Behind the scenes with ilya Sukhar

On today's episode we have ilya Sukhar (Partner at Matrix Partners) in conversation with Vikram Vaidyanathan (MD at Matrix Partners india). Through this conversation, we uncover ilya's journey from founder to VC, learnings & insights gained from his journey at Parse.com, YC, Facebook and Matrix. They then dive deep into ilya's investment in Fivetran -  leading provider of automated data integration. He takes us through the thought process behind the investment and much more. Tune-in

Vikram:

Welcome, everyone to another episode of The Matrix moments podcasts. it gives me great pleasure and welcoming ilya. He's our partner in the US and has been for the last five years. Super excited about this episode going into his journey of being an early-stage investor, a founder. And then about his recent investment in Fivetran, which has done well for us.

What stands out about ilya for me is that he's a got a extremely technical background, bachelor's in computer science and master's in computer science as well. i don't know if he finished his master, but i think it definitely.

ilya:

Yeah. i did.

Vikram:

Both at Cornell ended up in the startup space with a Y Combinator startup called E-tax, which was acquired by Salesforce and then ended up doing something on his own. One of the early batches at Y Combinator 2011. Startup was called Parse. i guess it was a dev tool startup providing mobile SDKs to devs.

At that time, pretty good exit, acquired by Facebook for 100 million in 2013. At that time, that was a big number. spent like three years at Facebook, you might have seen him at a lot of the Facebook dev conference videos, i was looking at an f8 Dev conference video, where i think you were part of the keynote there, did that till 2016. He was a prolific angel investor hanging around the Y Combinator community and as a part time partner at Y Combinator, and then ended up joining matrix in 2016, which we're very thankful for led a bunch of investments here around the core tech, and enterprise tech space, Fivetran, Fox safety, hype machine to name a few.

So thank you, ilya, for joining us and really appreciate the time, i'm going to start with sort of the past days, and maybe the sort of midpoint of the journey. Progress through that time. i know you were hanging out in the Y Combinator community, how did you come up with that idea? And also sort of the founding team?

ilya:

Yeah, i came out of school wanting to get into Startups. And, you know, that's mostly through our love of progrA Matrix Momenting and kind of an exposure to the startup ecosystem, because i happen to have grown up in the bay area here in California. And so, i came out of school, jumped into a company called Ooyala, where i was just an early engineer, and i loved it, you know, i learned a ton, you know, i built a ton.

But most importantly, i got the experience to see you know, all parts of the business, i got to see you know, what it's like to recruit engineers, i got to see what it's like to do a little bit of BD, what PMing is like, just kind of everything involved in getting something to market. And so i just kind of kept making progress on trying to go earlier and earlier and trying to get more and more responsibility and ownership and impact. And so like you said, the next stop on the journey was a company called E-tact, we were kind of making a consumer a personal CRM, which in retrospect, was a very difficult idea. i think we got a lot of users, we had a really hard time monetizing it, we had a nice soft landing at Salesforce. And, you know, i was very fortunate to have that outcome at the time that i did.

it definitely, you know, gave me a little bit of a cushion in the bank account and gave me a taste for kind of what the end journey could be, although it was pretty short. And so it really ended at Salesforce. And, you know, i there's a longer story, but i didn't want to stay at Salesforce.

And so, i did kind of the bare minimum there in terms of making sure our team got integrated and we got our project going, but i only stayed less than six months, i think. And you know, i felt like Okay, the next step is to really start a company that's my own from day one. You know, i want to be a CEO, i want to start a company. And so i was young. This was 2010, 2011 the kind of the hottest thing in Silicon Valley at the time was like mobile social local mobile apps things like Foursquare.

i think instagram was coming down the pipe. Just basically consumer mobile was really the hottest thing and so i pursued that i spent probably six months in my apartment in my one-bedroom apartment on my own, diving into mobile, building apps for my friends, you know, begging them to use the apps, seeing what the feedback was like. And i think, somewhere, you know, somewhere after six months of doing that, i think i came to the conclusion that i didn't have an innate talent for the consumer side of apps, i think, you know, a lot of these apps were about socializing and going out and about big friend groups. And honestly, i was not doing that i was sitting in my apartment writing code and struggling with XCode, and all that stuff. And that's kind of my personality.

i'm not, you know, the most out their social person. And so, you know, i kind of have this moment of intellectual honesty, it's like, okay, am i really going to be the one to come through with, you know, the next great app that competes with Foursquare. But what i did come to understand was the world of mobile tooling.

The kind of the work to be done in this new age of mobile, for developers who are trying to build these sorts of apps, was really an opportunity to work on. You know, everything that people relied on in terms of open source projects, in terms of cloud infrastructure, in terms of tooling, was really optimized for the web, like we were coming off, you know, 10 years of kind of web,  web 2.0, as people called it, you know, JavaScript apps as kind of the primary delivery mechanism for experiences.

