The Challenges a Repeat Founder Faces

by Y Combinator9/8/2017

Tikhon Bernstam cofounded Scribd (YC S06) and Parse (YC S11).


Google Play


Craig Cannon [00:00] – Hey this is Craig Cannon and you’re listening to Y Combinator’s podcast. Today’s episode is with Tikhon Bernstam. Tikhon is a a multi-time YC founder. The first time went through YC was the summer of 2006 with Scribd and then five years later during the summer of 2011 with Parse. Given that we’re about away from the winter 2018 application deadline, I thought it would be cool for Tikhon to come in, talk about his experience, talk about his decision to do YC a second time given that Scribd had already worked out pretty well. And talk about his experience has been like as an investor these past few years. As always if you want to read the transcrit or watch the video, those are both at If you’d like to apply for the winter 2018 that is at Alright here we go. Could you start by explaining how you first found YC?

Tikhon Bernstam [00:49] – Yeah, I actually found YC ’cause I was on Reddit. I was in graduate school, which meant I had a lot of free time. Reddit was super early, this was back when the top post might have like 10 points. This is like before it even had comments. Via that I found Paul Graham’s essays on how to do a startup, and to me, the idea of doing a company was basically impossible. This was after the bubble, or after the dot com crash. Everyone had basically declared the web was over, like that’s all done now. Paul Graham’s essays were really the only source that I had found that kind of said, actually, here’s the pathway to do it, and it can be done. I was super interested in that. I applied to YC on a, really on a longshot, I didn’t think I’d get in. I interviewed in Palo Alto, and although the actual program was still back in Boston in Cambridge. I got in, so I just dropped out of graduate school.

Craig Cannon [01:59] – What were you studying?

Tikhon Bernstam [02:01] – Physics, actually. Theoretical physics. But I had studied computer science and econ in undergrad, so that was kind of a small change. I dropped out of grad school and moved to Cambridge, and then me, Trip and Jared probably worked on Scribd for at least a year-and-a-half before we got real traction on it. This is back in summer 2006.

Craig Cannon [02:31] – What was the initial pitch for Scribd in the application?

Tikhon Bernstam [02:35] – Well, so Scribd was actually one of those pivot companies where we applied with different ideas, and then, much like Reddit was actually a pivot company. It was actually applied for as something else, like food delivery stuff.

Craig Cannon [02:48] – Really?

Tikhon Bernstam [02:49] – Paul’s like I love you guys, Steve and Alexis, but I hate that idea. How about this kind of thing? Steve’s talked about that before. Similarly, Scribd actually was the merger of me, Trip and Jared during YC. They were working on some other project and it was like a Craigslist for colleges. Back then Craigslist was super hot, so everyone was trying to do… Much like nowadays everyone’s doing Uber for X, back then it was Craigslist for X.

Craig Cannon [03:21] – Glory days on the internet.

Tikhon Bernstam [03:23] – But then YouTube sold for over a billion, and so people said that was insane, oh my God, Google overpaid! Which… Totally wrong. And so then YouTube became the hot one. So we said, let’s do YouTube for documents. Like that was literally like the thinking. And we also noticed this problem, where Trip’s dad, who’s at the Stanford Med School, was having trouble basically viewing the documents online that he wanted to read, and he didn’t have the right reader. Sometimes I’d send my mom a post script file and she’d be like, how the hell do I open this?

Craig Cannon [03:59] – People were just downloading it and then they needed to download Acrobat or something?

Tikhon Bernstam [04:03] – Yeah, ’cause back then if you had Adobe Reader, I don’t think that actually opened up like post script files back then. Maybe it did. But a lot of default viewers didn’t. It just depended on your OS and what sort of default software you had bundled. But anyway, we just decided why don’t we allow the documents to be embedded inside of the webpages? And we started off with Flash, but we always hoped that we could switch to HTML5, which we did much later.

Craig Cannon [04:29] – Okay. What was the YC experience, and so summer ’06, what was it like?

Tikhon Bernstam [04:34] – It was great. This was back in Cambridge. Honestly, we would never have made it without that program, and I think that’s true of actually most of those companies, and I’d argue probably most companies. Because without that community of like-minded people who were highly motivated and working on the same general pathway as you, and also people who’d done it before, like Paul, Jessica… Paul, JL, Trevor, and Robert who were basically there to, “Like no, that’s stupid, get back on course.” ‘Cause sometimes we would just go the wrong way for a whole week before we’d see them and we’d show them and they’d say, “What the hell are you guys doing?” And they were always right, is the thing. It saved us years and the company wouldn’t have worked without them. I’ll always be grateful for that. All the batches were small back then, I think Scribd summer ’06 was the second batch, so still my best friends are the people from summer ’05 and summer ’06 batch in Cambridge, and that’s going on over a decade. Just because those were the only people I knew doing startups, and they were the people I’d talk to, and it was great ’cause with startups, one day you think your company’s going to fail, you think you’re getting sued or something, someone’s threatening you, and the next day you think you’re going to sell to Google for a huge amount, just because some person at M&A accidentally emails you. Having other people going through the same thing, like the Twitch guys kind of working on their stuff, which took over a decade–and back then it was Kiko, which switched into–but like having those people around and that community and then the guidance of the partners, is probably the only reason that we survived.

