How to pull off the ‘zoom-in pivot,’ and other lessons in early-stage building with Luminai’s Kesava Kirupa Dinakaran

 thanks for having me, Todd. Excited to be here.

 So can you tell us a little bit, about your personal background and some of the different pieces that pushed you towards the founder path?

 Absolutely. So I grew up in the southern part of, India, and most of my sort of, generations of family, grew up on a coconut farm. For a long time, the sort of exposure or the kinds of things, that people within my family or within my community did, was very different to the world I'm living in today.

And so when my parents actually moved up to Chennai, which is the capital of, the state I'm from back in India. I stumbled upon Rubik Cubes and that was my first, interaction with, something that was maybe very unfamiliar to the community I'm from. And, it'll make sense when we go dig deeper, but, I'd walked into my, first ever sort of Rubik's cube competition cuz I had learned how to solve it from a friend of mine.

And, this was a competition that was, in one of the, I, its, in India and it's a, it's one of the sort of national technological universities, and. When I walked in, I walked into this room fiddled with CEOs and musicians and artists and engineers and doctors, and was like, whoa, I guess I can do like other

 I was, 11 years old, so just I was like seventh grade, that was the first time I was like, okay. So all these things that I'd learned, I was like, that's like maybe 0.0001% of what the world had to offer. And when I saw these people I was like, oh my God, these are people who are pursuing things that I've only seen in like movies.

And eventually got very into solving Rubik's cubes as I got closer and closer to the community. And that was my sort of journey into, at least learning something about tech and entrepreneurship.

 And then what was it that initially brought you to the us? When was that?

 Yeah. Basically Rubiks Cubes kind of the Indian government and a bunch of other sort of communities had funded me, or sponsored, some of my trips, abroad. So I'd gotten to go to the, world Rubiks Cube championships and, and I'd, I got the chance to captain the Indian National team.

And so got to travel across, some of these countries and, ended up going to this sort of high school called United World College. And it's this high school that brings together, people from 70 different countries to work on international peace and understanding. And I got into this, full scholarship to go there and it was not because I was good at schools because of this Ruby Soup stuff and, this whole,

what was interesting about the Rubik's Cube was what was interesting about UWC was the whole mission of the school was to promote peace. And, so I'd, decided to, right after I graduated from high school, I'd taken a gap year, because I didn't end up getting into any of these American colleges that I applied to.

So I wanted to live this United World College, mission of peace. And so ended up cycling from Europe, to Asia to promote this mission. And there was this foundation in New York City that sort of cut wind of it and then said, we'd love to fly you out for this like 10-day social entrepreneurship program.

And that was kinda the first time I ever came to the US back in March of

 Wow. Okay. And so when you came to the US in 2019, like my understanding is that initially you were living in a hacker house and getting by and living off of hacker money, and hackathon winnings. What can you tell us about that story?

 so when I first came to the US right? I'd come for this program, and I knew I was interested in tech. I planned my sort of return flight from San Francisco. And so I'd come out, to, to the, to Silicon Valley. And when I came here I was like, oh my God, these are like my people.

And I started to feel like it, it truly felt like home in so many ways cuz, people would be thinking and dreaming about the most wildest and crazy and impossible things. And that energy really struck with me because that was the energy that I was searching for.

And so I decided to stay and not go back. But at that point, I just, I'd had a Taurus Visa when I first came out. And so I had to figure out a way to, obviously I couldn't work here. And hackathons were, a very effective way to continue to legally be here, but like still win prize money to sur to survive essentially.

And, these hackathons were like two day events, for, to, to up to $10,000 in prize money. So you could win a weekend and live a coup, like at least like a month or two on one hackathon winning essentially. And so that was the first, snapshot into some of the Silicon Valley stuff.

 All. So tell me about some of those hackathons. What, where were they, what were the rules? What were you building?

 Yep. Yeah. Most of these hackathons, me and my co-founder, my sort of the co-founder started a company, the company with, we'd initially picked very strategically the ones that we wanted to go to. There were multiple of these hack multiple types of these hackathons, right? They're all the way from Europe, like sort of university hackathons to, to these large corporate hackathons.

