Windsurf: The Making of a Billion-Dollar AI Company

31:37So it's certainly like, I agree with you, like these rounds happen so fast. I think as a result, you have to like, just change how, you know, I don't think there's the traditional like, you've got four weeks and they're going to give you all the data up front. And no, I think you have to like figure out scrappy ways to get a lot of that yourself. Going back to our topic on competition and windsurf in particular. So they're in an increasingly competitive market. It is so competitive. We've Cursor, Copilot, Devin. there's so many in this environment so how have they been able to rise above the pack with this impending news maybe it happens by now maybe it doesn't but how how have they been able to gain that that that attention in that market share yeah it is like so it's interesting i mean there are endless like possibilities with this technology that we have right now with all the and multimodal models is just like endless the the what's going to end up happening but i mean there's like these near-term what i call like battleground spaces and you know cogen is probably like the like most most most most war zone right now you've got customer support um and then you've got like you know copywriting maybe some areas of like sales ai that are starting to get really competitive um but yeah cogen is just like an extremely extremely competitive space and so i I mean, I think like there's a few parts of Windsor strategy that have been really great.

33:05I mean, one, their team, their infrastructure experts. So they are just really, really good at being able to like optimize these models and the product for what users care about, latency, accurate results, things like that. And then I think the strategy of like, let's become an enterprise ready tool, but then also have this sort of like best in class PLG product to like get leads for the enterprise motion. I just think that's a winning approach. And, you know, I think the team just proves time and time again with every wave that they ship rapidly that they really, you know, care about usability and their users.

33:47And I was just kind of talking about this the other day. This isn't like a new take, but, you know, I think it's very unclear right now what out of all these companies who has a moat, but I do think like speed right now is kind of the only moat that you can have. And hopefully, hopefully that will translate into, okay, like you get into these enterprises, maybe you learn and you become more personalized to certain users' workflows, there's switching costs. Maybe that becomes a moat. Or maybe like there's some amount of network effect when you start to collaborate with other people on the Windsor platform or something like that.

34:19But at least today, you know, the moat for all of these companies is just like how fast they can move. And, you know, I think what's good about this space is it's huge. And, you know, as I was saying earlier, it's a lot bigger than I thought. The pendulum swings back. Before it was, okay, let's get to profitability as fast as possible, as fast as possible. Now it's growth at all costs, save nothing, go for it, go big or go home. Yeah. In some areas, there's always a nuance. But if this is true and this acquisition goes through a$3 billion acquisition by OpenAI, Lee Marie, you will be a legendary investor.

35:10You are a legendary investor and probably better than 99 % of VCs out there and everyone will be very jealous. I can't comment, but you're gassing me up. You really are. Just doing my job. Yep. No, no commenting for now. All right. Fair. Well, speaking of rising above the pack, the rise of AI was the theme of your latest fund announcement. I would love to go into this further. What are the key trends in AI infrastructure and ML applications that you're watching? For sure. Yeah. So we already talked about kind of the battleground spaces. And I think what's been, I mean, I'm very lucky that my partners, even before I got to KP two years ago, have been making investments in these spaces that have been doing really well.

36:00So, for example, we invested in Glean. Or I should say, I mean, it was more of a, we kind of incubated Glean. So we had a big role of like getting Glean off the ground. Another zero to a hundred million ARR track. Yeah, I mean, yeah, Mamoon, he's definitely, definitely a goat when it comes to just like amazing software companies. And this is no exception. I mean, Glean, we're super bullish on. And yeah, so that's like, I'd say that's in a battleground space. Like at this point now, you know, like enterprise, knowledge, discovery, I think that's becoming all of, you know, perplexity, all these other tools are now trying to kind of like get into that area too.

36:39So that's certainly like a battleground space. I mean, you have the medical scribe space. We're in Ambience, who's doing very well. So this is basically a tool that doctors use during doctor appointments to like listen in. And then instead of them having to spend countless hours like writing down notes into their EHR, like as a follow-up email summary, like it's just done automatically. And this is kind of like a no brainer. And then let's see, I think the only area that we don't have anything in right now that's like very obvious is customer support. And so, you know, we have a growth fund. We're looking, but, you know, I'd say those are the areas where it's the tech, clearly there, the customer need, clearly there.

37:22People are just duking it out, competing for the big prize. And then you've got, and this is what I'm excited about and what I spend the vast majority of my day on is, okay, well, what are the other really big companies that can use these models to create in-customer value, to help customers cut costs, basically, you know, to do work versus just kind of like be sort of a software tool. And I mean, I think there's like a few areas that I'm really excited about. I mean, I think we are just at the tip of the, you know, iceberg when it comes to enterprises actually using AI effectively. I think a lot of people can use ShadGPT, like want to use AI, but in terms of actually like automating tedious back office tasks or, you know, being able to like come up with like new ideas really quickly or analyze it like data.

38:19There's just all this stuff that they could be doing that they aren't because it's just really hard to integrate AI into like these, you know, giant enterprise data silos and things like that. So you're certainly looking for like companies that have a strong point of view on like, okay, this enterprise doesn't need to use BPOs for this sort of task. They can use maybe a BPO plus a combination with a software tool or maybe just like fully automating certain types of processes end to end. I think I'm excited about, there's probably a lot of opportunities in finance to automate like financial analyst workflows.