Craig Cannon [06:32] – How much money were you getting from YC at that point? It was not the same as it is now.

Tikhon Bernstam [06:37] – Yeah, back then it was fixed, it was $6,000 per person, so overall we took $18,000.

Craig Cannon [06:43] – Congrats, that’s huge!

Tikhon Bernstam [06:44] – We almost sold the company for that much.

Craig Cannon [06:48] – Obviously the money lasted. What did you do with it? People get way more now.

Tikhon Bernstam [06:55] – Yeah, I’m always amazed how much people get now, because the seed rounds today are bigger than the Series A that Dropbox and Airbnb did with Sequoia, which I think was roughly a mil each. And so I’m like, how did you guys spend all that money? It’s that people run out. But back then, we struggled. For a couple of years, I was living on savings. We were basically, I was basically living out of Trip’s parents’ house at Stanford and his mom was getting irritated ’cause I keep eating all the candy, and I was subsisting on Top Ramen, and whatever food was available in their cabinets. Trip, Jared and I worked for months basically out of his living room, and without the generosity of his parents to even let us do that…

Craig Cannon [07:39] – Did you raise after YC?

Tikhon Bernstam [07:40] – We did not.

Craig Cannon [07:42] – Or was it a while?

Tikhon Bernstam [07:43] – Back then, fundraising was very, very different than it is today. Also, just the environment in Cambridge is much different than the Valley’s. But back then the mood was so gloomy and startups are over sort of thing, that it took a really… There actually just weren’t that many angels around. There weren’t that many people who had exited, or who wanted to do that sort of thing. I mean now, Demo Day is oversubscribed and such, but back then it wasn’t. Back then I don’t think it was even a full room, actually.

Craig Cannon [08:15] – You were just basically living off savings? Like Scribd wasn’t making enough money?

Tikhon Bernstam [08:20] – We actually stretched that $18k pretty darned far. I mean if you can somehow, as long as you can get your rent low, you can live for a long time.

Craig Cannon [08:29] – Okay, do you know approximately how long that actually went?

Tikhon Bernstam [08:32] – Well, I mean, yeah. It was another year after YC, so we scratched by on that, and being a wealthy grad student, so having zero savings for really any of us, ’cause Trip had just left Harvard, Jared dropped out of Harvard to do this, so we just really tried to make every dime count.

Craig Cannon [08:54] – How long did you work on Scribd?

Tikhon Bernstam [08:58] – We started like I said, summer 2006. I left about five-and-a-half years later. I talked to the co-founders, I said, “Listen, I think if we’re not going to go more B2B, like more like enterprisey, I think that I should probably go, because I’m really interested in doing something in that space.” And so they were like, “Okay cool.” They actually invested in my next company, which was Parse, which we basically, actually two of us who left, it was me and James U. from Scribd. James was actually our first hire at Scribd. He and I were close friends and still are, and so we left and the very next day, we were coding and working on the prototype of, well it was not Parse, it was something else.

Craig Cannon [09:48] – What, okay, pause. What was that idea?

Tikhon Bernstam [09:51] – There were actually 12. We made 12 different landing pages. With Scribd we would spend about one month on every idea, and build out an app and then test it, And that’s kind of why it may have taken longer. With Parse, we said let’s not waste a year, let’s make 12 landing pages, kind of launch them, and then see if people will even give us an email.

Craig Cannon [10:12] – In B2B products?

Tikhon Bernstam [10:13] – Yeah, in things that made money, ’cause I think in B2C it’s much harder to gauge whether you have a long term traction or not, or if you’re Chatroulette or something. We basically had 12 different pages that we launched. One was an API for companies to send gifts to their best users, ’cause at Scribd I always wanted to send gifts to people that uploaded 100 documents. It would be nice to do that programmatically. I don’t even think that exists, still. So we just looked if people clicked on the pricing page, and if they’d signed up, put an email in, and then anyone who typed in their email, we would email them being like why did you sign up for this? This was a crappy landing page that we made in two hours.

Craig Cannon [10:57] – Wait, how were they getting to the landing page? Were you doing outbound marketing for any of this stuff?

Tikhon Bernstam [11:02] – Show HN was basically–this was before Product Hunt, so it was mostly Show HN, it was on Twitter.

Craig Cannon [11:09] – You were doing Show HN landing pages?

Tikhon Bernstam [11:10] – Yes, like Show HN the service exists. Actually Parse launched that way, too. Parse launched as a landing page.

Craig Cannon [11:17] – Really?

