EP 36: Fivetran CEO George Fraser and Young VCs Launching Funds

In episode 36 Logan and Zach talk with Fivetran Co-Founder and CEO George Fraser about how entrepreneurship is evolving and whether the era of iterating on product until you find your niche still exists. Plus investors leaving major firms to start their own funds, how to choose a VC firm, and figuring out the right time to sell your company. (0:00) Intro (0:53) Investors starting their own funds (10:05) Tiger (14:16) Choosing a VC firm (22:07) Facebook’s Yahoo offer (25:04) Welcome George Fraser (30:02) Plaid comparison (39:01) Salesforce data (42:32) Finding product/market fit (46:52) Era of iterating to success (51:55) What’s driving stack changes? (56:29) Mixing family and business (1:02:17) Academia to entrepreneurship (1:00:28) Lawrenceville Boarding School (1:05:56) Acquisition lessons learned (1:10:07) Closing Mixed and edited: Justin Hrabovsky Produced: Andrew Nadeau and Rashad Assir Executive Producer: Josh Machiz Music: Griff Lawson 🎙 Listen to the show Apple Podcasts: https://podcasts.apple.com/us/podcast... Spotify: https://open.spotify.com/show/5WqBqDb... Google Podcasts: https://podcasts.google.com/feed/aHR0... Follow on Socials 📸 Instagram - https://instagram.com/cartoonavatars 🐦 Twitter - https://twitter.com/cartoonavatars 🎬 Clips on TikTok - https://www.tiktok.com/@cartoonavatars About the Show A Saturday morning podcast hosted by Logan Bartlett (Partner and Managing Director at Redpoint Ventures) covering the tech news with his friends and other people with industry expertise. Executive Producer: Rashad Assir Producer: Leah Clapper Mixing and editing: Justin Hrabovsky Check out Unsupervised Learning, Redpoint's AI Podcast: https://www.youtube.com/@UCUl-s_Vp-Kkk_XVyDylNwLA 🎥 Subscribe on YouTube: https://www.youtube.com/channel/UCugS0jD5IAdoqzjaNYzns7w?sub_confirmation=1 Follow on Socials 📸 Instagram - https://www.instagram.com/theloganbartlettshow 📱 X - https://twitter.com/loganbartshow 🎬 Clips on TikTok - https://www.tiktok.com/@theloganbartlettshow About the Show Logan Bartlett is a Software Investor at Redpoint Ventures - a Silicon Valley-based VC with $6B AUM and investments in Snowflake, DraftKings, Twilio, and Netflix. In each episode of The Logan Bartlett Show, we sit down with the people behind today’s most important startups and extract the tactics, lessons, and frameworks they’ve learned the hard way. Conversations span hiring to GTM, product, growth, fundraising and everything in between - collectively forming the ultimate playbook to make you a better CEO, investor or board member. Tap follow and enable notifications to stay ahead of the game.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:05Logan Bartlett?

  Welcome to the 36th episode of Cartoon Avatars. I'm your host, Logan Bartlett, and what you're going to hear on this episode is a conversation Zach Weinberg and I had earlier this week about different VCs spinning out at established firms and starting their own, which happened a few times, perhaps coincidentally this week. So we go into that. And then in the middle of that conversation, George Fraser joins. George is the founder and CEO of Five Tran, which is a $5.6 billion private company. So George joins in the middle of that conversation due to some scheduling things. Zach and I were running along and he dives right in to the dialogue with us. So it's a really fun conversation.

• 0:47Logan Bartlett?

  George is a really thoughtful, entertaining, interesting guy, and so it was a fun one to do. Look forward to you hearing it now. I guess we're recording this on Tuesday night, October 4th. So one of the things I thought was interesting this week, and it's coincidental timing, I assume, or end of quarter for the hedge fund reporting, I guess, but Matt Mazzio, who I know a little bit from COTU, it was announced he was leaving to start his own fund. John Curtis, who has been one of the more prolific private investors over the course of the last two years at Tiger, it was announced that he was leaving to start his own fund. And then I think that was yesterday, so I assume both of those came out after 9:30

• 1:27Logan Bartlett?

  reporting and reporters got their hand on Tiger and COTU's letters for the quarter. But then also Sarah Guo, who I know from Greylock pretty well, closed her fund, I guess it was announced today, and she had been at Greylock for a number of years. So Zach, we were just talking before we hopped on, but interesting theme in general that I expect we're going to see a lot more of. I don't want to speak to any of these situations individually because I assume we both know or know of all these people, but it is interesting. All these people are 30s, mid 30s. I think Mazzio just turned 40 in that range. They had been at their firms a little while, but not, you know, 15 years or whatever.

• 2:11Logan Bartlett?

  So kind of interesting to have all this fall at the same time.

• 2:14Speaker 01

  I'm gonna, I'm gonna do something that I despise when people do it, I'm just gonna do it anyway, which is like speculate on something I literally know absolutely nothing about.

• 2:23Logan Bartlett?

  We can keep it in generalities, right? More than any of these specific people. Like, I, I think both of us think this is gonna happen more and more outside of these specific people and the reasons, I'm sure each of them have their own reasons for leaving, but I do think this is gonna happen a lot.

• 2:37Speaker 01

  Yeah, maybe they do, maybe they're similar. I actually, right as you mentioned, so I didn't, I hadn't been paying attention, so I didn't hear about this, but I went and quickly looked how long each of them had been at those funds. Uh, I think Sarah was like a few years, uh, and I've actually, I've met her before, but don't, don't know her well. John, it looks like was there for five. Uh, and then Matt, I just pulled it up here, joined in 2018.

• 3:01Logan Bartlett?

  Yeah.

• 3:01Speaker 01

  So there's like, it all kinda like joined like around the same time, basically, it seems, like sometime late 2018, 2017, 2019 time.

• 3:08Logan Bartlett?

  I think Sarah's been there since maybe 2014 or 13.

• 3:12Speaker 01

  Okay, she's a little longer.

• 3:13Logan Bartlett?

  Yeah.

• 3:13Speaker 01

  Yeah, I mean, it's interesting, right? Like if you join one of these giant funds or later-stage funds, uh, in that like 2017-plus timeframe or 2018-plus timeframe, and so you're writing checks in those funds at inflated prices, and you probably have seen little to no liquidity on any of those, would be my guess, uh, or very limited, and, you know, you've got like, what is the unvested, vested carry like actually really worth?

• 3:41Logan Bartlett?

  Yeah.

• 3:41Speaker 01

  Right? If you thought it was gonna be a 3X fund, and now stuff's getting marked down, and it looks like it's, I don't know, one and a half or something like that, so all of a sudden, it's like your stock options are underwater, right? It's the same idea. So if you're leaving, you're not leaving as much on the table, so it makes it easier to go.

• 3:56Logan Bartlett?

  For the most part, the way, I mean, I guess just to give a little bit of inside baseball into how this stuff works, there's typically a management company at the top of these firms that have essentially a profit share of the net income, right? And how that gets divvied up or whatever is different for each firm individually. But then for the most part, and so that's kinda like, hey, Zach's the owner of his company, and so Zach gets the spoils every quarter, or every year gets to pay him out the dividend that comes with it. But then for the most part, like, everyone else that's not in that structure is an employee, and they have a salary.

• 4:29Speaker 01

  Like a hired gun with a... You get a cut, you get typically a predetermined cut of that profit share on a per-fund basis that you negotiate per-fund, right?

• 4:42Logan Bartlett?

  Well, even a lot of these people just have salaries, right? It's just a plu-, it's like, you know, rank-and-file executives. Maybe you have, you get some percentage of it if you're that senior.

• 4:52Speaker 01

  Well, at this, at this level, those three in particular, I would assume have carry.

• 4:57Logan Bartlett?

  Oh, sorry, carry. Sorry, I was talking, I was, sorry, we were conflating. So I was talking just about the income stream that comes from management fees.

• 5:07Speaker 01

  Oh, just the management fee. Yeah, yeah, that just pays your, your day-to-day salary, right?

• 5:12Logan Bartlett?

  Yeah, exactly. So most of these people are just getting paid salary. But then, yes, to your point, there's some carried interest, which is essentially for rank-and-file employees, you could think of as stock options, right? Essentially, and this is your interest in all the investments that are being made.