And, you know, suddenly Apple launches the App Store, Apple has, you know, a tool chain for compiling the apps on their devices, but they really did very little. And the whole ecosystem had very little for bridging the gap between the sort of work you have to do on the client side to build a nice experience, you know, get all the screens, right, do the transitions, the gradients, the UX, the polished, you know, the Apple way of doing things on the client, and all the stuff underneath the hood.

And the stuff underneath the hood exploded in complexity relative to the web, because you can't just reload the app to shift a new version, you know, the phones aren't always online connectivity back then was not as good. it's still you know, not, something you can totally rely on.

And so, you have to deal with all these new challenges of as a developer, there's a bunch of devices out there, they're running different versions of your app, they're not always connected. You need to cache things you need to deal with this whole new level of complexity.

And so basically, the idea of Parse came from that, which is like, okay people are going to start focusing on how to deliver great consumer experiences, but there's just a set of back-end services that they all need, there's just an 80% commonality between the sorts of things that people are building on the back end, and underneath the hood. So flexible data storage, push notifications, lightweight Analytics, you know, off all that kind of stuff. And so, yeah, that was where the idea of Parse came from.

Vikram:

So, i think i read somewhere, or maybe i heard one of your podcasts that you sort of found your founders in sort of strangers or ad-hoc introductions that were made from the Y Combinator community. And that's exactly the opposite of what we tell others to do, which is to find your co-founders in your shared history. So how did you go about doing that? And what advice do you have for founders who find themselves in that situation where they sort of connect with somebody but it's a fresh introduction? You don't really know that person? And then you have to go on this founding journey together?

ilya:

Yeah, i mean, i definitely don't think it's the optimal way to do it. i wouldn’t do it again. And i wouldn't recommend other people do it. But, you know, i was young, and i wanted to get going. And so the way it really happened was, i had applied to Y Combinator. So, E-taxt was a Y Combinator company. i had joined them right after they got out of demo day.

So i wasn't like a true co-founder, but i was kind of like a late add co-founder, i was, you know, there early, and i was definitely in the mix of the YC ecosystem as a result. So, after that, i applied to YC i applied by myself, i actually applied with one of these consumer ideas. And when i did the interview, i think it went fine.

But they didn't love the idea clearly. And so i walked out. i walked out thinking like i'm probably not going to get in. i'll try again later. You know, my heart was not in the idea, to be perfectly honest. And so i wasn't super dejected. i was like, Okay, now i know what it's like. i know what it's like to interview. it'll be better next time. But actually, as i was walking to my car, it's, uh, you know, got a drive home or whatnot. Paul Graham ran after me. He like ran he like, you know, he's like, got his like, Birkenstocks and his shorts and his, like, orange polo or whatever. And he's running after me. And i was like, oh, man, what did i do, did i leave something behind? Like, why is he running after me, and you know, that the short of it is, he was like Look, we really like you, we hate your idea.

We're gonna let you in, we're strongly encouraged two things. Find a new idea. And find some co-founders. Because they at the time, and i think even today, you know, they're pretty anti’ solo founders. And so there was a period of time between, you know, the interviews when you get in, and when the progrA Matrix Momented actually starts, and during that period is kind of one i, like i said, i kind of had this, you know, actual sit down with myself and was like, Okay, this consumer stuffs not working.

Maybe i'll work on something like Parse. And during that time, Paul Graham, you know, kept in touch with me. And i kept him updated on kind of the ideas and what i was thinking about. And he introduced me to Kevin Lacker, who had also done a prior YC startup, like a Facebook gaming startup. We met up and i remember, we were sitting in like a mission, burrito shop.

And it was kind of this, this moment where we were kind of in the same place where we had this, you know, prior set of projects that we were trying to move beyond, we had a set of new ideas, and we were just pitching them to each other. And i think, you know, he hated the Parse idea, the least of all. And so we decided to try to work together and went through a period of, you know, sitting in my apartment, writing code together and kind of trying to get on the same page. i think we were lucky, we really, i think, found a good fit there. And kind of went to YC and got going. So that was that was a crazy thing we did.  i can talk about the next crazy move we did with regards to founders.

Vikram:

Can i pause you there. Also heard you say somewhere that you always wanted to figure out early, whether you could have arguments with your co founders? And how important was that period of, you know, working together, having those arguments? And what would you advise founders to do in that sort of, you know, critical dating period?

ilya:

Yeah, you know, i think it's helpful that Kevin, and i have both kind of strong opinions, and i think enjoy a little bit of debate. And so it kind of came out naturally. And i think we found ways to productively work through some kind of differences of opinion on various things. And so that was super important. i think in that moment. in that time, i can't say i did it very deliberately, you know, we were just trying to build something, see what happens, you know, that the stakes are low, right?