And we focused on our kind of end goal was to survive, right? Here in the us. And so it was like, let's just go to the hackathons that has the most price, money and the ones that had the most price, money war, these sort of corporate hackathons. These were hackathons organized by companies like Twitter and Twilio and, shell, and they all had their sort of own versions of a hackathon that they ran one year that, you know, they tried to promote and run around a specific theme.

And so we focused very specifically on these type of corporate hackathons when we're building products, that were very relevant to the companies that ended up organizing

 And if you were relying on Hackathon prize money to support yourself, you probably got pretty good at winning these hackathons or you were able to do it reliably somehow. How did you do that?

 Yeah, I think we basically won almost every hackathon that, we did. So the only one we actually did not win was like, TechCrunch. And the, I feel like the reason why we didn't win TechCrunch was because all the rules that applied to these corporate hackathons did not apply to TechCrunch. Cuz in TechCrunch, they didn't care about some of these other strategies we used in some of these corporate hackathons where we built products for the companies themselves.

But in TechCrunch thousands of people participating on building the best product for that weekend.

 But it sounds like you had a formula for that, for these corporate hackathons.

 Yeah. What's really interesting is there's almost like an analogy that I can, you know, drive between how we ran in these hackathons and how we won them to, how we built the first, how we built at Lumina for the first year or two years. And, the thing about these hackathons, they are generally.

Advertised as this is an excellent place to learn and you should just come build whatever you like. But corporations who are that size aren't just doing hackathons for the sake of charity. And what was really interesting is, what we discovered was as we started to talk to some of these like hackathon organizers, there were two reasons why they were organizing the hackathon.

It was like one was either they wanted to recruit you into their companies. Or two is they'd want you to test, something that they'd launched recently or, an API that they'd put out or some version of that. And so what we do is just before the hackathon, the day of the hackathon, let's say the hackathon started on a Saturday morning, we'd call up the organizers on Friday afternoon, and Sweet Dock, our way to figuring out, why they were organizing the hackathon.

It was fairly easy to understand if, like they were organizing it for the sake of recruiting or for the, for us to test, their APIs. And so then when the or actual hackathon started, in the first three hours we'd ideate and try to think of something that was very specifically for that goal, whether if it was recruiting related, then we'd basically study all of their, engineering job descriptions, all the different types of, the stack that they were using and what they were looking for.

Or if it was something related to the API or some, something they've launched, then we'd build a product around that, being the core, of what we were, trying to do for the weekend. And in the first three hours, once we did that, This was the key piece of the puzzle, which was, we'd go and pick this idea back to the organizer, or the organizers, right?

And we'd get their feedback and then we'd come back and we'd build a little and then go back to them. And we did this over and over again in that sort of 48 hour time period while everybody else was just coding away, right? And what ends up happening in most of these hackathons basically is these organizers who are just serving you food and just like hanging out there, end up being the judges of the hackathon.

So when you'd go and actually pitch, you would've built exactly what they wanted to see, and basically almost had a hundred percent hit rate. Which makes sense why it didn't work in TechCrunch. Cuz in TechCrunch the organizers aren't actually the judges and they don't really care, what it is that

 That is brilliant. So you basically reverse engineered how to understand what the judges were looking for and then built that iteratively over a three day hackathon and then we're winning them fairly reliably. That's amazing. So then how did you then get from that to Lumina or, I think at the time you had initially called the product digital brain.

Did that come out of a hackathon or what was the original concept for the product?

 The first initial version, started at, this hackathon organized by this large German public company called, optimal Systems. Maybe they had changed, they called it the Uvis Hackathon, which was this, specific, cloud API that they'd released.

And they had a bunch of different tools around, OCR and, some other, APIs related to some of the stuff that we're building on today. And we'd initially started to talk to the main organizer at this hackathon, and they were talking about their want and focus, to build much better customer experiences, not just for their users, but helping, their customers give it to their end users, right?

And so that was the first understanding, even remotely what this problem space of sort of customer service and customer experience was. And that was the hackathon where the initial version of back then called, digital Brain

 after that first hackathon, what did the product look like? What did the first version look like?