• 5:28Speaker 01

  Yeah, it's your cut of the profit share, right? So the fund itself, like, the management company gets typically, it depends on the fund, but typically 20% of the profit of a fund, and sometimes it's higher. And then how they divvy that profit share up amongst the partners and junior employees is basically up to the owners, right? They decide this, and they tend to negotiate it upfront per-fund, right? So like, you sign up as, you know, one of the partners or rank-and-file employees. Everyone's a little different, but you tend to only get it like one fund at a time. You sign up on a per-fund basis, and so you're constantly having to prove your worth, because if you don't do well, you don't, you're not guaranteed the carry on the next fund, or the next fund, or the next fund.

• 6:13Speaker 01

  And so, you know, when you walk from one of these funds, you're basically giving up the unvested carry, like, the stuff that hasn't been paid out on the fund you've been a part of, and then what you would have made in a future fund theoretically, right?

• 6:26Logan Bartlett?

  And so, so the way this actually works is like, so let's say Logan joined a fund in 2015, right? And I had $100,000 of carry in that fund, right? And that's a, that, that, when you talk about it in that dollar amount, that typically means, hey, if this fund 2X's, and everyone just sort of uses 2X as like the framework around it, if this fund 2X's, then I should, my interest should be from 100 to 200K, right? And you hope the fund's going to 10X, and so actually, that will be worth, whatever, 900K and not, not, not just the, or a million in that case, right? And so that's the way you think about it. And then there is a vest. It could be everything

• 7:06Logan Bartlett?

  from two years, four years, five years, 10 years in a lot of cases with these GPs because they basically want all these things to stack. And so think about it as an employee that your vesting schedule lasts for 10 years, right? In a lot of the cases. And so basically, I'm only getting 10 grand a year of that amount. And then if I leave before that, I'm giving that up. And so now it stacks, we go raise another fund and I get this, this time I negotiate with Zach and I get 150K in that fund or 200K in that fund or whatever it is. So all of them stack, but then the vesting restarts, right?

• 7:45Speaker 01

  Yeah, you, you earn into it per fund and you know, sometimes the number, you know, that is done usually as a percentage of the total, right? It's necessarily a dollar, but same idea.

• 7:55Logan Bartlett?

  Which is actually a funny thing, like funds will often try to go back and forth between percentage and dollar, right? Because percentage scales linearly and dollar doesn't. You know, dollar is the funds get bigger, the dollar amount you can grow, not at the same rate as the percentage.

• 8:13Speaker 01

  Yeah. And there's, I mean, you can kind of basically do whatever you'd like within the rules of the law. You know, it's your fund. You can structure the profit sharing how you'd like. I think for many of these, and again, I don't know the situation in any of these.

• 8:25Logan Bartlett?

  Yeah, yeah, we're talking generalities.

• 8:27Speaker 01

  But like, look, if you're one of the newer partners, you're younger, you're definitely not part of that management company. So you are effectively like a hired gun partner, you know, and you've only been doing it for four to five years. The unvested, right? The thing, the profit share you haven't earned into yet that you thought you had is probably a lot lower now than it was just a year ago. So, you know, the amount of money you are theoretically walking away from is cut quite a bit. Plus, I have to guess, again, I don't know these funds,

• 8:58Logan Bartlett?

  Yeah.

• 8:58Speaker 01

  That now, when they're negotiating for the next fund, you know, if you haven't been the single best person, you're gonna get chopped, right? Like, you may not get that carry bonus, or that carry increase, rather, that you expected, because things are gonna get a little tighter, and fund returns are gonna get a little tighter. And I think, you know, the management companies may be a little less generous with some of the junior employees, so.

• 9:21Logan Bartlett?

  Or the other way, right? Or they say, hey, that fund sucked. And, you know, you told me I had this, but now I don't have that, so I deserve 2X that, 3X that. And there's, therein lies the tension.

• 9:35Speaker 01

  It's funny, right? It's actually on both ends of the spectrum. If it's like, you're really good, and you don't believe in the rest of the fund partner team, time to go, right? Do your own thing, control your own destiny. It actually is also the other way around, too, which is if you're not that good, and the fund is kinda going like, I don't think you're gonna make the senior level of partner, your carry's gonna be fixed, or even lower on the next one. You also go, so you can read these reports, and it's kind of like, was that the really good one? Or the really bad one? But it's, from the outside, it's impossible to know.

• 10:05Logan Bartlett?

  Well, I think the most interesting one here is, I mean, Tiger, like, so many things. I mean, John was the point person for hundreds of portfolio companies within Tiger, right? He was their line and all of that. And so, that's an interesting dynamic as well. I mean, I had this experience when I left Battery is, when you, and I had, I don't know, eight investments or something, right? And when you leave a fund, like, your relationship with the company, I mean, you build great relationships, you're on the board, you have helped them talk to those CEOs every day, right? Or once a week, or once a month, or once a quarter, whatever it is. But you walk away from.

• 10:47Logan Bartlett?

  All those relationships, pretty much. And like, you still have the friendships in place. I'm still friends with a lot of people I invested in, but you don't have the formality of engaging with them about their business anymore, right? And it's super tough and to do that, to walk away from it, if you're like, an empathetic human being, even in the tiniest way, when you have like, three, four, five, six companies you've invested in, I can only imagine 100, 200, 300, obviously Tiger's relationship with their companies is different than the average VC, but I, there's probably a lot of VCs or portfolio companies right now, just being like, wait, who do I call?

• 11:26Speaker 01

  I mean, I, there, you know, Tiger tended historically to take the approach of, we're not going to take board seats, we're mostly just capital, you know, we're not working deeply with each of these companies and because of that, we can move really fast and do, you know, so I would, I don't know what level of relationship typically the Tiger partners would have with their, with their companies.

• 11:46Logan Bartlett?

  You still need someone to call, like, hey, we need these docs signed, or whatever, right? You know, or, hey, we're raising a round.

• 11:54Speaker 01

  As a founder, I, it's funny, I always appreciated, you know, and I, we never raised money from Tiger and as to disclose, I am an LP in one of their funds.

• 12:03Logan Bartlett?

  I didn't know.

• 12:05Speaker 01

  Yeah, yeah, yeah, one of the, a few years ago, one of their funds. And so again, like, I, I don't really know anything specifically, but I always appreciated the, the Tiger approach as a founder, which was like, they're not, they're not there trying to bullshit you on like, we're going to do all these strategic things, we're going to help you with XYZ on the board, or introduce you, or, they were kind of like, we're going to give you a bunch of money at a reasonably high price, and we're going to do our homework and do it really fast. And then if you need more money and the business is doing well, we will give it to you. And if you don't, it's great.

• 12:37Logan Bartlett?

  By the way, for someone like you, repeat founder, like, great product, hey, leave me alone and give me capital at as expensive a price, right? I think it's, you know, the, the LP side of things is, hey, does that math work from a return standpoint, as well as the other founders that like, don't, haven't sold two businesses for whatever, to, to successful acquirers out there, right? Though, that's where it gets a little more nebulous, I would say, but for you, I think it's an amazing product. I get it for honestly, all founders, I get it. It's not the path I would choose universally for people, but.

• 13:12Speaker 01

  Yeah, I mean, look, there are, you know, even if you haven't done this before and you're

• 13:17Speaker 01

  Uncomfortable in certain situations because you've never seen them. There are other ways, frankly, potentially better ways of getting strategic advice and unfiltered opinions around the table through angel investors, through your strategics that, you know, maybe don't have to write the lead check, but write a big enough check from their own personal dollars or have a big enough stake that it matters. And, you know, in some way, it's actually kind of better because you can triangulate across multiple opinions as opposed to just who happened to write the big check in your round. So, you know, I'm always conflicted in this because I've always had very good investors I really liked. You know, we had Josh Koppelman from First Round and Chris Fralich from First Round in our first company, and they were great. We had Krishna from Google Ventures in the second round, Josh also in Flatiron as well.

• 14:02Speaker 01

  So I always really liked them. But generally speaking, like, you know, did the VCs really help in the business in any material way that like you kind of probably could have just done on your own? I think it's overblown.

• 14:17Logan Bartlett?

  Yeah, no, I mean, that was definitely the whole value prop of Tiger, right? Like, I think it totally makes sense. And the help thing, it's one of those things that like, it's almost like selling your house in some ways where like, sure, you could pick the worst real estate agent that's going to charge you less. You could do that, right? Like, there's probably someone, you might not even need a real estate agent, but like, when push comes to shove, you would probably prefer the better one just because the delta of even if it is helping you see around some slight corner and help with some higher. Like, and again, this is the whole nebulous thing of like arguing about VCs as partners

• 14:58Logan Bartlett?

  versus not is like, I agree with you, probably 75% at best are like, maybe neutral. And, you know, there's probably the bottom 25% of that 75 that are bad, but there are, I have seen a lot of people be super helpful if only for, you know, introductions or strategic advice or console or therapy session, like all that stuff, right?