Tikhon Bernstam [11:18] – Yeah, back then it was called ZStack, and we didn’t own the domain, which we only got later thanks to the Stripe founders, and Stripe CTO. It was called ZStack. We launched a landing page again, just, we wanted to make sure people actually wanted this thing. ‘Cause we knew from our own experience building mobile apps at Scribd, at other companies, that everyone was sort of reinventing the wheel, reinventing the same code that push notification type stuff that really could be generalized. Back then AWS was still very much in its infancy, but you could see the need for… Heroku had just sold for $250 million to SalesForce, so the pitch of Heroku for mobile was a very… I actually dislike X for Y pitches now, like all these now Uber for Xs…

Craig Cannon [12:10] – But you were riding those waves in the past?

Tikhon Bernstam [12:12] – Yes. I am 100% guilty of doing those in the past. I think the pitch deck I used at YC at the Demo Day was Heroku for Mobile, the actual title of what the company was.

Craig Cannon [12:29] – Okay, and just stepping back a little bit, what made you decide to go with ZStack versus the other 11?

Tikhon Bernstam [12:37] – It was actually talking to the users. We had talked to the users on basically…

Craig Cannon [12:42] – But to be clear, there was no product for any of them?

Tikhon Bernstam [12:44] – Parse is more complicated. Parse was actually the merger of three independent entities. So Ilya was a solo founder in Parse, working on a find your friends app sort of thing, which was very popular at the time. And Kevin wasn’t even in YC. Kevin emailed–Kevin had been in YC previously, and then was working at Google, emails Paul and says listen, I want to get back into startups. If you have anything, let me know. Paul emails the list like there’s this great guy, great hacker, you guys should talk to him if you’re looking for a co-founder. And so James and I met with Kevin. Ilya met with Kevin. We ended up all trying to work together for like a couple of weeks just to see if we harmonized. I don’t advise people work with people they don’t know. But hey it worked for Twitch.

Craig Cannon [13:32] – No, it’s worked for some people. It hasn’t worked for a lot of people. Okay, so what were those two ideas, then, that merged?

Tikhon Bernstam [13:41] – Yeah, so we were actually more in the payments space. James and I were interested in recurring subscription billing space, where people were, people still have trouble getting recurring billing working. It’s getting better now, thanks to Stripe, but stuff for normal people basically take payments like that, that were recurring. We were still very much in the ideation phase, where we were very happy to pivot to whatever we felt would be the best type of business. Then basically we all just started working on Parse and we basically launched that landing page pretty quickly into YC. And we launched it on Hacker News, it was a Show HN, it was called I think at the time.

Craig Cannon [14:35] – And so what were, like what was the pitch?

Tikhon Bernstam [14:37] – It was literally a ThemeForest theme. It was like a $9 theme forest theme and it has these hot air balloons on it. You can still buy it today. And just it was basically Heroku for mobile. Like basically we’re going to take away the dev ops and the back end part of it, so you can just focus on the front end native code, and not have to worry about thinking about your database and all that stuff, abstracting it away the SDKs.

Craig Cannon [15:06] – I mean I’ve used the product before, but just to be clear for everyone else who’s listening, could you just give the quick explanation of what it actually became?

Tikhon Bernstam [15:13] – Yeah, Parse is a platform that helps you make mobile applications much, much faster. We estimate roughly 10x faster. Normally if you’re making a mobile app, you’d have to write all of the code, the native client interface on the phone that the person would play with, but then you’d have to think about how does that app talk to some server in Virginia, and what does it send, and then where’s that data going? How am I querying it? What happens if that server goes down? Parse took care of that whole second part of this stack, so that you just did the client stuff. You make a pretty native UI. We didn’t believe much at the time in the HTML5 approach, which now is working better, but the sort of PhoneGap type of thing. We believed that native apps at the time were definitely superior in experience. We just tried to abstract a way, basically the really hard 80% of stuff that people shouldn’t have to know how to do in order to make apps.

Craig Cannon [16:22] – Had you guys decided that you wanted to go through YC again?

Tikhon Bernstam [16:26] – Oh yeah, I talked to Paul almost immediately.

Craig Cannon [16:29] – On leaving Scribd?

Tikhon Bernstam [16:31] – Me and James have these ideas, we’re not sure exactly which one we’re going to do. Paul was like, all right, let’s do this again.

Craig Cannon [16:41] – Really? So, okay, you applied kind of blind? There was no product still?

Tikhon Bernstam [16:47] – Well, we agreed with Paul that we would focus on a company called PintPay, which was the subscription billing service. Back then, billing and payments companies were super hot, like WePay just raised a bunch of money. Stripe was doing famously well. We definitely never wanted to compete with Stripe. We were looking for other angles inside of payments that you could do things, like anti-fraud type stuff on payments.

Craig Cannon [17:12] – Basically another idea was sort of funded through YC, and then you just changed it, and you brought people on in the meantime?