 Yeah. So fundamentally what we realized was that customer service teams, or customer service folks, Were the first ever interaction that an end user would have. once someone bought a product, right? Or even before buying a product. And they were the brand or the face of an organization.

But what was really interesting was, I reflected on my own experiences with sort of customer service, and me and my co-founder, we deflected on our own experiences with customer service. And what we discovered was that, in the end of the day, all we care about, was how quickly can we get an answer or get some.

Issue resolved. And so what ended up happening was when we dug deeper, we also lived in this hacker house and one of these companies that, had raised a massive growth round, grew from one support agent to sort of 50 support agents over two quarters. And when that sort of hackathon had just happened, and when this friend of ours had built this company that, had to basically 50 x their headcount, we were like, okay, there's something, there's some weird like correlation between the linear scaling of teams to growth and that didn't feel

and so when we went and spoke to this founder, he was like, Hey, you should just come, sit in our support org and see what's going on. And so when we went and just sat in this, company's support org, what we discovered was that. Support agents were, they had four monitors. It looked like they were traders, like quantitative traders or something.

And they were jumping between these different systems to one, gather information, but two, perform the actions associated with these tickets that were coming in. And what was interesting was on, 20% of, or 30% of tickets, they spent 80% of their time. And that 80% of their time was actually actions and workflows that they did basically the same way every single time.

And it was like six to eight workflows that were somewhat bespoke to each and every company. And it felt If we helped accelerate that or turbocharge that part, then one, we're impacting that end user, experience. But two, we're helping the agents on many different other levels, beyond just making them better at their own job, but, reducing them from burnout and making them much more, less prone to even like mistakes that they're doing, on a day-to-day basis.

 And so I imagine though, you, there were so many hackathons that you participated in, and this digital brain was one idea from one hackathon. How did that, how did you know that was the idea that you wanted to build a company around? It sounds like you connected that with the customer service observation you were doing, or how did that become the idea for you?

 it's a great question. So what ended up happening was, this was probably the first time where it started to feel like. What we were building, could actually apply in the real world. And what I mean by that is the organizer of the hackathon that we built the product for, came back to us after one week.

So they'd already given us the prize money. They said, congrats, you win this hackathon. And then one week later they'd come back and the organizer emailed me and was like, Hey, our sort of customer experience organization wants to talk to you. And I was like, this is odd. Maybe they're just doing it cuz they want to interview us for something or maybe it's just an extension of this hackathon.

This is new, we'll take a call. So we take this phone call and this org is Hey, we actually want to deploy this internally. Is there any way you could license this technology to us? And that was that moment, from between vague hearing of the hacker house friend who was like, oh my God, I hate my support org to building this for, optimal and then, for optimal systems and then to the company being like, Hey, we're actually interested in, in using this internally.

That thread is what sort of, put us on this like idea ma of okay, maybe there's something here and we should go in and dig deeper.

 Wow. Okay. And so this was like end of 2019 kind of timeframe? Early 2020.

 That's right. Yeah. It's like September, October, November 20th,

 Okay. And so then what was the first thing you did getting started in 2020, in those first few months?

 Yeah, the moment we heard that, we were like, okay, great. Let's just figure out what we need to actually do to, license this. And then it involved a small set of, logistical tasks, that as immigrants, we had a little more, things to do, on at least the, incorporation side.

Getting a bank account set up and figuring out what a contract looks like, and, making sure that we're not like, dropping the ball anywhere.

 so you actually signed like a paying contract with them? They were. They were effectively your first customer.

And that was the first, version where we started to itrate with them towards the first version of what, Lumina ended up looking

 And then I believe that you applied to join Y Combinator in summer of 2020. So what led

 That's right. Yeah, I think as sort of young founders with no Silicon Valley network, the things that, you know, initially, even brought us to startups or helped us understand what company building was, was why C'S posts and pgs, essays.

And it was a dream in many ways, to get the chance to participate. And so we'd applied and, we'd actually, applied a second time. So we'd applied when we were doing those hackathons. And we actually had just done the hackathon with, with optimal systems. But, when we'd done the, we'd applied again, with a much more crisp, clear, and one sort of customer who was willing to engage with us.