There was a very high chance that we just kind of parted ways before YC started, so it was fine. i will say i think in later, moments, i've done that much more deliberately. i think other people, especially when there's not this kind of equal dynamic, or you're jumping in together, it's like one person is joining another in terms of, you know, following a lead, i do think it's super important to find ways to bring up potential disagreements. And so when i talk to founders today, or when, you know, i interview people.

i try to be very, you know, nice about it, but part of what i'm trying to do is, pick a little bit of an argument, try to question some of the assumptions and things that built in the past or career moves they've made or, you know, decisions that they feel very strongly about going forward. And i see how people engage on that, you know, do they, it's all over the map, right? Some people clamp up and don't want to engage, some people get very defensive. And i think at the end of the day, good, productive working relationships depend on an openness to revisiting things and working through them.

Vikram:

Go to the next crazy thing that you did?

ilya:

Yeah. The thing we did later, as we were in YC, we're kind of halfway through the three-month batch. i think we've launched our V1, you know, relatively speaking from getting from a standing stop to where we were, you know, we had momentum. And we felt pretty good about what we were working on. i think Kevin and i were a good working duo. We'd gotten to know, you know, it was small batches back then.

So we got to know a lot of people. And there was another team that i think was going through a, you know, kind of a rethinking moment, they were working on something that basically was Stripe, but Stripe at the time already existed, i think Stripe was maybe 18 months old, it was still private beta, but there was kind of a sense of like, do you really want to do that same idea. it's a small community, i think competing amongst the YC ecosystem at the time was not like a super common thing to do.

And so, they were like, we should probably work this duo James and Tecon, were like, we should probably do something else. And so, the crazy thing we did is we basically said, come on board, we'll go from two founders to four founders, we need help you guys are good hackers. Let's do this, like we need more firepower, we're aggressive, let's go. And that was super crazy. i do not recommend doing that.

it sort of worked. it sort of didn't. But yeah, that was kind of the mode we were in, it's like we're gonna, make this thing happen or die trying. And we were not thinking necessarily about the long-term implications of what it's like to have, you know, go from two decision makers that only know each other for a few months to four decision makers that you know each other for a few months. So, in the end, it worked. But i certainly do not recommend that.

Vikram:

An amazing, memorable time. Fast forward, things go well, you guys get acquired by Facebook, you're doing a bunch of amazing things. They're building out their developer platform, you're investing in YC Startups part time as a as a venture partner? So what made you interested in early stage investing first? And then what made you want to do it sort of institutionally and become an institutional investor?

ilya:

Yeah, i think so kind of, i think it's started during my time at Facebook, where i was fortunate to have a pretty big role there, you know, run Parse semi autonomously also be the product leader for the whole developer group, all the APi's, all of the stuff that happened at f8, every year.

And look, at the end of the day, that job is very different from kind of where i was coming from, right, i was, a hacker that loved shipping product, and like the fast feedback, loop of that, and, you know, being in a senior-ish role at Facebook is a lot of people stuff, a lot of resource allocation, a lot of alignment with other parts of the company. You know, it's big company life, although at the time, the company was much smaller than it is today.

And so, i looked at Angel investing as a way to just keep my mind in the place where i had the most fun. And so, you know, meeting up with teams of two that are just getting going, you know, writing them a small check, meeting up with them periodically trying to help them. i didn't think that was gonna be a career move. i didn't think that was gonna make me any money. Honestly, it was like a hedge against, you know, spending eight hours a day, 10 hours a day in meetings.

Vikram:

it was what you did for fun..

ilya:

Yeah, it was just a fun little hobby. And i was fortunate to get asked to be that part time partner at YC. And so, i, you know, i did interviews, i did office hours, and that was still just fun. Like, i don't think there was a clear path to going full time there. At the time. it's a bit of a family business. And so, i was just, you know, just having fun. i think the way it turned into a more serious thing was after i left Facebook, i came out of Facebook thinking i would start another company. i had a bit of a chip on my shoulder with regards to the parts journey, because in the end, i think we sold too early.

i think we had a really big opportunity at Facebook and we're seizing it but for good reason for lots of reasons i can get into. The company basically moved on from our project, that sort of strategic underpinnings of acquisition, you know that the desire to give us headcount and kind of  fund our project, over three years kind of went away.

