I know that my value depends on actually your value. And if you think about it,
02:00:14
Speaker
uh, philosophically, right? I think different roles you can do ah across times, right? So, so imagine if you were in the 1700s and I, I always believe that a very exciting thing to do would have been to be like Columbus and say, that's how I think of startup, right? I think that, that you say, okay, of let's, let's, i mean, this whole ocean thing, let's, let's go out there and see what we can find. And maybe we can find some other continents, et cetera, or find better parts to pass tools these places.
02:00:44
Speaker
so So I think when you ah ah ah when you ah see all of these like different different roles that have evolved, i think like building ah and think we have the best structure as a startup. What I mean is that because now we have ESOPs and things like that. right So it's not just theoretical concept.
02:01:04
Speaker
right There's actually a logical reason ah in which that truly my valuation is increasing as your valuation is increasing. So So like if you are a leader today who is, uh, let's say more obsessed with a certain problem, then, then if you do really well, then, you know, I win also. Right. So I think, so I think that kind of, uh, um, uh, Ubuntu as a concept, I'm trying to really like get everyone in my team to adopt to it.
02:01:36
Speaker
And, uh, and, uh, And yeah, so i think I think that's what. So I think there are so many like different concepts around within love, I think, that can really be leveraged.
02:01:48
Speaker
And then you can create something like really magical, right? I think when you really enjoy every day when you come. And and by the way, when I say all this, like we fight. like You should see us fight, right? I think.
02:02:01
Speaker
So so we we we fight exactly like, by the way, how i used to fight with my uh with all the kids i used to play with in the playground right uh so you fight as passionately as that and and it's not like you know uh any of us can pull off some shit and we're not going to be called out for it So it's absolutely very much there, right? But it is like the same thing, I think. I think the emotion of sports, like why are you here?
02:02:29
Speaker
but and and And that's my life philosophy also, by the way. I think like I'm here to play, right? I mean, and if you, again, like philosophy, you think about step up. I think it's a dream job to be a founder because, i mean, you basically have this weird philosophy of how the world should look like and you know some crazy idea and and then And then you are able to get bunch of other people to join you in that expedition. and And then maybe some i mean maybe if you are lucky enough, then you know people will give you capital also and crores of rupees. OK, go play.
02:03:02
Speaker
go You're like Columbus. Yeah, exactly. And then you're playing also, right? And actually, you're not like the numbers. I mean, let's poor guy, they were like, i mean, we're going to die, right? If they take a step bad move. And we we we keep saying we'll die, but physically, nothing will happen. I mean, we'll go on our next expedition tomorrow. So I think it's such a beautiful... ah That's what I was telling you, man. I'm i'm quite a nerdy purist with this whole startup thing ah that I just feel like it's just so cool, right? I think you can...
02:03:32
Speaker
ah I mean, it's it's just play. Really, i think there's no better way to say that. Fascinating. Thank you for your time, Apoor. It was a real fun conversation.
02:03:45
Speaker
Likewise. Thank you so much, Akshay. And I love the depth of your questions as well, man. I think it's, I mean, I can't imagine how someone can be so well-read to ask. I mean, you how you literally understood cloud telephony and all that. Okay.