 So what was in the application? What were you, what was the thing you were pitching at that time? How were you describing it?

 Yeah, we called it, superhuman for customer support, superhuman, which is still the email client I still use, first round company. We, we basically realized we were going into this age of, design first software. In many ways because like when we talk to sort of customer support agents, There were mostly people who are our age, meaning folks who are, under 30 in most cases, if not younger, and they were starting their career. But for people who grew up in the advent of Instagram and, and Facebook and tools that, were consumer like software. Customer service was the opposite of that. It was dominated by tools that were predominantly born much, much earlier than, where the support agents sort of demographics were.

And so we ended up kind of building the first version, which looked similar basically to how Superhuman was, but for the customer service team. So it was just a layer on top of, at that time, Zendesk, where it transformed the workspace, into something that was a lot more, quicker, much more usable and super fast and had a bunch of other features that, that at least made their, we thought made their job a lot easier.

 Okay. And so during yc, you're iterating on this product, building this product, getting some very early customers. What were some of the things that you learned in YC that were most helpful to you?

 it's really a cliche, but when people talk about, the only things that matter, actually talking to customers and building products, the amount of help that actually, or the amount of justice, that it does by hearing it every single day over and over again, it deeply implants, that philosophy.

I remember like even a 24 hour period or a 36 hour period where you went into just build mode. I started to feel the sense of, hey, Am I putting myself in the shoe of this end user who is actually, going to see and use, what it is that we're building for them?

And I am not that person. I am not that person who is going to use this product. So are you thinking of yourself as someone who's just graduated, college, taking on their first job and has no idea about, the company, they're just starting to work for, are you imagining yourself as that person while building this?

So it's like small little titbits like that sort of added up, into us, like being able to get that first, initial, prototype or v p out and that was something that I think was drilled in, a lot more deeply, by YC

 Are there any specific things that you did or like tactics that you could share about how to put yourself in the shoes of the customer so deeply?

 Yeah, I think a lot of this is going to come down to becoming friends with your users. And I think it's becoming a lot more common where. Ycs advice was always like, give your phone number, like right off the bat. If when you onboard someone or when you're even talking to them, because when they're thinking about something that's even remotely related to you, you wanna make the friction, between them and you as small as possible.

And so in our case, it was like, how do we, how do we get them, to feel like we were their peer and that if we were sitting right next to them, they'd talk to us the same exact way, compared to if we got on a Zoom call, or whatever. And so that was the first start of just almost understanding them as a whole human being.

Like why did they pick the job? What was their sort of motivation? What are their kind of like long-term aspirations for what they wanna do? And in a lot of cases, It might really surprise you. A lot of these folks, like not everybody is like looking for Hey, I'm, I wanna, I want to become the CEO of this company.

I'm starting to work out as a support agent. Like a lot of them, wanted to just be a yoga teacher and support was a way to, live life while they pursue other passions. And so it's it's almost knowing where their intrinsic motivations are to help eventually, design and build what we ended up building for

 so how many were you, were they like texting you and calling you when you gave 'em your phone number? And how many folks were you working

 wasn't as often. it wasn't as often as I would've liked, I think. But at least I think we at least got one or two interactions per week from the sort of first four to five customers that we'd initially signed. when we were doing

 Okay, so the first version was this superhuman for customer support. What was the reaction to that product when you put it in customer's hands.

 the easiest features were just things like keyboard shortcuts and getting as much data in about the end customer they were talking to right in front of them. And giving them a very easy and accessible kind of interface to interact with that felt a lot more natural than an old school Zendesk interface or a ticketing system.

And so when we put that in front of them, I think in a lot of cases we saw in incredible sort of delight or want to use it, right? But that was at least our understanding of when we put it in front of them.

when we saw their faces light up it made us feel like, oh my God, like this is it. That was the first reactions that made us

 And what was the thing that made their faces light up? That it was just so much faster, easier, better designed? What was the thing that really was the aha.

 so this is what led us to do this, like zoom in, pivot. Maybe I'm like jumping the gun here, but basically when we gave the product their, eyes lit up and all of these things happened, but it wasn't until I remember this very specific user we were working with this company that was doing ride sharing fora sort of bike and scooter ride sharing company that was local based in LA and one of the sort of support managers.