Tikhon Bernstam [17:19] – Yes, that actually happens a surprising amount of the time.

Craig Cannon [17:24] – Yeah, and so by the end of the summer, you’re Heroku…

Tikhon Bernstam [17:28] – We’re Heroku for mobile, or we later switched to AWS for mobile. But basically like an AWS tailored for developers of mobile applications.

Craig Cannon [17:38] – Okay, and so why did you want to do it a second time, YC?

Tikhon Bernstam [17:43] – Oh, I mean, first off, the first time we would never have made it without YC, and that’s likely true the second time as well. Again, it’s largely for the same reasons as the first time. I still don’t think I know anything about startups. The more I do this, the more I’m convinced how little I know. Having that community of people who are highly motivated, having my first customers there right there in the batch, where I could say hey, you’re making a mobile app, use this. That’s really, really helpful. ‘Cause then you suddenly get feedback and then every dinner, every week, you go to them and say hey, I launched this new stuff, check this out. And then we bring our laptops and show what we’re doing and half the time it was stupid and people told us that and then we stopped. And the other half, they were like why not expand on this that way? And we were like, oh! ‘Cause that’s the problem you’re having in mobile, it’s not that.

Craig Cannon [18:38] – How did you start getting customers afterward? Because it makes sense when you’re surrounded by startups you can just grab a ton of people. I found Parse by going to hack-a-thons, because people were using it all the time. But how did you get people who would pay for it?

Tikhon Bernstam [18:51] – We didn’t have pricing for actually quite a while. It was a free service. We didn’t know how to do pricing properly back them. We were mostly focused on can we get, are we making something that’s making a few people really happy? So we would talk to the developers who had signed up and started using our beta. We started off just basically like sending data to the cloud to like S3, say, and by talking to the developers and seeing do we have a product where people are really disappointed if we went away? Which I think’s a really good test. If you can’t find some niche that is absolutely like fanatic about your product then you likely don’t have strong product-market fit. And most of the times that I’ve failed in startups was trying to push growth before I had product-market fit. We found this core segment of developers who really wanted to make native apps much, much faster, ranging from agencies who were making apps for everyone from Green Bay Packers to the Travel Channel, Weather Channel, and Cadillac, who actually all ended up using Parse. But we found people who really cared about this, who were like, this makes my life much, much better. Then we actually talked to them and used their feedback to make the product better in the ways that it should be better. We actually moved really fast. All four of us were coding. We worked and we kept kind of launching futures constantly, and we got the name thanks to Greg, the CTO of Stripe, ’cause I was over at their office, and I asked them what was your guys’ second choice? ‘Cause like Stripe’s an awesome name. And if Stripe didn’t work, we were going to be Parse. I was like, that’s available?

Craig Cannon [20:52] – They just gave it to you? For 10 bucks or something?

Tikhon Bernstam [20:55] – Well they didn’t own it. They had written a script to email all the owners of one-word English, one syllable domains like Stripe, Parse, things like that, and asked them like hey, we’re a couple people working on a startup, are you interested in selling? Most never replied, but Parse and Stripe owners did. I won’t say how much Stripe cost, but Parse was only $15k,

Craig Cannon [21:25] – That’s not bad. All things considered. Going through YC the second time, what became valuable to you? I assume it was different–well first of all, it was like way more people, right? In summer ’11, right? But what was valuable the second time?

Tikhon Bernstam [21:41] – Well again, having customers or users inside of the batch where I could talk to them, having that community, ’cause it actually gets quite lonely when you’re just sitting there coding 16 hours a day. Even if you’re with the same three people every day. It’s nice to have that community of other people who are doing the same thing. The partners smacked us around a bit like, no, no, you guys are going the wrong way. They were almost always right. And so we got a lot of good– it kept us honest about shipping every week, shipping new stuff, not being embarrassed to ship too early, which is really, really common, especially amongst developers who just want to keep coding instead of shipping. So the partners’ advice is priceless, and then at Demo Day, 1000X fundraising was easier. I can’t imagine the fundraising minus the YC Demo Day. We were wildly oversubscribed, but we only took $1.5 (million), we stopped there for our seed round. We really were able to get some of the best investors and some of the most famous Valley type people, and it was really awesome.

Craig Cannon [23:00] – Yeah, and then it was like two years and then Facebook acquired you guys?

Tikhon Bernstam [23:04] – Yeah, so we started Parse in 2011, Facebook acquired the company in 2013. We actually had M&A discussions with several other companies along the way, like in 2012 and early 2013, but we actually raised our Series A right after Demo Day, we basically went and started to raise our A, because we actually thought we already had basically enough traction to start doing that. It was a fairly quick process. We raised 5.5 million from Ignition, so in total we raised 7 million, and then we started hiring, and we were able to recruit some just amazing developers. A lot of people off the Google Search Team, for instance. People, a lot of developers were passionate about the problem and liked working on the tools that solve their own problems, so that was helpful. Being such a developer focused brand also helped us, and YC was great for that. Being able to post hey, we’re hiring, on Hacker News back then.