• 15:21Speaker 01

  Yeah, maybe, maybe they change the final outcome of the business by 10 to 20%, maybe.

• 15:28Speaker 01

  Maybe. Yeah. And that's actually, I mean, usually when founders ask me some version of this question of like, hey, I've got two term sheets or three term sheets and they're within, you know, these not like, how do you think about it? And usually what I'd say is if it's within 10%, like if the price is within 10%, then let's go to the other factors, right? Let's start to look at, you know, the track record of that fund supporting founders in a bad time, does, you know, so on and so forth.

• 15:55Logan Bartlett?

  And it probably depends on stage too. Like I would probably give even 20% at the series A or the seed or something, right? Versus the series D or E or whatever, I probably wouldn't even give five or whatever, eight percent, whatever the number is.

• 16:10Speaker 01

  Right. When the business is more mature, you don't need the help. Yeah, there's some window where like the help is more useful. I think that's right. It's, you know, it's, it's somewhere in a, in a range, in a delta. Unless, you know, the, obviously the one big red flag is like, is this a partner that has had a history of doing the wrong thing?

• 16:30Logan Bartlett?

  Yes.

• 16:30Speaker 01

  You know, when something goes wrong. The thing I will say about, everyone's always talks about that situation, like how, what did the partner do in a really bad setup? And the reality is, we haven't really had like too many bad situations in venture for like 10 years. So nobody has a track record of doing the wrong thing because there hadn't really been the opportunity to do the wrong thing. Cause like the market was just going crazy and now we're going to see it.

• 16:52Logan Bartlett?

  Or the people who have, are like really bad human beings, like flawed, fundamentally fucked up people. Like if you've done something that over this, like, you know, whatever, gumdrops and rainbows and a period of time of like a 15 year bull market. Like if you've done shitty things, then at your core, you're probably like a really, really bad human being. But like, now we're entering the points of like incentives just not being aligned. You could have two good people whose like incentives aren't aligned. And, you know, I'm sure there's going to be a lot of versions of like, well, we were doing this recap round and he said he would support it, but you know, at the last minute, couldn't get his

• 17:35Logan Bartlett?

  partners over the line and, and we were left in a lurch. And so we couldn't close the round. And is that actually? A bad person, you don't know what actually happened. Maybe the, maybe inside the, the partnership, the, the rest of the partners, the person doesn't have the clout and the rest of the partners told them to fuck off and they thought they had it. And, you know, are they a bad person? Or is this just a tough time that a lot of companies are going to be going through? And so like, that's, you know, those are the other little inputs that you don't hear about when everything's up into the right.

• 18:06Speaker 01

  Yeah, we're going to find out in the next two years when all this comes to fruition. It's funny, in the, in the, in the last 10 years, we almost had like the reverse problem, which is you had a bunch of companies that were doing things that needed drained in. And, and, you know, the, the, the hands off VC was a little too hands off. And, you know, that's how you end up with like WeWork style systems and, and, and so it's, it's like it's, we're watching the world flip very, very quickly, which is kind of fascinating. I continue to believe like, you know, look, as a, as a VC, like you don't have to always be supportive. You don't have to always do what the founder says. What you need to do is be transparent.

• 18:46Logan Bartlett?

  Yeah.

• 18:46Speaker 01

  Right? You need to set expectations and say, look, this is what I'm capable of doing. This is what I'm not capable of doing. These are the situations where I can support you. These are the ones where like, look, I just have to, I'm an economic actor. I'm going to have to like look out for my own interests. And I never felt like I needed somebody to always say yes to what I wanted, but I just wanted somebody that wasn't going to like lie to my face or stretch the truth or something like that.

• 19:09Logan Bartlett?

  It's like, it's like the two best outcomes in a fundraise for founders are like to raise money from that person or to get told no very quickly. And like in the middle is a bunch of bad acting that can happen of like, you know, hiding the ball and leading people on and all that. And transparency is really going to be the most important thing over the course of the next like two, three years. Different funds are going to have different parrot rules. Different funds are going to have different, uh, levels of support from, you know, whatever, dry powder that they can put to work, like just communicating and not backing out. But to some extent, that also ties back to the credibility those people have within the partnership, the credibility that partnership has with their limited partners, right?

• 19:54Logan Bartlett?

  Like all these things, it's kind of a dynamic chess, chessboard. And so that's one of the interesting things. Whenever I'm like talking to people about the venture industry. It it it's hard because it's one person's opinion at one point in time in this very three dimensional moving thing that like you could ask me who my top competitors are. And it's a different answer today than it was a year ago because the market looks different, right? And so it's hard to internalize all these factors, right?

• 20:21Speaker 01

  Yeah, also like you catch a fund after they had an exit and like you get a lot more yeses than the one where they just it's just like what what mood did you catch?

• 20:31Logan Bartlett?

  Totally. They're clonizer guys, but moving right now is like, you know, I I I imagine they're they're having a good time over there. Fifty times air NTMA or R, right? For whatever a $2.3 billion acquisition, so.

• 20:46Speaker 01

  I always remember at our first company in 2000 at Invite Media. That we never talked about this story publicly, but we actually had an offer to sell the company. This was like about a year before we actually ended up selling it to Google. And we had this other company come in and try and buy us. They had offered like, I think it was like 25 million or something. And we ended up selling it later for about three or four X that. But at the 25, you know, we were like super young and didn't know what we were doing. And so I remember Nat and I saying, yeah, like, we're gonna take this deal. You know, it was like the first time like $5 in my bank account. So it was like the first time, you know, someone's offering you any real money. And I remember telling our board, we're gonna take this deal.

• 21:28Speaker 01

  And Josh Gollum, who's again, like, just the best at this, was like, I think you're doing the wrong thing. I don't agree with this decision. You know, you're undervaluing the business, you're at an inflection point, so on and so forth. And I'm gonna say yes anyway, if you want to do it. So just so you know, if you want to do it, you can sell it. Like, I'm not gonna stop you, but I just want to tell you all the reasons why I think you're wrong. Anyway, we ended up ignoring his advice because we're idiots and and trying to take the deal. And then we lucked out at this one point where actually the company that was trying to buy us got bought while it was happening. Uh, and we had this unique opportunity to walk away from the deal.

• 22:06Speaker 02

  Like the fish eating each other.

• 22:08Speaker 01

  Yeah, exactly. Yeah. The minnow by the other fish by the other fish. And we were like the, the fucking goldfish. This lot, whatever the smallest fish is you could find. Uh, and, uh, we ended up being able to walk from the deal, go back to Josh and be like, yeah, I think you were right, man. And then, you know, nine or whatever months later, Google bought it for, for like four X or so. So.

• 22:29Logan Bartlett?

  You remember the story? There's the famous like Mark Zuckerberg billion dollar offer from Yahoo. And I think it was Andreessen, uh, who like, talked him out of doing it. Everyone was like, dude, billion dollars. And this is, this is 2006, right?

• 22:43Speaker 02

  Like. The story he tells is that he had no, he never, and he was not interested in it whatsoever. Because like, as he put it, what was that? What else was I gonna do?

• 22:52Logan Bartlett?

  Who Zuckerberg's business or, or Mark's?

• 22:55Speaker 02

  Yeah. Like, what was that? What would I, what would I do then? I'm, I'm only interested in making Facebook.

• 23:00Logan Bartlett?

  I mean, it's the right, at some point, the perverse incentives of like founders at young ages that are like, hey, I've got the tiger by the tail. I might as well keep going. One of my, one of my friends is the CEO of a public company and, uh, he, uh, he, anytime acquisition interest comes in, or like, you know, when people ask him about, and he's like, what, I'm, it took me 12 years to get here. It's like, I'll just go restart. I'll have to go start this whole thing over just to get back right where I am today. Why would I ever be interested in doing that? Right?