And so i had a bit of a chip on my shoulder, and i was like, i'm going to do it again, i'm going to surpass myself, i'm going to go much further in the journey. And i spent like six months trying to spend on projects. And again, now, having had even more experience, i had a moment where i was like, i don't think any of these projects are, the sort of project i want to spend another 5, 10 years of my life on. So maybe i need to step back. And maybe i'll start another company again, one day, but i'm not gonna start it right now. i don't want to force it. i've done early-stage Startups, you can't force it. if you do it just to do it, you're in for a world of pain. And so during that time, again, i was for fun meeting up with people writing Angel cheques.

And i was getting hit up by VCs, like, you know, real VCs with funds and cheque books. Mostly, i think, because they were curious what i was going to work on next. And in those conversations, i think i started to reveal like, Hey, i thought i was gonna start another company. That's why you're here with me. But i don't think it's gonna happen right now you're wasting your time, i'll let you know. And a lot of people were like, cool.

Thanks for letting me know. i'll see you later. And but some people were like, Hey, we noticed you on a bunch of cap tables that are interesting. We noticed you angel investing out there. You know, have you ever thought about doing this more seriously? And my first reaction was like, No, i'm not, you know, i'm a builder. i'm not gonna, what are you talking about?

Like, i don't know how to do that. But i think after some conversations, and after some reflection, i was like, actually, i am doing that, you know, i'm out there, trying to find good opportunities, trying to help them, trying to get on the cap table, i think relatively successful at it. And so i looked at it as like, Okay, i didn't really set out to be a VC. But i'm kind of voting with my feet, i'm doing it with my own checking account. i'm doing it with my time. i don't need to be doing this. i can do a lot of things. So i must enjoy it. And so i had this moment where i was like, Okay, i'll explore it. And then through conversations, i ended up joining matrix.

Vikram:

So, it's been about five years as a VC, what makes you stay?

ilya:

Well, what makes me stay. i enjoy it, i think, this might be a little bit controversial. And there's definitely flip sides of that. But i think that's actually much more intellectually stimulating than being a founder, in a lot of ways, like, being a founder is very rewarding, in many other ways. But you've got the blinders on, you're working on your narrow slice of the world, you're very focused on your team, you're very focused on just pushing the ball forward, you know, inch by inch foot by foot. i think the role i'm in today, i get paid to learn.

And i think that is super special and rare, right? Like, i can keep expanding my horizons, i can keep learning new things. And i can be on a number of journeys, you know, at the same time, and so i just really enjoy that. Like, in the last three weeks, i've been getting a little bit more interested in the intersection of technology and Healthcare, not a place i know anything about. i spent the last three weeks going deep on it.

And i still don't know anything, but i know a little bit more. And i think that's remarkable. i mean, i think the other thing is, at this point five years in, i've got a portfolio that i really care about, there's you know, multiyear relationships, some of them are five years old, some of them are three, four years old, where i really want to see how those stories end. it's, a special thing to like, build those relationships beyond those boards, see the progress. See people get through hard times, and there's more to come.

Vikram:

Awesome. You know, i've asked this question many different ways to different VCs. And it's amazing how consistent the reasons are. The one thing everyone says is that you get paid to learn and grow. And then you can be relevant across cycles. And the second it's the relationships and journeys that you get to go on with founders and at least for me that's what keeps me going. it this is sort of a great segue into Fivetran. And i'm just going to introduce FiveTran, for our listeners, it's the best way to describe it. it's a data integration software company. it's essentially does data connectors.

And we'll get into what those connectors are, takes data from diverse sources for large enterprise, figured out automatic connectors, and how to update those, that set of data into a data warehouse. And then it's good to go for your data scientists and your business analysts. Now, it sounds pretty simple.

But it's got pretty nuanced, and complicated technology in it, ilya led the series A, 15 million in December ‘18. And we'll get into the history of that company till December 18. And since then, the company has just been on a tear, there is 5.65 million from Andreessen and GC earlier this year, at a 5.6 billion valuation. i was just congratulating iiya on that investment, and that journey. So, let's get into that story.

And i want to start with George as a as a founder, i had the pleasure of listening to him at one of our annual meetings. He seemed like an incredibly unassuming, humble guy went back to his background, and i was surprised to see a bachelor's in cognitive science, a PhD in neurobiology, didn't seem that and then he just got into startup. So really didn't seem like the background to start, like a core tech company. And seems like there's like an amazing founder story there. So, what's the story?

ilya:

Yeah, that's right. He is a scientist really kind of, by background, i think immediately prior to starting this company, he was in a therapeutics company. it's an interesting journey. And i think it has some parallels to mine, which i think we talked about and kind of hit it off on. So he and his childhood friend, Taylor started a company that, you know, has like vague overlap with what Fivetran does today, but really started out quite different. So, they started out, basically trying to build a new interface for data science.