Craig Cannon [24:14] – ‘Cause that’s always a question I wonder, because people are looking for those jobs, but you guys are also just looking to find people. What were the effective strategies for finding developers to come work on Parse? You just made a cool product or what?

Tikhon Bernstam [24:29] – We often would talk to a lot of our actually users. There were some amazing engineers who were using Parse, so even if we couldn’t hire them, we could talk to their friends, but also once we got a couple of amazing hires out of the Google Search Team, we just kept asking every… We asked every single hire who’s the best person you worked with? Who would you co-found a company with? Then we would just go to that person and ask the same question to that person, and even if I couldn’t get that individual, you just kind of kept moving down the tree. So that actually worked really, really well. We were able to hire a lot of friends or people we’d worked with previously. I think Kevin had worked with several of the Google folks previously. They already knew and trusted us, and we had kind of a good, friendly group that wasn’t all just kind of strangers. Eventually we ended up hiring everyone from… We hired… Some of our best people were actually users of Parse. The guy who ran all our community and such, was actually making Parse apps, he had just graduated, and that was actually another great strategy. Hacker News jobs post was a really good one. I guess that alone actually worked really, really well.

Craig Cannon [25:55] – Now Parse is shut down, right?

Tikhon Bernstam [25:58] – It is. Just the other day, I think the end of January, Facebook finally… Facebook open sourced the entire project and now there’re actually little companies that provide Parse hosting, much like…

Craig Cannon [26:11] – Oh man, really?

Tikhon Bernstam [26:13] – Yeah.

Craig Cannon [26:15] – It’s AWS for Parse?

Tikhon Bernstam [26:17] – Well ’cause there’s actually very little profitable businesses, all those folks who host WordPress for businesses, ’cause WordPress is an open source project, now Parse is, too. There are people who just took over the hosting part of it. The Parse team, to their credit, they kept building and working and adding features to the open source project, it’s really, really good. There’s a lot of people who switched over. We actually helped everyone migrate over from using our servers to the open source project and how to do that. The company did an amazing job with that. I think it was one of the best shutdowns of a company that I’ve seen. So often users are really angry and this wasn’t really graceful, long, over a year notice, that we were going to be shut down, open sourced it with migration tools. Too many startups end up just kind of closing.

Craig Cannon [27:14] – Oh, for sure, that’s happened. Do you think there’s another Parse coming down the pipeline? Like someone’s going to try it again?

Tikhon Bernstam [27:20] – Well Firebase did, they were actually in our batch. They were a different company at the time. They were doing basically chat for every website. You know how Twitch has a little chat box, they were trying to do that for everyone. And they ended up pivoting later into doing basically Parse again. Then Google bought them. So Google has their own Parse and Facebook had one.

Craig Cannon [27:48] – You can use Firebase, I guess, or Parse hosted.

Tikhon Bernstam [27:52] – Firebase is great, actually.

Craig Cannon [27:53] – Yeah. You didn’t go to Facebook, right? What happens after Parse? What did you do?

Tikhon Bernstam [28:00] – I was like fairly burned out after. I had no break between Scribd and Parse. Literally it was the next day I just started working again. After many, many years, I guess, I’d always wanted to travel, I’d never done so, so I’ve spent about two years traveling. Backpacked South America without a computer which was great.

Craig Cannon [28:25] – What was that like?

Tikhon Bernstam [28:26] – Yeah, and lived in Europe for a while, did a lot of speaking at conferences. That sort of thing. I tried to learn more about the international startup scene. It feels like every city internationally, or a lot of them, think like “How do we build the Silicon Valley of South America?” Or “How do we build the Silicon Valley of Peru?” It’s like, that’s not really the right approach, I don’t think. They’d be better off focusing on being really friendly in one niche. Like we’re going to be really friendly to Bitcoin people. We’re going to make the banks be friendly to bitcoin startups. You might get more people going there and not dealing with all the difficulty that drones have here, like with all the regulations. You could see like a Drone Valley and a Bitcoin Valley,

Craig Cannon [29:14] – Shhh! Stop. No, I completely agree with you. There’s no reason why all the pharmaceutical startups couldn’t be in Germany or something. What did you learn from, you’re speaking at all these conferences and stuff, and you eventually stop doing all that speaking, right? Did you pick anything up along the way?

Tikhon Bernstam [29:35] – Yeah, the number one question that I got was how do we raise money? How did you do that? It’s really unfortunate because much like the angel ecosystem that had to build in Silicon Valley, it took a long time. Back in even ’06 there were not that many investors and terms were not good. I mean nowadays the typical seed run out of YC is higher than the Series A post money that Sequoia did in Dropbox and Airbnb. So you have to envision this client was completely different. There was not this flood of capital chasing every good startup. So that’s kind of how it feels in most international areas.