• 23:34Speaker 01

  Yeah, but here's the thing. We hear these stories because these are the ones that worked out. Right? Zuck, Evan Spiegel's thing with Snapchat, whatever, you know, you're like, yeah, always go for it. You know, there, there's, I mean, I know a bunch of people personally where, you know, at, at peak were offered a significant amount of money for the business. Wanted to, you know, do the Zuck thing, as George mentioned, and let's just say made the wrong decision, or at least it looks like they made the wrong decision. Uh, and so, but it's funny because you don't hear about those because the founders never want to talk about them. So.

• 24:10Speaker 02

  You know, what's funny is I say that I've been saying this frequently recently to members of the leadership team that we are just now reaching the size.

• 24:21Speaker 02

  That Slack reached when Microsoft crushed them. Like, there's a, there's the final category of shit that can happen. It's not just necessarily those players, but other stuff is now. Like, all that we've done, there's one more death match.

• 24:38Speaker 02

  And all that we've done and all of our luck has bought us the ticket to the final death match.

• 24:42Speaker 01

  My favorite in that death match is like, you have to run the risk analysis of like, does giant company A offer a reasonably equivalent product for free? You know, that one, like, that's a big, I mean, and that stuff happens. What happened with Slack and they got, obviously timed it perfectly and got out. But this stuff does get competitive when the big guys get interested.

• 25:04Logan Bartlett?

  That's a great segue right here. George Fraser, thank you for joining. So for context, George, founder and CEO of Five Tran, which is a, George, Five Tran was founded in 2012. Is that right?

• 25:17Speaker 02

  That's right.

• 25:18Logan Bartlett?

  And $5.6 billion private company most recently. George has the, I think the, the unenviable distinction of being the only founder I've given two term sheets to and still not had the chance to work with, but he is one of my.

• 25:35Speaker 01

  Come on, I can't be the only one.

• 25:37Logan Bartlett?

  I think the only, like, anyone else, I had enough pride to be like, I'm not giving this to you. If, if we, but with you, I was just like, your business is so good. I, I don't care if you, if you say no, right? So George, maybe tell us what, what does Five Tran actually do in your words?

• 25:56Speaker 02

  Five Tran

• 26:00Speaker 02

  goes and gets all the data of your, your, all your company's data from all the places that it lives. So your Salesforce instance, your finance system, your production database, your everything, your ad networks, and we make a copy of that data in your data warehouse. And then we keep that copy up to date continuously if the data changes in the source systems for as long as you keep hanging about. And then you can use this data to understand what the heck is going on in your business.

• 26:30Logan Bartlett?

  So, I mean, one of the things that we were just talking about is, to your point on the Microsoft coming in with Teams and Slack and Scale, I know when, when you were fundraising in the early days, there were a lot of people that said, hey, wouldn't Amazon just do this? Won't Microsoft just do this? Right? And there was always this threat, both from DC's perspective, right? But also like actually this threat of a competitor coming in all the time and offering this. And so we were just talking about the, the Microsoft example with, with Slack.

• 27:15Logan Bartlett?

  But that was something you guys, I mean, had to get around for a long time, right?

• 27:19Speaker 02

  It was the big thing people were worried about for a long time. And the one answer is they have, but they're just not very good. And, and this is actually a really hard problem. And it's not hard in the way that most hard problems are hard problems. Like, you know, making an analytical database like Snowflake, that's something people do PhD pieces in many.

• 27:48Speaker 02

  What makes Five Train hard is not like any single deep computer science problem. It's just this tremendous amount of incidental complexity that we have to deal with in order to, because all of these data sources are very quirky. And, and the thing that makes Five Train great is our willingness to run down all of these little quirks so that you as a user don't have to understand how this, how this all happens. You just, you see, you know, your.

• 28:17Speaker 02

  Your charges table in Stripe and you don't have to know that Stripe contains an eventual consistency bug in its events API endpoint that only comes up once in a blue moon at extremely high transaction volumes. And every time you pull it, you have to go back an extra 60 seconds from the last time you pulled it to make sure that you cover for that bug. You know, which we discovered because a customer, a big B to C customer ran into it years ago. That bug took two years to discover. That's kind of an extreme case. That was the most, that was probably the hardest to chase bug we ever had. And our willingness to do that is what makes Five Trans special.

• 28:55Logan Bartlett?

  Yeah. And there's this economies of scale to that where, where now when you get more customers, you're spending more time QAing these things for your customers. Right? And so it's actually, it's not like a hard, I mean, I'm sure it's a hard math problem. But it's also just like, it's a lot of fucking work. Right?

• 29:15Speaker 02

  It's a huge amount of work in it, but it does have this incredible, as you put it, this QA economy of scale. Because we have many customers who are using every, all the popular connectors. And so we, we get many chances to observe the same bugs. The other interesting phenomenon in our space is that the less popular data sources always tend, tend to have cleaner APIs. Funny. It's the strangest observation, but it's like, if your API is beautiful and perfect, then your business is terrible. Because you were spending too much time on your API or something. I don't know why this is, but it's like a universal principle out there. That like the, the best APIs have hardly any customers.

• 29:57Speaker 02

  And so the fact that we don't have the QA economy of scale there is actually sort of okay.

• 30:02Speaker 01

  It reminds me so much. We had the Plaid founder, one of the two Plaid founders that I've done for a very long time on a while ago. Just talking about like the, the long tail of just like shitty data integrations they had to do. And that's obviously what drove the value there. And I asked him this question. I always like to ask founders as just, because it's like the inside baseball. No one ever figures out is like, was there a product decision early on in the cycle that you can attribute? You can point out the little nuanced product decision that you felt. Drove the differentiation between you and the, I'm sure, 10 other companies that were trying to do something similar and the big guys who were trying to do it poorly. Was there one thing or two things that really changed the game or was it just like chasing each one of these edge cases down?

• 30:46Speaker 02

  Well, there is an answer to that, but first I just want to say that Plaid is a good analogy. Plaid is very similar in this regard and I've met the Plaid founders and discussed this with them. Our companies are sort of cousins. There's another cousin company in this group, Zapier, very similar. And we've talked to the Zapier founders. We've even compared notes about how we do like troubleshooting processes to try to learn from each other.

• 31:09Speaker 02

  And there's yet another cousin company who sort of pretends not to be one of us and it's Stripe. Stripe, a lot of what they're actually doing behind the scenes is dealing with all of these different like banks and credit unions and stuff. And they're basically maintaining integrations to all of these, all of these financial institutions. But I think they don't like to let people know that that's what they're really, really doing back there. So they kind of pretend that they're, they're solving something more glamorous and they, they don't, they're, they're, they're one of us, but they, they pretend not to be.

• 31:40Speaker 02

  But to answer your question, there was a key product decision and it was delivering a normalized schema, which is a little bit like in the weeds of data stuff. But when you deliver data to a data warehouse, traditionally you deliver the data in what's called a denormalized schema, where it's optimized for consumption. So you have relatively few tables, like a simple schema, relatively few tables, lots of columns on each table, and it's just easier to understand and work with. And that's why you do it.

• 32:13Speaker 02

  The problem with this is that everyone's denormalized schema is different. A denormalized schema that works for one company is not going to work for another company. Because these, the simplification that you are imposing on the data is always sort of domain specific. And so what we said is, well, And because of this, every every ETL tool ever made was highly customizable so that everyone can have a different schema. And then that meant fundamentally, you didn't get that QA economy of scale, because everyone was doing something a little bit different. It was these things were basically platforms. They were not actually selling, you know, something that you could just take out of the box and use. So denormalized schema was the traditional way. We did something really weird. We said, we're going to deliver a normalized

• 32:57Speaker 02

  schema. We're going to deliver the data in its full complexity, one table for every concept. And it's gonna, you know, there's gonna be some work there when you have first contact with the schema to just understand what all these things are. But the huge advantage of this is that the normalized schema is the same for everyone. So we could build standardized connectors where everyone is truly running the same code. And then we we got this QA economy of scale. And then after a couple of years, it started to really work well.

• 33:24Speaker 01

  Do you know, like, were there specific competitors, because you guys have been around now for about 10 years, along the way that made a different decision or went a different route? And you felt like you could tell it hurt them in market? And now they're, whatever, fifth place or not around? Like, was that clear?

• 33:41Speaker 02

  Well, in the early days, our big competitor was Stitch, who eventually sold to Talend. And I got a great quote secondhand. So you'll have to fact check this with him, whether it's actually true, but from Bob Moore, who's now running Crossbeam, which is a really cool company. But he said that he sold Talend because, so I just didn't want to spend the next 10 years of my life competing with those guys fixing bugs. Like, yes, that's right.

• 34:03Logan Bartlett?