I remember her very specifically, looking at one thing that was part of the product that made her go oh my God take my money. And it was specifically this aspect around automating away some of their repetitive workflows. And in this case it was a very simple workflow that we'd integrated where they could click refund on a subscription that someone you know had if someone requested a refund on a subscription, then you could do that refund through back then Digital Brain in, in just one click.

And when I heard that reaction from this one customer, we recorded all of our calls until that point. We ended up going back to actually every single call and notice when their, when they had that moment of aha. And what was interesting was they weren't impressed by the keyboard shortcuts.

They weren't impressed by the sort of how fast the interface was. They weren't impressed by, how pretty the product looked. What they were actually impressed by was the things that made them feel incredibly tired and bored and very manual and very cognitively heavy when we shored one use case of that being taken away from their day-to-day, that's when their eyes, would light up and you'd see their eyes become bigger and you'd hear their like, voice become different.

And that's when it started to feel like, okay, like we've seen this now with six or seven customers. What if we like just stripped everything else out and just focused on this like one specific use case. And and do real research to figure out if this is

 And just to be clear, it's because the act of issuing a refund prior to digital brain involved go, I assume going to different systems and clicking different buttons and different applications and wherever the orders are, wherever the payments happen. Is that why it took so, it was such a better experience.

 Yeah, so here's just an example. So we work with a customer. We work with a customer called ordering. It's like a sleep ring company. They have hundreds of sort of support folks and, to process a refund, traditionally. And this is not just for ordering. This is actually a majority of retail customers, if you look at them, they, what they're doing basically is, once a customer writes in, they're copying their email, they're going to an order management system that's homegrown.

Generally, homegrown admin systems are like super slow to load and get started. They're going in, and they're searching for this user. They're finding that specific order. They're clicking cancel on that. Then they're figuring out how much to refund or they're figuring out the amount that needs to be refunded because they're they can't give refunds on shipping and they can't they have to figure out what the sales tax was based on Which specific state this person was from, or which country they're from.

So doing, they're doing all the math and then going to a payment processor and then if it's like a much more clunky payment processor, then they're like navigating that own system and then doing the same set of actions basically to do the refund. Then they're going to a shipping provider if you go have to work with, Canada Mail or u p s or something, to even navigate that interface is a job of its own.

And then you're creating this shipping label. You're downloading the shipping label, you're going back to the end user on Zendesk and then attaching it and saying, Hey, your refund has been processed. Can you make sure to attach the shipping label and Send the product back to us?

So when you're dealing with these like five different systems and like 40, 50 different clicks to process one refund. That's incredibly tiring and a huge weight on the sort of support agent where they've signed up to talk to customers, but 50 or 60% of their time is spent on doing things like that are like this

 So digital brain was able to get that all down to one click. How did that work?

 Yeah. So this is the magic of what we offer, even today. So even about two and a half, three years in this is what our focus area is. And. We were figuring out like, what are the ways we can solve this problem of unifying these multitude of systems.

So we discovered, many different approaches to these problems. And the one that was the most exciting, or the one that, blew my mind is this whole world called robotic process automation and rpa. That's right. And RPA has been around for a long time UiPath and automation area.

These are all like public companies valued in the, billions of dollars in some cases, tens of billions of dollars very large businesses. But they've not necessarily doubled down or focused on sort of customer service or contact center. And the concept of sort of RPA is you're able to record what it is that you do and.

646632a1ea021acd36d2995d

Kesava Kirupa Dinakaran is the co-founder and CEO of Luminai, an early-stage software startup that automates workflows by taking any multi-click process and making it one click. Formerly a world-renowned Rubik’s Cube champion and a back-to-back corporate Hackathon winner, Dinakaran shares nuggets of wisdom from his personal journey to tech and Luminai’s path to finding product-market fit for other founders.  