Craig Cannon [30:22] – What do you tell people? Come out here?

Tikhon Bernstam [30:25] – What I was telling folks… Yeah, I said if you can, raising out in the Valley is… For instance, Algolia was in Paris the same time I was there and they ended up moving here. Front App from Paris also moved here, Front App I’m an investor in. Yhe honest truth is like I still think living in a startup here is very beneficial. It has certain downsides, it’s very expensive here. But in terms of fundraising, that certainly helps. Investors are largely averse to non-Delaware C Corps, which a lot of these companies outside of the U.S. are. For instance I, as an investor now in like over 60 or 70 companies, I don’t understand the legal ramifications of investing in a company in South America.

Craig Cannon [31:17] – You basically just avoid them?

Tikhon Bernstam [31:19] – Well, it would be a large legal headache to go have lawyers review this and see… With the Delaware C Corp stuff, you already have that figured out. I actually just encourage folks to apply to Y Combinator, or if they can’t get into YC, apply to 500 Startups, whatever. But these programs are so perfect for people internationally and I love what Stripe is doing with Atlas. Like helping people. They’re letting anyone be an entrepreneur anywhere around the world. That is such an awesome ambitious goal and solving that problem. I think things have gotten much, much better than back when I was hearing that in 2014. Largely due to efforts like that. YC does a really good job of taking– there’s a great company called Razorpay that I’m invested in, that’s kind of like Stripe for India. And YC helped them with the whole transition.

Craig Cannon [32:16] – That’s super valuable. That’s probably a good transition. You said like 60 angel investments at this point? Do you have a particular strategy, an area you’re focusing on? How’s it going?

Tikhon Bernstam [32:27] – Yeah, so far I’ve invested in Cruise, which just sold to GM, Gusto, LendUp, Checkr, Optimizely, Reddit, things like that. For me in the early stage investing, I really, the first thing I’ll look for at least is like is there a large market, because without a market, it’s really hard to be successful even if you build a great product. Is there a large market, and are these founders a really relentless team that is determined? I also have this joke question I sometimes, I don’t actually ask founders this, but I’m like hey, there’s this conference in the UAE, and someone just dropped out, do you want to go give a talk? It’ll take you a week. They were kind of like the conferences that I was actually going to.

Craig Cannon [33:30] – I guess you weren’t a founder at the time, but still.

Tikhon Bernstam [33:32] – But like often if founders are willing to take long distractions in doing things that are not relevant to their companies, that’s a bad sign. You learn to identify the folks who are relentless.

Craig Cannon [33:46] – What are other signals you see in people?

Tikhon Bernstam [33:49] – I actually like solo founders. I know that’s an unconventional thing in the Valley, but Drew was initially a solo founder with Dropbox. And a lot of other great companies were as well, or had one founder who really drove the project through. I think people who are willing to go through it solo, it’s already hard enough. People who can persevere for a year or longer by themselves, I really admire that. You can kind of… References in some sense. If you ask people who are the best people you ever worked with and they name someone, that’s probably someone that’s pretty good.

Craig Cannon [34:28] – As a technical person, do you vet the product, the tech side of it at all?

Tikhon Bernstam [34:32] – Yes, but I kind of believe, and I’m stealing a lot of this from Marc Andreessen, but a really strong team in a really great market will eventually build the right product. They’ll figure it out because the market will drag the right product out of them, whether it’s kicking and screaming. Like Docker came out of dotCloud. There’s a lot of those things. Twitch came out of had this one vertical of gaming that was really growing and they actually ditched the rest of it and just did the gaming vertical. Often good growing markets will actually eventually steer you in the right direction.

Craig Cannon [35:16] – As an investor, have you been burned yet? I mean you’ve only been doing it for a couple years, right? Are there any hard lessons you’ve had to learn?

Tikhon Bernstam [35:26] – I think people complain about valuations every single year. I’ve heard this since 2005. I remember when Optimizely’s seed round was at four. People were like that’s so expensive, I’m never going to pay that. Like, now it’s a multi-billion dollar company. I don’t listen to that anymore. I think a lot of the best deals can feel expensive. As Gary Tan recently said, “Investors make their money basically on being in the deal everyone wants to be in or being in the deal nobody wants to be in.” I think that’s largely true. If you look at Dropbox and Airbnb, they took over a year to go from Demo Day to their Series A. You don’t have to actually rush into things at Demo Day. There’s a lot of room in the post-seed world where you can invest maybe at a slightly higher valuation but before the Series A. I’ve done a lot of that as well.

Craig Cannon [36:26] – How are you finding those deals?

Tikhon Bernstam [36:28] – Often they’re products I use or things that actually solve problems that I had. Like in terms of the areas I focus on, I should mention, a lot of B2B and enterprise, a lot of autonomous vehicle tech, much like Cruise. A lot in marketplaces. I’m still bullish on… I don’t think there are enough marketplaces for all the different verticals yet.