  That's actually a direct quote. And I'm the one that told you that. So I will, I will take. So I'm on the board of Crossbeam. And he just, he said that Bob's the best. But he said, like, that's a really fucking hard problem. And I'm gonna have to spend my next 10 years of my life, like, just QAing all this stuff. And so, I mean, you guys were both like so early to the cloud, and index to Snowflake, and Redshift, and all these trends, right? And Bob, I don't know how much he says that tongue in cheek, but it is pretty funny.

• 34:36Speaker 02

  So we, we were both like, right there at that. And then, and some of those, in that key early year of like, 2015, 2016. And the different decision that we made is

• 34:49Speaker 02

  We said that we are gonna, we are gonna promise that the data will be 100% correct and we will chase all the bugs and we will design a normalized schema that is possible to deliver for each data source. And Stitch said, we're gonna make a slightly less ambitious promise. We're basically gonna say, we're gonna call the API endpoints. And whatever comes out, we're gonna load into the data warehouse. And that's kind of the definition of our job. And if that does not result in a correct replica, well, that's kind of the sources problem. That's not, that's not our fault. That's the sources fault, which is fair enough. It's, and, and that, that was like a critical decision that we each made differently.

• 35:32Speaker 02

  And, and then it took, you know, it took a while for that sort of bug fixing flywheel to, to do its thing, but eventually there was this pretty significant quality difference in the, in the correctness of the data that you would get with Byotran versus Stitch. And that I think allowed us to pull ahead in the long run, but it was not obvious in 2015 or 2016 that that was right.

• 35:53Speaker 01

  When you were making that call, who was in the room on your team? Like, how did it happen? Like, was it, was it a real debate? Was there a spec? Was there like, you know, because what we asked William the same question about, he has a different version of the same answer and in their context and his take was, everybody thought I was wrong and I wanted to do it anyway. And so we did it and here we are. I'm curious what your version of that looks like.

• 36:18Speaker 02

  Um, it was only three of us at the time and it was me and Taylor and Mel.

• 36:23Speaker 01

  Oh, so this was early.

• 36:24Speaker 02

  This was really early in the design. And it was, we never considered the alternative. We were just like, well, this is how it has to be. You, you can't, we have to take responsibility for correctness here because people were filing bug reports and we were like, well, that seems like a bug. Uh, I guess we should fix that. That was it. It, it was never, we never actually perceived it as a decision.

• 36:46Speaker 01

  They're like, hey, listen, you know, we, we want this to be right.

• 36:49Speaker 02

  The reality is there's a whole other category of work that we abdicated and said, we're not going to touch that. That's your problem. Which is like rationalizing the data that comes out of the system. There's all these crazy properties that the data has, like in Salesforce, like if the account is this one particular ID, like that's the test account. And so whatever you see for that, you gotta filter it out because that's not real revenue. That's test data. And that's like a super simple example. The realities are like way more complicated than horrible. And we just said, like, We're not touching that problem with a 10 foot pole like we deliver the data. If it's a correct copy of the source, our job is done. And then you have to go and figure out, uh, like, how to make sense of all this.

• 37:29Speaker 02

  And that was also, yeah, that was also different than the traditional ETL vendors where they would tackle that. And they would say that was the primary value was tackling that problem. And so they, they would use the same tools to solve the sort of business rule problem. And to solve the, uh, like data integrity, correct replication problem. And what we did is we came along and we said, okay, the business rule problem is actually impossible for a vendor to solve. Like only the customer can figure this out and they should use like a different set of tools to solve this problem. Basically just SQL queries. We looked at, we're like, the SQL queries that come with your database are the right solution to this problem. And we have nothing to add.

• 38:11Speaker 02

  Eventually DBT would actually add something big to that. But we were like, yeah, we're not, we're, we are declaring this problem unsolvable and we're not getting involved. And then this other problem, we're declaring it completely solvable. We're saying not only are we going to solve it, but like, we're not maybe, you don't even know about it. Like if you observe any violation of correct replica, that's just a bug and you just tell us and we'll go fix it. So it was, it was, you know, it was like triage on the, uh, sometimes it was like selling outcomes, but the outcome is a correct replica in your destination. And the outcome they actually care about is, uh, you know, the answer of how much revenue did I actually book in, you know, the Western region and the growth segment for the new product that I just launched last week. And like,

• 38:56Speaker 01

  The business rules.

• 38:57Speaker 02

  There's still one way to go to get to that, but we can't get you all the way to that outcome. You have to do that.

• 39:01Speaker 01

  When you were first going to like, obviously there's the technical components of this of, I've got to like connect all the source systems. I got to deal with the weird edge cases in their APIs or where the documentation is wrong. But then there's the like, the use case that drives demand. What did you think it was gonna be? Like, was it a marketing use case or a sales, you know, the actual business case behind it? And then what did it actually turn out to be that drove the like, what I think of as like, the first stage of demand, the like, people recognizing, ooh, this is actually a problem I want to solve?

• 39:32Speaker 02

  Salesforce was the original tent pole connector. That was, that was the most popular connector for so long. For years, it remained the most popular connector. And now, depending on how you measure it, it's been displaced by ad networks or by databases. But that was the original tent pole. You'd always go after Salesforce first. Get me a copy of all my Salesforce data in the data warehouse. And it's sort of interesting because like, Salesforce is kind of an exception to that rule about what I said, the best businesses have the worst API. Salesforce's API is really pretty good. It's funny, people often complain about it and they're like, it's so hard to write a Salesforce connector. And I'm like, are you kidding? Like, have you tried writing a Marketo connector? Salesforce is nothing. So it was, and Stitcher Salesforce

• 40:16Speaker 02

  connector was good. You know, that's part of why the competition was so close run for those first couple of years. They were pretty similar. It was not much different. Well, and that was the tent pole. That was what you would go in with. So it's like, ah, well, in the second month, our connectors are better. That's a hard, hard edge to drive, right? So that, that was the original, you know, first connector everyone would set up.

• 40:40Speaker 01

  Sorry, now I'm really curious. Was it like, they just wanted the Salesforce data so they could write SQL against it and they didn't want to understand the API? Or was it, they wanted to merge it? Was it like, I always thought that the idea was like, and I want to combine with X other data set and, and you can provide both the ETL for, you know, the combination of the two? Or was it really just, I just want my Salesforce data?

• 41:02Speaker 02

  Yeah, at that time, probably for most people, it was just, I just want my Salesforce data. I just want to be able to write a query against my Salesforce data and have it run fast. I mean, literally, that was our first customer, their use case. It was like,

• 41:17Speaker 02

  Our first customer was Nenefets.

• 41:19Logan Bartlett?

  Was there never a question at Salesforce of like, hey, we'll build this as like another product that we bundle up the like, you know, the data warehouse representation of your data that's really fast. They just never considered building it?

• 41:29Speaker 02

  They've had for a couple of years. Um, the problem is that they haven't done a good job with a bunch of things. You know, I shouldn't say exactly what, because I don't want to help them. Uh, it's funny. I, I always kind of assumed that they would. Uh, I actually modeled this in some of our funding rounds. It's like, okay, if like Salesforce builds a high quality connector, how much of a problem is that for us? How much do we lose? And it was like 5% of our revenue or something. And, um, because by that point, we were pretty diversified, right? It was still the most popular connector, but nonetheless, it was the most, there was, it was, there were hundreds in the race, right? And the funny thing is to this day, their connector kind of sucks. Uh, it, it, they've gotten tripped up on other aspects of the problem.

• 42:13Logan Bartlett?

  Interesting.

• 42:14Speaker 02

  And we continue to make our Salesforce connector better. Like now we have support for formula fields, which is truly unique. We wrote a formula field to SQL translator, uh, so that customers could get formula fields in their data warehouse, um, which we're the only ones who have ever done that. That was like the equivalent of building 10 connectors, that project.

• 42:32Logan Bartlett?

  One of the things I think is interesting about your company is you had a fairly normal, like, start to it, uh, going through YC. And then like when we got to know each other, when you raised a series A, like from that point on, it was, it's been pretty serendipitous and pretty normal, but like the, the period in between there, and I don't know exactly what it is, like 2013 to 2015 or something, 2012 to 2015. You, you iterated quite a bit, right? Like what you went to YC with was very different than, uh, what you ultimately kind of iterated on. Can you, can you take us through that journey of like going with MATLAB in the cloud to Paul Graham to, you know,

• 43:16Logan Bartlett?

  actually finding product market fit, what, what that path was, uh, looked like and how you actually went about doing it?