Like a whole new UX for it, i think it came from his experience struggling with kind of incumbent tooling to get his job done as a bench scientist in a lab. And they didn't have much success with it, you know, it was the sort of thing that i think they built probably a little bit in a vacuum, a little bit based on their conception of what people would want. And i think they went through a really hard time trying to get traction for it to not get a lot of love coming out of YC kind of, you know, graduated and had, you know, the immediate trough of sorrow that progrA Matrix Momente talks about.

And i think the emergence of Fivetran came again, you know, this is where the parallel is to Parse and in my journey from, thinking through, like, what are the underlying problems that actually have, you know, more tractable solutions that they can go target that will maybe one day enable kind of the original vision of making data science easier. And so the way it boils down to, is they went out to a bunch of customers trying to pitch this new platform, the customers like, we're like, you know, our actual problem is, getting data from you know, one place to another.

We are adopting these new cloud data warehouses, redshift, Snowflake, Big Query. Theoretically, we are on new HBi, tooling, new age data science tooling. You know, maybe we'll talk to you later, but we were just in the early innings of replatforming on this kind of new stack that at the time, had no real term for it. it was really just these cloud data warehouses that people were excited about.

And so they started to strive to like, solve whatever problem they could for those customers. And i think that basically meant building, bespoke connectors to take data from Salesforce and get it into Snowflake data from Zendesk and get it into Snowflake. And they had a really unique approach to that relative to kind of the alternatives that we can get into. But i think that is the evolution of the company and so Fivetran is not one of those stories that you know, , a brilliant idea from day one, it is a multi-year journey.

Vikram:

A little bit of the journey, they started up in 2013 ended up coming out of YC. Maybe around that time. And then they really didn't go anywhere till 2017, 18. And, you know, as we see is you usually have a bias against if this hasn't gone somewhere quick, then it might not go anywhere. So how did you get around that bias? And how did you get to backing these guys? And what sort of stood out about especially about George and his team that you backed them then?

ilya:

i mean, i think what helped is i was coming from a place of seeking out solutions to the problem that they were now solving. And so my intersection with them was, i had had, through other conversations become aware of the impact that these new cloud data warehouses were having on kind of the entire stack.

So it actually started with Redshift. Right, i kept hearing Redshift come up. And then i started talking to companies working on really just kind of solving a lot of the problems with Redshift. So there was a wave of companies that are basically APM and tooling for Redshift, because Redshift has a lot of cost and performance issues, i didn't end up investing in any of those, then i started to learn about snowflake and Big Query.

And kind of understand that this movement was happening you know, what do you need to do to actually make use of these cloud data warehouses, you have to get the data in there, it's pretty simple. At the end of the day, there's a whole you know, incumbent world that deals and ETL, i had built ETL pipelines in my first company out of Ooyala, with like the Hadoop ecosystem, hive, HDFS, like writing Java code.

So, i was familiar with the problem space, i met some of Fivetrans competitors first actually, there was another YC company called etiLi, there was a Luma, which was an israeli company, there was Xplenty, i didn't meet all of them. But i met a subset of kind of half a dozen competitors that Fivetran had. And actually, Fivetran was probably the last one that we ended up meeting. And so i think we came to the conversation, primed and believing there is a problem to be solved here. And there's an opportunity. And i think, what blew me away, what blew the team away was, George and Taylor had a very different approach to the problem very. i, you know, in some ways, at the time controversial approach to the problem, i think that the entire, history of ETL was flexibility, was the most important.

Vikram:

So i'm just gonna pause you there. And i think we're gonna get into the meat of this. So ETL for our listeners is extract, transform, load. And that's everyone was doing it. So you have all these data, warehouses,Ssnowflake, and so on, are getting cheaper. And then you had all these data scientists working on them.

And there was sort of a established way of getting data into these warehouses, which was extract, then transform sort of at the source, and then you load it into the Data Warehouse. And at least the way i understand it, what Fivetran came up with was flip it on its head, and instead of ETL to ELT, in some way, and do transform at destination, and then i'm going to hand over to you because that's all i understand.

ilya:

Yeah. So i think what's important to understand about kind of the prior ETL approaches is all of the people who had done it before, you know, when i was writing the Java code, and my first job out of college, you're working with different constraints, you're working with constraints where the data warehouse is really hard to manage.

it's expensive. it's difficult to scale. And so the transform is really important in that in environment, where you'd have tones of source data. And then you use the kind of whatever the transform layer is in your stack, whether it's, you know, proprietary product like informatica or its, you know, Hadoop code, like i'd worked with, to whittle the data down to like, what you actually need out of it, in that moment, in the roll ups and in the dimensions that you need in that moment, so that you store kind of only what you need, you don't like, you know, explode your data warehouse with resource demands.