Craig Cannon [36:55] – For example?

Tikhon Bernstam [36:58] – Okay, I’m an investor in Quartzy, which is a great startup, just did their Series B, and they are bringing for life science laboratories like at universities or say at Genentech, they’re uniting, they’re fixing the whole inventory system problem, because most folks just have a chalkboard with writing down what they have, or whatever, or some paper binder. They are uniting the life science labs with the suppliers. That’s very valuable. Much like Airbnb. A lot of people said why do we need Airbnb? We already have Craigslist. But often these experiences are much better when they’re custom tailored.

Craig Cannon [37:40] – Yeah, I think it was Chris Dixon who made that graphic a while ago of breaking Craigslist out into all these billion dollar companies, something like that. I think it’s true, there so many companies. We see it with YC. Jared actually wrote a post about it last year, the startup zeitgeist, all the trends. People often don’t pay attention to the unsexy but giant markets.

Tikhon Bernstam [38:03] – Yeah, and those are often the most interesting ones in terms of investments. And so a lot of my investments didn’t seem sexy at the time, but worked out.

Craig Cannon [38:16] – Where are the markets you think that people, in addition to those marketplaces, you think people ought to be paying attention to? Because right now there’s a lot of noise around for example like VR or machine learning AI stuff. Do you think that those things are worth startups going for right now or are there other things that you think are good opportunities?

Tikhon Bernstam [38:40] – I actually really like existing huge markets that are growing. Not everyone agrees with this. Some people say look for a small market and expand it. I think in Zero to One, Peter Thiel kind of talks about this. That certainly can work. Dropbox is a great example. A lot of famous investors I won’t name passed on Dropbox at a $2 million valuation, saying that space it too crowded. But when you see a huge market that’s fragmented with lots of companies, many were public already doing the same kind of thing just badly, if you can make a 10x better product, you can really win. And there’s room for lots of winners, right? Box is public, they started roughly in the same era. Dropbox is doing really, really well, I think they just hit a billion dollar run rate or something. And there’s room for lots of winners in huge markets. I’m guilty of following trends and saying YouTube is hot, let’s do YouTube for X.

Craig Cannon [39:42] – Like, it’s Cruise for X.

Tikhon Bernstam [39:44] – But don’t do that. Because that’s rarely how the best companies were formed. When Stripe started, everyone said why another payments company? Everyone already has a payment system. But they made a 10x better product that developers loved. They found a niche of people who loved what they were doing. Developer friendly and they assign merchant IDs on the fly. That was huge. Dropbox had the same kind of reaction. Airbnb, I mean, as … posted, all the famous investors passed on that, because they didn’t necessarily understand the problem. Like, why would I rent out the living room of my brownstone in Manhattan?

Craig Cannon [40:24] – Do you think it is an education problem? Because that happens with YC a lot. Like there’re people that are seemingly entering a crowded space, but in actuality they are going to be the winner. If you are the founder of that company, how do you educate people if you do in fact need money from them? Or is it just about education?

Tikhon Bernstam [40:45] – I think YC has actually done a really admirable job in picking relentless individuals like the Airbnb team, which was a team that was never going to give up. Even when they’re not sure about the market or the idea, they’ll bet on really amazing people who are relentless. I think that’s one of the genius parts of YC and why YC has worked out so well. I suspect nowadays I think you have to ship at least part of that 10x better product, just because the application… Nowadays I don’t even think I would get into YC. If I applied right now. ‘Cause it’s much harder to get into. There’s much more applications now than there were back then. I think if you can show some traction or some product-market fit in a large market, that’s always good, but in terms… I just think investors are wrong on this. A great way to find startup ideas is to look at the Inc. 500 magazine. And it shows the fastest growing startups in terms of like revenue, like say year over year, or over three years. And then just look at those markets, and see if you have founder market fit in any of them. Do you know any of those markets really well? And then if you have a secret, as Peter Thiel would say, like if you know something about… Much like the way that Drew knew that we can make a much better backup cloud storage type product than this, and it’s ridiculous people are using USB drives for backing up stuff. That’s because all the products suck, even though there are public companies in the space. So he just sat down and built a 10x better product. I really like looking at the software that you use every day and seeing which ones annoy you, which ones make you cringe? I always write those down. Because if you’re using that it means it’s probably one of the better products or it’s probably what people are forced to use, and if you can build a 10x better product, people will switch.

Craig Cannon [42:46] – Totally. I think paying attention to market size makes a huge difference. I’m still shocked that there aren’t more, not Cruise look-alikes, but self-driving in general. Like personal transportation is so massive that there’s room for like 10 of them.