• 43:22Speaker 02

  Well, MATLAB in the cloud, which was the original, original idea lasted all of the, but, but although that is where the name comes from. If you think about it, five tran, which is a pun on four tran, makes a lot of sense as a name.

• 43:34Speaker 02

  That did not last long, though. It was pretty, it became pretty clear that like I was the only person in the world who wanted that. It was like what I would have wanted as a scientist. That's what we were making. And then it was like, okay, I was actually the only one. And the funniest thing is we actually met two guys who had tried the exact same idea a couple of years earlier. We were introduced by Warren Hogarth at Sequoia. And now I forget the two guys' names, which I feel so bad about. Great favor he did us and they did us because they just went through it. And like, everything they described is like, this was our exact plan. Like, this, this, this is actually better than our plan. Like, we were learning a bunch

• 44:16Speaker 02

  of things that like, we would have taken six months to figure out, but it did not work whatsoever. And they described exactly how it did work. And like, that doesn't mean that just because some, just because someone tried it before and it didn't work, doesn't mean it can't work. But in this case, it was like, we have no answers to the obstacles you ran into. And there was some other evidence that we were on the wrong path. So we pretty quickly dropped that idea. And we said, okay, we're gonna make a vertically integrated data analysis tool for business people. And then that was the start of a long series of iteration. So it was spreadsheets for big data, and then it was more of a BI tool. And data integration was part of that story. So data integration was like part of that stack of what was inside, what we were working on for those whole two years. And then eventually, it became clear

• 45:01Speaker 02

  that if you took just, if you stripped it down to just data integration, that was something people actually really wanted. But that was because Redshift came up. See, we were too early for FiveTran. Like, you couldn't have made FiveTran in 2013. If I could go back in time, I'd have to tell myself, just like, wait around for a little while, like, practice programming. Like, the year to start FiveTran is basically like, the end of 2014 is the earliest time you can start to sell it, because you need Redshift to take off. Because that gives you all of these.

• 45:31Speaker 02

  All of these West Coast tech companies that are pretty small, setting up their first data warehouse for the first time. They don't have any preconceived notions of how ETL should work. They'll talk, they'll, they'll consider anything. So they're perfect, like, um, early adopters. And they, there was just wave of that year, you know, from, because of red shift taking off and the next year. And that, that was the opening that we went into. So there was a very much a timing element of it, but you know, we were too early and we solved that by just like cranking away on various still faded ideas, um, ill fated variations of this vertically integrated data analysis tool until the time it became right.

• 46:10Speaker 01

  There's like this window almost in history, kind of like right when you guys were starting through maybe a few years ago where like enterprise data tools, and it's like a horrible generalization, but I'm just going to use it for a second. Where you have companies had begun to shift. They had on-prem data. They had SaaS data. They had all these kind of like, you know, alternative data sources where, where things lived. And there was just like a whole bunch of problems that created that, you know, they almost like the enterprise data stack had to be rebuilt. And you had a lot of companies that were just kind of what you guys did, right? Smart, technical people who were just in the game in the beginning at the right time.

• 46:48Speaker 02

  Yeah. And just found a piece of that and grabbed onto it. Yeah.

• 46:51Speaker 01

  Yeah, exactly. And it worked right. Clearly. Like if you were really good at what you did, you, you could kind of like iterate your way to the idea. Do you think that time is over?

• 47:00Speaker 02

  No.

• 47:00Speaker 01

  Do you have to like really know what you're doing? Or is there just so much free endless field out there that that can still work?

• 47:07Speaker 02

  I think it's not over. Um, I think that, uh, the actual like execution engines that run all these workloads are based on ideas from.

• 47:21Speaker 02

  Um, what did the, the first one was like, uh, Monet DB X 100 from like 2003 or something. And some, a lot of things have changed in hardware. And I, I don't mean to be too um, dismissive or anything here, like some, some big changes have been adopted, like the separation of compute from storage is a really important idea that comes from changes in the hardware. But there have been some other changes in the hardware that have some pretty profound implications for uh, how you build an analytical system that

• 48:00Speaker 02

  uh, I would be worried about if I were selling an analytical database, uh, like

• 48:07Speaker 02

  the, the, the, the ground is shifting beneath our feet, or it shifted and people are noticing that it shifted.

• 48:12Speaker 01

  Interesting. It's good that like the hardware changes the, what software is capable of and that changes the ability to deliver and you get that like little cycle that

• 48:20Speaker 02

  And it changes the trade offs, you know, the something, something where you have a trade off where like on the one hand, uh, pro and con, and it drives you to a particular decision, but then the size of the pros are shrinking, you know, the the weights of the things start to change and, and when they, they cross the threshold and now it flips the other way. And it's like, oh, now you want to do it the other way. And suddenly all of this work is invalidated. Uh, so, but it's going to be good for the ecosystem and good for the customer. Ultimately, all of us vendors just have to figure out how to navigate it. And the things I think you're going to see are you're going to see costs go down, you're going to see latency go down, you're going to see um, performance go up, you know, it, it'll all become cheaper and faster and easier to use and have fewer compromises.

• 49:04Speaker 02

  I, I really look forward to the point when you can like go into Salesforce and like change something. And then go over to your other tab and like refresh your dashboard and the time it takes you to do that.

• 49:18Speaker 02

  Like the, the data that you just edited has made its way over, like that is, it's super hard to figure out, to, to imagine that with today's technology, but people really want that. And I, I think it does not like, it does not violate physics. Uh, like it can be done. It's just gonna take a huge amount of work, but people will love it once it is, even if it isn't that necessary, um, people will love it. And so whoever can deliver on that will win.

• 49:44Speaker 01

  It's coming. And it's not just in your, it's not just in your dashboard, right? The way it gets updated in like every connected SaaS system and it kind of, you know, perpetuates out through the entire stack and it's immediate and there's no, uh, there's no like downstream QA issues anytime you make like a scheme of change and just, you know, your entire system blows. I mean, these are all the things I remember doing in 2007, like just getting some of our own internal data from different systems into a warehouse that we could rely on and wasn't breaking all the time. Anytime any source system made a change. I mean, it was a disaster, total disaster.

• 50:19Speaker 02

  Some of the source change stuff is really tough because it's fundamentally about workflows and semantics and there is no technical solution. Somebody has to talk to somebody else and explain what this thing means now. Um, so some of that stuff isn't going to go away. Like the broken dashboard because somebody changed the source system. It's just, it's fundamentally a workflow problem. I guess it goes away. If that, you know, I shouldn't say that. It's just a different kind of problem. You have to build tools that allow people to coordinate these activities so that they don't break each other. So that right before I, I joke, it's like the Salesforce administrator, right before they make the change, they get like an electric shock from their chair. Like, what are you doing? You're about to break everything downstream of you.

• 51:01Speaker 02

  And maybe that's a slight exaggeration, but somehow they find out like, oh, you can't just change that. Like you have to go and coordinate with all these other parties.

• 51:08Speaker 01

  Yeah, even just, even just the linkage of, of, of what this change will downstream affect because, you know, especially in these large organizations, you're, you're making a change over here. You don't, you, you're not aware of the 5,000 things that ultimately rely on that data and just being able to say, hey, if you make this change and ideally into like a queue where someone has to approve it. Here are the 35 things that are going to break. And by the way, here are the people who are responsible for those things. We're going to make them aware before, you know, when they have like in nuclear war, like both people have to turn the switch at the same time, kind of thing. They just like knowing that that kind of traverse what is connected to what and to who. That in itself, right there is like a complicated problem where you have like

• 51:51Speaker 01

  services teams internally at some of these big enterprises just map that out for you.

• 51:55Speaker 02

  The hard part about that, well, the part that I, the thing I don't have an answer to for that is I think in order for that to work well, you really need to have testing be native to the workflow of each tool. Like the Salesforce administrator really needs to see the error in Salesforce sandbox after they make the change, but before they deploy it to production. Right? It doesn't, I don't know if it's going to work if it's in like this data lineage, whatever tool, but somewhere else, everyone's going to log into that. And they're going to like change the sandbox and they're going to have a separate five train connection for the sandbox, the lineage tool, and then the run test. I, it's just, I find it a little

• 52:36Speaker 02

  hard to believe that people are going to set up all these extra links in order to support testing, testing, you know, it's like eating your vegetables. People don't really want to do it. You have to make it really convenient. You have to like, you know, toast those Brussels sprouts and make them look really nice to get people to eat them. And I, it's really hard for me to picture how this happens. I think Salesforce changes are like the hardest piece of this. And so anyone who's working on this, like start there. Don't start with your production database. That's bullshit. I hate it when people, they're like, I have this great solution, data contracts or whatever for my production database. I'm like, yeah, yeah, that's the easy part. Tell me how you solve this for Salesforce.