And you let kind of the consumers of the data, you know, have it in the format that they need to just get going and work on top of it because You know, you want them to do really simple queries to kind of then work with it later down the road. i think what changed, and what George and Taylor really saw was the Cloud Data Warehouses just turned that upside down. it's actually quite cheap to just dump all your data in there. it's quite performant, to query it. downstream. However, you need to, you know, crunch it later.

And it's just not that expensive at the end of the day. And so you just get to a place where you can store as much as available in approximately good format, the best format that you can generically imagine, then why wouldn't you? And so that's basically the Fivetran ELT approach at the time really, they only did ENL. You know, they talked about ELT, because you don't want to just drop the t and have people asking where it went. He was like, yeah, just do it later, do it in in SQL on top of snowflake, or whatnot.

But they really did ENL. And so, their approach was like look. Well, the other thing that's worth mentioning is that when you look at kind of where data lives in an enterprise 20 years ago, the vast majority of it lives in your own databases, right? It's in your proprietary databases that are generated by your own code. And in the schema that you designed for your business, and that still exists today. But like, actually, where data lives, in enterprises all over the place, like everyone's using a bunch of SaaS tools. And so when you think about centralizing your data and running analytics on your business, you can't just be looking at your own database, you have to go pull data out of Salesforce, and Zendesk and Stripe and NetApp, and whatnot.

And so the combination of those two big shifts is the insight that they had, it's like, Okay, we're gonna build a bespoke connectors for all of these data sources. So you're going to give us your Salesforce instance. And we're going to give you, at the end of the pipeline, a set of tables in your Snowflake instance, that have all of your Salesforce data, in a format that is very, you know, nicely laid out, it's very thought through, we've thought through all the edge cases, we've thought through all the error handling, we're gonna give it to you kind of in this golden schema format, it's not gonna be perfect for you, because we don't know what you want to do with it, but it's gonna get you quite far. And I think doing it that way is super controversial, because it assumes a very different set of constraints.

And it basically takes all the control away from the customer. It says to the customer, you know, you're used to building these connectors, yourself, you're used to tones of knobs and tuning, you don't really need that. Our advantage, Fivetran’s advantage is, we see everyone's Salesforce instance, we know the variety of things that are quirky or painful about their API.

We know the pitfalls you can fall into, as you lay out this data in your data warehouse and try to do things with it, we're gonna apply a lot of opinion and judgement, that is unique to that data source, in building this connector, and it's just going to be, you know, a one click thing. There's nothing to configure about it. You author the instance, and there it is.

Vikram:

So, one of the things that stands out, for me, at least looking from afar, is the is the focus that the company had on just doing that, which is, you know, beginning and I saw some of the key metrics that they published, it was focused around 99.9% reliability.

And then the number of these bespoke connectors that they were offering, and that was it. Right, and maybe there were a few others, but that kind of focus usually indicates something special happening in the company, which is a very high clarity of focus on what they want to do and what they don't more importantly, and second, that sort of seeping into the underlying culture. Can you sort of comment and share more on that?

Ilya:

Yeah, I mean, that is absolutely a big cultural thing at the company. I think George likes to talk about it like, you know, we are like a utility company like you give us connections to your data, and we move it to the place you want it to go. We want it to be as reliable as simple as you know. Generic as you know, the wall socket giving you electricity. We are you know, laser focused on just solving that problem.

And I think a lot of people you know, three, four years ago would say that's not enough, or there's not enough complexity there. And I think what they understood is that there is plenty of complexity there. And doing building a connector for a data source, that is like 95% good, is like 10% of the work of, you know, building a connector, that's 99% good. Because these API's are not as like, they don't behave as predictably, as you think there are all sorts of edge cases, there's all sorts of errors that come up, every customers instance of Salesforce and Stripe or Netapp or whatnot, ends up actually being quite different.

And so they have always been focused on just climbing that frontier of quality, and just always trying to be better and better and better. So that the next incremental customer, when they hook up their source, it just works. It's not like a thing that you just finished doing. Because the sources evolve, the way they generate data evolves, eras evolve. And there's no magic to it. Yes, you can build frameworks, you can build tooling that makes it easier to deal with certain types of problems. And, and they have done a lot of that.

But you have to innately understand these data sources. And yeah, they've always been laser focused on that

Vikram:

So that’s a good segway into the next topic that I wanted to talk about it’s an extremely technical founder working on an enterprise tech product. Usually, when I see that combination, they struggle with sort of the GTM, right? The marketing message, the sales motion, but it seemed like these guys figured it out pretty quickly. So what are some of the learnings from that period? And what sort of worked for them? How does that sort of work for technical founders trying to build this today from India?