Tikhon Bernstam [43:00] – Yeah Sebasitian Thrun recently was quoted as saying roughly a self driving engineer type person is worth roughly like $10 million. In an acquisition, in like M&A. Cruise was 50 people and it sold for a billion. But keep in mind, GM’s stock price actually went up on the news. The market actually viewed this positively. ‘Cause GM wisely noted that they were going to either become a software and technology company or become a dinosaur relic. I believe this is true of all companies. All companies will either become software companies or they’re going to die.

Craig Cannon [43:41] – As an investor now, do you have a fear of choosing the right idea if you were to do a start up again? Are you scared of doing one or would you want to?

Tikhon Bernstam [43:52] – That’s such a great question. Yes, absolutely. I actually think repeat founders are actually at a disadvantage in many ways, because you feel this pressure to come up with some genius idea, when usually the best start ups are not a genius idea. They’re not some brilliant flash in the pan. It’s just doing something that… Facebook. MySpace already existed. Friendster already existed. But Mark just really nailed it. He started with colleges, he built something awesome that people loved. I think often doing things better is not a bad strategy. But as a second time or third time founder, I feel a pressure to– it’s like the sophomore slump that Richard Feinman talks about in physics, where he wasn’t allowed to work on small ideas anymore, he had to work on some gigantic thing, but all the best discoveries in physics came from like the “That doesn’t seem right, it’s kind of odd.” Maxwell’s equations say that the speed of light’s always constant…

Craig Cannon [44:51] – What do you do to actively combat that in your brain and the people around you as well?

Tikhon Bernstam [44:55] – I think you have to I almost want to say not work in secret, but I wouldn’t even talk too much about my idea, probably, because it would sound so stupid. But the best companies do sound kind of dumb. Also being willing to experiment with lots of dumb ideas and maybe focusing on markets where there already is a so-called winner. I think there’s room for lots of winners in lots of markets.

Craig Cannon [45:29] – It’s really tough, too, because I know tons of successful people that work on all these little side projects, but are embarrassed to put them out, because they’re just like stupid little toys, and it would be inappropriate, or just not cool for them to put out some Chrome extension that they just built for themselves.

Tikhon Bernstam [45:46] – This actually gets worse. Then you kind of get tempted to take the money people offer you in fundraising, you’re like okay I’ll take a couple million dollars just to like spread the seed around, but then all of the sudden, you’re stuck in whatever market and product that you’ve picked, and you’ve probably picked wrong, just like I did every time. Knowing that you’ve picked wrong, you kind of get stuck and it’s too awkward to tell your investors that you’re changing and pivoting, and so I think second time founders raise money too fast. I am not going to raise money until I have strong product-market fit and I know what I’m doing. A friend of mine who actually raised money too early was just lamenting how that probably killed his second company.

Craig Cannon [46:30] – What about those people who are successful and then decide that they want to create a lab where they have seven startups to their own?

Tikhon Bernstam [46:39] – Yeah, that’s really interesting, like Max Levchin’s done really well with that, like he had Yelp come out of that, Slide, which he sold to Google. I think Affirm, and then there’s some great ones. That model’s really worked out for him.

Craig Cannon [46:53] – I feel like he’s an outlier.

Tikhon Bernstam [46:55] – Yes.

Craig Cannon [46:58] – But I don’t know, I could be wrong.

Tikhon Bernstam [47:00] – The other labs I don’t think I’ve seen be as successful, like I think Y Combinator used to not like being called an incubator, because most incubators haven’t really worked out that well, Max’s aside. I think it’s hard to really honestly focus on that many different ideas that are all in progress. Landing pages are one thing, ’cause they’re all the same template, you’re modifying text, effectively. Like, often. But building, having teams that are working on six, seven different projects, it’s hard to figure out that really strong product market-fit, which is what most startups fail. Most startups fail because they never get strong product-market fit.

Craig Cannon [47:43] – Agreed. So kind of wrapping up. If you as a multi-time founder, if you were to give some advice to someone who was starting their first company, what would you tell them in terms of don’t spend time doing this, spend time doing that?

Tikhon Bernstam [47:55] – This is to a first-time founder?

Craig Cannon [47:57] – Why don’t we do both? So first time and second time.

Tikhon Bernstam [48:02] – Everything that’s not working on your product and finding product market fit is a waste of time. That’s the only thing that you should be doing. Don’t worry about fundraising, don’t worry about everything else. The folks who do fundraise too early, typically get into real trouble, because they never are able to get to product-market fit. Really focus on finding that huge market. Go to that Inc. 5000 list, find the fastest growing companies, find a market that you know really, really well, better than anyone else. Have founder-market fit. And make it a big one that’s growing, and then focus on building a 10x better product in that space.

Craig Cannon [48:38] – Alright thanks for listening. As always, please rate and subscribe to the show. If you want to read the transcript or check out the video they’re both at Okay, see you next time.


  • Y Combinator

    Y Combinator created a new model for funding early stage startups. Twice a year we invest a small amount of money ($150k) in a large number of startups (recently 200). The startups move to Silicon