• 53:14Speaker 01

  It's really interesting to me that like, when I ask the, you know, is the low hanging fruit kind of gone? You actually point to hardware changes that are happening more than anything else. It's just like a very nuanced answer that I wouldn't have been able to answer myself. I kind of like appreciate the nuance there. Now I'm curious, but I'm not going to ask you on this call because I'm sure you don't want to give it all away, but that that's like actually driving some of the, the stack changes that you expect to see.

• 53:39Speaker 02

  No, I'll give it away. I think the fundamental and, and that, you know, I think that the the other problem you identified, which is the sort of testing problem is actually just as big. It's just not one that I tend to grab it. I tend to like pay attention to because I'm such a like nerd. And it's more of a workflow problem, but it's, it's just as big. But the hardware change that I think is really invalidating a lot of the assumptions on which these systems were built is that almost that, that you can get such a large VM. You can get a VM with like two terabytes of memory. And

• 54:15Speaker 02

  That almost no one actually needs a massively parallel distributed processing system. You just need many of those, right? Like, no one, there isn't hardly anyone has a query that can't fit on one node. And that is just a very different, that unlocks a very different set of decisions that have huge benefits. If you only run on one node, then you don't have to do shuffles and you don't have to do all kinds of things. And you can get a lot of efficiencies and things can run faster. And so that is just a huge change that I think is going to really like mess with a lot of people.

• 54:56Speaker 01

  That's fascinating. Like, we made this assumption for a while that all of these giant data questions we had had to be distributed in nature because the machines weren't big enough to actually like hold it in and run the query. And that was true. And now it is no longer true. And you've got like all these distribut-- yeah, that's interesting.

• 55:15Speaker 02

  People fool themselves on this because they look and they say, oh, look at all this workload in this big warehouse. It's like, wait a minute. How big is the individual query? Because if you just have lots of queries that need to run simultaneously, you can distribute them onto a farm of independent nodes that each don't actually share information with one another. And then there are other things you can do, even if you have one query that's too big. If it's only like 10 times too big, there's some tricks you can pull in order to take pieces of the workload and distribute it without having to tackle the really costly pieces of true distributed processing. It's like this giant fork in the road and everyone for years and years and years has gone down this path. And I actually think for like 99.9% of companies, the other path is probably in theory now the correct answer. The problem is you have to do like all years and years of just engineering

• 55:59Speaker 02

  work in order to sort of realize that promise. So it's not, it doesn't just happen one day. But I think that's really a space to watch.

• 56:07Speaker 01

  Sorry, we bored Logan with the nitty gritty, but you know, I love this stuff because like for me, this is the shit that actually matters when you're a founder. It's like all those little nitty gritty details that nobody actually talks about in the press release or the website, but that's the nuance. That's the stuff that differentiates, you know, Five Tran from the company that got bought for one one hundredth of the price.

• 56:29Logan Bartlett?

  The question I wanted to ask both of you, I guess, George, you and Taylor, your co-founder, have known each other, or your families have known each other for, what, I've heard the story, a hundred years or something, like same cabin, same.

• 56:44Speaker 02

  More in fact, yeah. The cabins were built a hundred years ago, but the families knew each other before that, yeah.

• 56:51Logan Bartlett?

  You guys, your families have known each other forever. And Zach, I mean, you and Nat are, were what, friends in, in. College. College. And, and so one of the things like I feel people often say is like, don't risk your friendships or whatever going in on these businesses, particularly when there's generational relationships in place, right? And you're, it's one thing if it's like your, whatever, your, your next door neighbor that you play video games with. It's another if there's like family dynasties tied in together. I'm interested in, was it, was that something that when you were starting this, that you were actually concerned about? Or was it refreshing

• 57:34Logan Bartlett?

  to have a founder that you knew so well and there was such mutual trust?

• 57:39Speaker 02

  For us, it was mostly just upside. There was a lot of, there's like a deep reservoir of trust and that was really important. And funnily enough, because of this extensive, it was not just us who knew each other, it's all these relatives. Because of this, everyone knew what we were doing and was like, all on top of it. And we realized, oh my God, we have to make this. It made that much more pressure to succeed. Because we would just never live it down. We'd be hearing, you know, I'd be asking when I saw Aunt whoever once a year, she'd be asking, oh, it was too bad how that Five Tran thing didn't work out. I'd hear that, like. Every year at that time of year, the rest of my life, like that would have happened. And, and so it created a lot of pressure beyond what is already there. And you know what?

• 58:23Speaker 02

  Fear of failure is a very powerful motivator. So for us, it was, it was really all upside. I can understand how it, it could go wrong, but it's worked really well.

• 58:34Speaker 01

  I think these like rules about, you know, co-founders or solo founder or married founders or, you know, related. It's all bullshit. Like you can do this in a variety of ways as long as the people are just really, really smart. And there's like three things that matter, in my opinion, like the, they've got to be very smart. They've got to have just like a drive to do, get shit done constantly. And that's whatever, whether it's like my aunt is going to be, you know, never letting me live this down, or I dropped out of college in my case, or whatever. Like, there's just got to be something in the back of your head going like, don't fuck this up, don't fuck this up, don't fuck this up. And you have to have the capacity to do it on, on the skill side.

• 59:19Speaker 01

  And then I think there's like an ethical component, which is especially from co-founders, which is like, you're going to make a bunch of decisions that affect the other person. And you just need to have trust that like, even if you make the wrong decision, you did it for the right reasons. And I think if you have any, like some combination of the three of those things, doesn't really matter if you're, you know, hundreds of generations back, that's a kind of a crazy story.

• 59:41Speaker 02

  A hundred.

• 59:42Speaker 01

  Or you just met in college, or you're, or you're married.

• 59:44Speaker 02

  Five, I think.

• 59:45Speaker 01

  Whatever, you know, like the back in where the Vikings were fighting, but like, yeah, I just think you can kind of do it in a bunch of ways. It's more about the personality traits of the people than the like, you know, the setup. I, I, that's at least been my experience.

• 1:00:01Speaker 02

  Towards your earlier point, it's really the details that matter. These broad principles are always wrong.

• 1:00:06Speaker 01

  Yeah.

• 1:00:07Speaker 02

  Don't, this co-founder configuration is incorrect. For every form of that statement, you can go by the counter example. It's, it's, it's the details that matter.

• 1:00:14Logan Bartlett?

  Well, Cisco, they were married, right? And that's one of the biggest enterprise companies ever. So.

• 1:00:19Speaker 01

  You can't ever have co-CEOs. Well, have you heard of Workday? Right? Like you should never take the founders, uh, and, and put some like old school MBA in charge, Eric Schmidt.

• 1:00:28Logan Bartlett?

  Now, uh, one other one I just picked up along the way, but you had. You went to Lawrenceville, right? You grew up in New York, uh, or you grew up in Rye. You were born in New York. Is that right?

• 1:00:39Speaker 02

  Uh huh. That's right.

• 1:00:40Logan Bartlett?

  Lawrenceville Boarding School, uh, in New Jersey. Um, it seemed like you, you had family that had gone there over the different, maybe your dad went there. Is that?

• 1:00:50Speaker 02

  Uh huh. Yep.

• 1:00:51Logan Bartlett?

  It seemed like that was a, uh, somewhat impactful formative, like high school is for a lot of people, but it seems like you had a reverence to that experience in some way. Can you talk a little bit about like boarding school and, uh, why, why it was such a good experience for you and any impact it maybe had today?

• 1:01:10Speaker 02

  Yeah. I mean, it was, it was a very special place, a very special institution. Um, gave a great education. Uh, really that was my liberal arts education in college and graduate school were something totally different.

• 1:01:24Speaker 02

  And, uh, it also, um,

• 1:01:30Speaker 02

  you know, there, there's a tremendous sense of history at, at these old East Coast boarding schools and of like destiny almost, you know, you go there and you feel like you're meant to do great things. Everyone here is meant to do something important. And I think that really stays with you. So for me, it was, uh, it was a great experience.