Ilya:

Yeah, I mean, I think one thing that's worth mentioning is, George is a very technical guy, right? Again, he's kind of a scientist, by training, but his co-founder Taylor is not, his co-founder Taylor, I think is probably a design mind. But by training, but also the kind of personality that's good at, like hearing people out that is good at the like, early, you know, nuances of early sales. And so I think, when their first idea didn't work, I think the way that they sort of split the responsibility is was like, Taylor you go talk to these people who aren't buying our product, about their problems, and I'll go build a solution.

And we'll just iterate from there. And so, I mean, the lesson learned is, at the end of the day, you know, it is a mix of skills, and they were fortunate to have that mix of skills and lots of trust they are childhood friends and have known each other forever. And so they had that trust to kind of split the tasks and keep charging forward. But yeah, this is something that I think I struggle with as an investor, which is, I'm a technical person, I think, you know, I became decent at sales, you know, this job, in some ways is about sales that we're in. But then they it's not kind of my strength area of a personality.

And it's not the place where I spend my most, you know, precious brain cycles thinking about go to market tactics, but you must do it right, you can't get around it, it is a key part of getting something off the ground. And so one thing I've come around to looking for is that balance of skills on that early team. Or at least they desire, a self-awareness on behalf of the technical founder to compliment themselves to bring on go to market talent, people are going to do those early sales like I think there is much to be learned in the early days by technical people attempting the sales themselves. But I think there is a certain hardheadedness about that I try to avoid, because I've had some painful lessons learned myself and seen in other companies where success comes from a mix of skills for sure.

Vikram:

Let me ask you a more specific question on just selling sort of enterprise tech products, the, you know, true moment of value in these products comes after you've made a bunch of investment as a buyer, in buying the product and implementing it. And then you know, whether it really works or not. And so, oftentimes, in the early sales process, the buyers trying to make up their mind on whether they should spend that time and money.

And so it seems like it comes down to some technical differentiation in that pre sales process. And if you hit the nail on the head on that differentiation, then it sort of works- is that the best way to think about it for sort of founders were trying to convince their first two, three customers to, you know, co-develop, pilot it, whatever?

Ilya:

Yeah, I think so. I think in the case of Fivetran, for example, actually, I think this is a core strength, which is the incumbents, the old ways of doing ETL are things that you have to buy into and then invest a lot of time into, I think the Fivetran experiences is totally different way which is like, you know, give us some credentials, authorise one of these data sources and, you know, go away for 10 minutes, and we'll come back and show you what you've got in your Snowflake instance.

And so it is a very fast, you know, demonstration of that aha moment where, you know, yeah, it's not going to be all of your data immediately. There is not going to have all of your connectors. But you see the value of their approach and kind of what's different about it? And like you said, the technical differentiation. And so yeah, I think that's what I look for.

And I think that's today what's most successful out in the market is you don't have to solve the customer's problem entirely in that pre sales process. Your early product, you know, probably haven't built enough to even try. But I think you have to find a way to demonstrate that inkling of value that makes people intrigued, that makes them appreciate what you bring that's different to the table, whether it's technical, or it's through some workflow, UX, innovation, or whatnot, where they'll spend the time with you, I think, just pitching them on the end value prop. And sort of the old traditional sale of, you know, we're gonna, give you this big behemoth, and we're gonna configure it for you, and you're gonna buy it all at once. And six months later, it might be delivering value to you like, maybe it works, but I don't spend any time talking about the sorts of companies.

Vikram:

So this has just been a amazing story. And you know, when I look at where Fivetran is, where you have all of these data connectors, to all of these big products on one side, and then these data warehouse, the other side, just seems like they've got like an unbelievable position on this data highway. And I can't wait to see what they do next. And what you do next with those insights, and what kind of companies you invest behind. So, I'm looking forward to that. I want to come to my last part of the podcast.

And, you know, you've been sort of tinkering around with me and helping me think through what we're doing in the India SaaS corridor, and especially the Indo US corridor, and some of the core tech founders that are coming out of India. And I think you're pretty plugged in to the India diaspora in the US as well. So, do you see things changing with these Indian startups coming, especially with pretty technical ideas? What excites you? What worries you?

Ilya:

It's a good question. I mean, definitely, the quality of the founders I encounter is, you know, is excellent. I feel like there's, you know, I think there's less of like a big disconnect. Right, I think the people that may have found their way to the US. Now, you know, they're connected to the US, they are maybe fundraising in the US to some degree, or they're looking for early customers. But I think to our detriment, I think they are less likely to, come here. And I think they build for their market for what makes sense for a broader world. And so, you know, I wish I had some grand insight, I think the interesting thing that kind of intersects with some of my overall investment areas is I think, we're just getting to a place from an infrastructure perspective and a data perspective where the world for better or for worse is balkanized.

A little bit like it matters you know, where your data lives and it matters, how you build your product relative to the regulation and consumer demands of various markets.

And so, I think infrastructure, companies, Developer Tools, company, I think, have more of an opportunity to take a lot of the UX insights, a lot of the like, aesthetic and best practices that have developed in the US for, I don't know, 5, 10 years now, and apply them all over the world in a way that leverages what's unique about those markets. What's unique about the applications that people are building on top, what's unique about the demands of enterprises in India, in China, wherever you end up with different products, right?

I think there's a lot of things we look at together, that I'm like, oh, this is like the Indian version of XYZ. And when you dig into it, it's like, no, yes, it is that Indian version of XYZ at a very high level, but there's lots of nuance and insight that's unique to the Indian market that you know, makes it a big, big opportunity. That's, interesting for us to invest in.

Vikram:

And it's first thing that you see that one and the second is, I think what's different for me is today, they're able to find their first customer, that they're sort of co-developing the product with, which has the same level of complexity or maybe greater sometimes, because the concurrency of these platforms in India, as the US counterpart, and therefore, or the level of product that they're building seems to measure up versus being sort of like a low-cost version of what somebody is building the US.

And I think the second thing, at least what we're seeing is at least one founder moving to the US quickly. And like you said, because you know, especially around the data stack, people are sort of balkanizing, I think that's the word you used. It just makes sense for you to be sort of multilocation or multi-Geo? Probably from day one. Definitely, from like, three months in.

So the last question, more sort of broad thoughts on future of data stack, it seemed like you're spending a lot of time on the intersection of health and tech, I still spending more time on the data stack. What is sort of big areas of interest?

Ilya:

Yeah, for sure. I think there's kind of two interesting things to say there. One is, I think, more of an investment lesson learned, which is, I think, it was great, you know, what I'm really proud of, in regards to the Fivetran investment is, we were early to that ecosystem, not the earliest, by any means, you know, obviously, Sutter Hill did a really nice job of Snowflake, and they've, reaped the rewards there. But I think relatively speaking, in terms of appreciating, you know, what this new movement is, what the implications are, what are the pieces required to actually make it happen? We were pretty early to that.

And I think that the Fivetran investment is testament to it, I think, you know, when the lessons learned is, all good investment areas become pretty consensus over a couple of years. And so certainly, after Snowflake went public, you know, this whole modern data stack term was coined, there's been a lot of investment in the space, you know, every little box and a diagram of architecture is, you know, has like 3 or 4, 5, 6, you know, very well-funded competitors. And so, I spent a lot of time looking at many of them. Um, we've made a couple more investments in the space, but I think a big lesson learned is at the end of the day.

This job is about being early to big trends. And I think the best trends, you know, become consensus, and the investment opportunities become less attractive. And so I think we're seeing a big overfunding of the space and a lot of ways, and I think we're gonna end up seeing a lot of consolidation. And a lot of you know, a lot of battles play out over the coming years in the space. And so, yeah, one of the things I'm always focused on, which is why I'm dabbling in health or other things as like, I think I don't want to make a career out of one space.

That said, the second thing we're saying about it is I'm really excited about what happens once this kind of movement as it is today gets digested by a lot of enterprises, and you go into any kind of serious company, or any startup really, and it's like, they've got the stack, they've got Fivetran. They've got DBT, they've got one of the cloud data warehouses, it's just it becomes kind of this, like, assumption that it's there. And they have a Modern BI stack on top of it. The question is, what's next, you know, like, the data's there, they have all the data that they gathered from all over the enterprise, it's well organized, it's scalable, it's cheap.

It's fast to access. Everyone's gonna look at a bunch of new charts as a result, which is fine. Like we've made some investments there. I think there's good opportunities there. But I think the biggest question is what's next? I think the data stack as it is today, the data warehouse becomes potentially a big operational leverage point, that starts to flow back to, you know, how does the consumer experience the product, what decisions are made real time whether it's FinTech underwriting, risk, or, you know, optimization, that is the result of the fusion of this data in one place that was very hard to accomplish?

And prior generations of data like that is where I'm spending my time is, you know, now what do we need to do is to make this movement happen in a two year time frame, but what's going to happen once it's there, and people want to do more with the data on a five year time frame or a 10 year time frame? And so that's why I'm spending my time on.

Vikram:

Ilya this has been a fascinating conversation. Thank you for spending the time it's clear, you know, given your background what you've done over the last five years, you have a way of seeing what's going to come next. And I'm really looking forward to the kind of investment you're going to make and looking forward to finding things to work on together.

Ilya:

Yeah, appreciate it. Thanks for the conversation.

Vikram:

Thanks.

Salonie: Thanks for tuning in for more matrix moments episodes you can head to www.matrixpartners.in/blog You can also follow us on twitter LinkedIn and YouTube for more updates.

‍