• 1:01:48Logan Bartlett?

  I'm picturing, uh, Hogwarts and Dead Poet Society and, and all of that stuff.

• 1:01:55Speaker 02

  Definitely people refer to it as like Harry Potter with no magic. That's not inaccurate. And, uh, one of the characters in Dead Poet Society supposedly this, you know, maybe bullshit, but supposedly was based on a friend of my dad's from Princeton, uh, who went to not Lawrenceville, but a different boarding school. So, you know, you're, you're on target.

• 1:02:16Logan Bartlett?

  Interesting. You spent six years getting your PhD, right? In neurobiology, uh, which for another

• 1:02:23Speaker 02

  Five and a half.

• 1:02:24Logan Bartlett?

  Five and a half. And one of the things I heard you say was that the randomness of academia was, was a, a, almost like a daunting thing, or it just seemed too, uh, abstract in terms of how people actually succeed. And so you, you wanted to go down the entrepreneur path because of that. Can you talk a little bit about like academia and sort of be doing that for five and, I mean, five and a half years is a long time, right? Uh, and, and going down that path and then ultimately deciding to be an entrepreneur?

• 1:02:55Speaker 02

  Well, I went to graduate school and I did graduate. I am Dr. Fraser.

• 1:03:01Speaker 02

  But no one calls me that except my boyfriend because he also has a PhD that no one calls him by.

• 1:03:07Speaker 02

  And he calls me Dr. Boyfriend, so.

• 1:03:13Speaker 02

  I but by the time I graduated, I had decided to go into the private sector. And the reason was basically what you said. I I watched other people who were further along. And who actually succeeded in getting the, you know, tenure track faculty positions that you need in order to be a research and be a full time researcher.

• 1:03:34Speaker 02

  Seemed extremely random. Like there were some great people who got these jobs, but there were also terrible people who got these jobs. I was like, that guy got a tenure track faculty job? How? And vice versa, you know, I would see great people just fail and fail and fail. And it was like, what are the rules of the system? I don't understand. And, um, you know, startups.

• 1:04:01Speaker 02

  Well.

• 1:04:03Speaker 02

  Also containing a large degree of luck and randomness and also having a very low success rate. They do follow rules. Like, at the end of the day. You know, it's the customer who decides whether to buy, whether they want to buy your thing or not for themselves. It's not someone passing judgment on whether you are good, right? And so that there was something appealing about that. It's kind of you make your own destiny.

• 1:04:26Speaker 01

  The other really interesting thing about academia, I think always. I found really interesting. I'm curious if you felt this was. The promotion decisions similar to what you were talking about. Were not based on any sort of like rigorous. Interview, any sort of like structured evaluation process, all the things you end up having to put in place in a company because. You know, if you don't have standard, clear processes for how people get promoted or paid, there's. Politics and cultural issues and the whole system kind of breaks down. But academia kind of doesn't really have any of that. There's almost no structure and process. It's completely amorphous, if you will. And it's just really shocking to me that

• 1:05:09Speaker 01

  we haven't come up with a like, hey, if we're gonna think about who gets like a tenure track position in chemistry, maybe we should actually test their understanding of chemistry at that level. Like, we should have rigorous interviewing. And it doesn't seem to exist in any way. You just look at like, the papers they published, essentially.

• 1:05:28Speaker 02

  Well, I think once the people get tenure track positions, it is fairly, that is fairly straightforward how it works. It's publishing papers, high impact journals.

• 1:05:37Speaker 01

  Yeah.

• 1:05:38Speaker 02

  And getting grants. That's it.

• 1:05:39Speaker 01

  I don't know. Just that felt like.

• 1:05:41Speaker 02

  But I, I, I have no quarrel with that. It's more kind of the initial gate that was bizarre to me. Although then at the second layer, the grant, the grant process and the paper process is also very bizarre. Like, what succeeds and what doesn't.

• 1:05:56Logan Bartlett?

  Hey, one thing I wanted to ask about was, you guys acquired a fairly big company, what, about a year ago now?

• 1:06:05Speaker 02

  It was one year. The deal closed one year ago, like, three days ago, something like that.

• 1:06:10Logan Bartlett?

  So your round was 5.6 billion at the time. You cash in stock of 700 million, you guys disclosed at the time for HVR. So, rough math, 12.5%. Typically, not to do the VC platitude thing that I, that I did earlier, but I mean, private to private acquisition, acquisitions in general are hard, right? Acquisitions of a big scale are hard. And doing this type of thing in general is hard. You're, you're a year in now. Any lessons that, like, things that you didn't expect going in to doing that, that you maybe got lucky with that worked out and things you wish you had done differently?

• 1:06:55Speaker 02

  I don't know. It's gone pretty well. Most of the advice we got about this is what make acquisitions work or not work was right. And we like, very diligently followed it.

• 1:07:07Logan Bartlett?

  Which was what? What did people tell you?

• 1:07:09Speaker 02

  There's so many different pieces of it. That the compatibility of the culture is very important. It's not just about the sort of market and the technology and that logic. But if the company cultures are not compatible, one of them will just. Eat the other.

• 1:07:25Logan Bartlett?

  How many people were on each side when the acquisition happened?

• 1:07:28Speaker 02

  Man, HBR was like a hundred and we were like,

• 1:07:33Speaker 02

  I don't know, six hundred, something like that.

• 1:07:36Logan Bartlett?

  So proportional roughly to?

• 1:07:38Speaker 02

  Maybe a little, maybe a little more than that.

• 1:07:40Logan Bartlett?

  Yeah.

• 1:07:41Speaker 02

  And, and we went in knowing that the cultures were mostly compatible and there were maybe some pockets that were going to be a little different and that those predictions pretty much turned out right.

• 1:07:51Logan Bartlett?

  Uh, yeah.

• 1:07:52Speaker 02

  There was one big surprise. The technology was surprisingly compatible. Uh, we knew that there were a lot of similarities under the hood, but it turned out, once we actually looked at the source code, that uh, that, that both systems convert all, all the data they ingest into like an intermediate representation internally. That is like an update or an insert and a bunch of other things. The details are very important and, and very specific, uh, and they were identical. It was like, you know how alligators and crocodiles actually are not related. They just both evolved to like live in swamps and so they look the same. It was like that. We had just

• 1:08:33Speaker 02

  co-invented the exact same internal representation with one exception. There's a thing that happens in APIs that just never happened in databases. They were, HBR was all about databases. Um, so, uh, but that, that was really convenient and helpful. Um, in, in terms of getting a integrated product up and running quickly and quickly meant like six months, right? But that's fast.

• 1:08:55Logan Bartlett?

  Was doing it during COVID helpful, uh, or was it harder? Uh, I mean, you know, we, we were still sort of in the middle of Omicron a year ago, right? And people were definitely working from home. Did that make it easier to get everyone on the same page and not have to worry about locations to the same extent? Or was it harder?

• 1:09:13Speaker 02

  Well, I, I, the definition of COVID being over, I think varies from person to person. And my view COVID was over at that time. And I think most people at. The combined company felt that way. And so we spent a lot of time together. And, and I know there are some people who are going to feel that COVID is not over forever. But I think I would, I would disagree with the premise of that question like we, we had a giant in-person event for the entire company. Not that long after the acquisition closed, Camp Five Trend, which is a very special part of our culture that we've done for many years. And it was conveniently timed to be right after the acquisition. That worked out well.

• 1:09:51Logan Bartlett?

  Yeah.

• 1:09:52Speaker 02

  And we had sales kickoff the next year. You know, same as usual. And that was a great opportunity for people to meet and spread Omicron to one another.

• 1:10:01Logan Bartlett?

  Yeah, totally. That's the way all good company cultures bond over the last couple of years.

• 1:10:06Speaker 02

  Right?

• 1:10:07Logan Bartlett?

  Well, thanks for doing this. This is, uh, this was great. Uh, thanks for letting us go in so many unstructured ways here.

• 1:10:13Speaker 02

  George, good to be here. No problem. It was fun.

• 1:10:19Logan Bartlett?

  So that'll do it for the 36th episode of Cartoon Avatars. Uh, thank you to Zach Weinberg. Thank you to George Fraser for, for coming on. Uh, really fun conversation. Next week, we have one of the white whales of guests that I've tried to pin down for a long time. Supposedly coming on, we'll see if we actually record it. Uh, it's, it's been a long time coming. So thanks everyone for listening and have a good weekend.