38:55We don't yet have a bet in that space, but we have a lot of people that are on the KP investment team that used to work in investment making or private equity or hedge funds and the amount of like PowerPoints they had to put together or like Excel workflows or all that stuff, that's all been relatively untouched by AI. And you could imagine, especially as the models develop better quantitative understandings, like just being able to deal numbers better, you could imagine those workflows being automated away. I mean, I think there's just a lot of like very domain specific stuff, whether it's in supply chain, whether it's in like biotech, biopharma, whether it's in like other parts of legal.

39:34I mean, we're in Harvey and they're going really well, but there's just so many workflows there. Yeah, I just think like even the models today, and hopefully the cost will continue to go down by 10X every year and it will be able to unlock a lot of this. But there's just a lot today that is very rote and doesn't require a high degree of like, you know, creativity or whatever that these models can can automate yeah there's a lot and it's it's great because you have such a severe expertise in all of it and you've been watching it for a while that i've learned a lot even in this short amount of time that we've been talking but i know we didn't get a chance to go deep into your background yet so i want to do that I will hype you up again, Leigh Marie.

40:20So part of your portfolio is Windsurf, Nooks, Chronosphere, Neon, and Persona. Can we just go back to Leigh Marie, Scale.ai, and go forward from there? Let's talk about your background. Let's do it. Let's do it. So yeah, I guess the way that I get to Scale.ai was again through composition math, which is hilarious, and also poker. Alex and I got very close on the MIT poker club before he dropped out. I think his sophomore year at a found scale. So I was at scale. Let's see. Pretty early on. I was like an intern when we were like three. Wow. And then. Yeah. Yeah. Yeah. When it was like tiny, tiny.

41:01Oh my gosh. Yeah. I mean, it was a scale was the API for human labor back then. And I called it the Alex API because we didn't yet have like a task force. So if you put in, is there still an API and you could like, you know basically there's an end point where it's like you know they'll we'll make you a phone call or like order your pizza and like alice will be like the back end like doing that which i'm like you know this i i mean this is one of the greatest skills but he's always very very in the details and um you know just worked it works extremely extremely hard um but basically honestly when we were like three i was probably full-time when we were like around seven started as an engineer.

41:41This is back when, you know, I think scale very quickly realized, like you can't be the best API for everything you have to pick. And, you know, there were YC company. And so a lot of these self-driving car companies, especially through YC, started using the API. And that was just this incredible customer pull. Like, okay, this is like, if you want to build email platform of the future, where's the biggest pain point right now? It's providing labeled image data for self-driving cars. And so kind of rethought the whole product to just be around that, providing it basically end to end. You don't have to manage the labelers.

42:19You don't have to manage the payments. You don't have to manage the quality. We do all of that for you. And I think that value prop really resonated. And then very quickly, Alex realized there's an even bigger opportunity in LiDAR or 3D labeling for the same companies. So the vast majority of companies outside of Tesla believe that you have to use this 3D sensor or LIDAR to have enough redundancy to make a really safe self-driving car. And so I, as you know, the product service area got bigger. I had always kind of wanted to be a PM at scale. I started as an engineer. Alex is very product-minded.

42:52But as the product service area got bigger, you know, there was an opportunity for me to come in and, like, product manage. And at first I was just like, I get to code and I get to do product management tasks, you know, and I get to organize a gear. um but eventually that became you know like okay where where are we taking this product and how do we integrate ml to reduce our cost and things like that um which was extremely interesting um and so i was there for around four years like in that time scale started working with basically basically all the software in car companies um also made inroads in defense which was back when it was like yet you know you know that's like not cool yeah yeah there's like in this recent thing where it's cool to want to work with the government the gundo yeah yeah there's like a term now american dynamism dynamism but like back then it was like it was like not cool um and alex i think really long sales cycle hard to sell into morally wrong you know like there was just like all this evil yeah evil to help the u.s government achieve the u.s government's goal but um yeah so it was like in back then too so that's even more admirable um but that's now become like a big part of the business um and then you know this was all pre or maybe i was there maybe at the tail end of this but you know we actually helped open ai with gpt2 um and i came thinking like like what is this like you know i don't even know if this will ever like what are what are like lms and transformers like what yeah and then and then you know i left and then you know a few a few years later that now it's like um this is scales like main main business line um and i think that you You know, one thing I always say about scale, Alex and his ability to put himself in like the right rooms, know the right people and be able to see like, okay, where's all the like, like revenue going to come from, from the next wave of like what's going on in machine learning, like human intelligence and I'm biased, but I think human intelligence and data will always be needed.

44:46And I have like full confidence because he's done it time and time again, that Alex is always going to be able to figure out like where is that place where scale needs to like build a product um to sort of like be able to bring the humans um to to the models um and so they like you can't ask for a better first startup experience like every six months the company changed dramatically it grew i got to learn things about sales marketing i get to learn just in general like how chaotic a startup is um and i get all these people that last there like haley marie like my startup oh just like it feels like we're changing things all the time and everything's on fire all the time and i'm like great because if it's not there's like no customer need like if you you are not like feeling like you are running around all the time and everything's on fire then like there's probably not product market like there's probably it's it's not there's not that the only like hyper growth startups even the ones that seem like they have their shit together they don't um it's just like uh it's always a constant fire and And I think it's nice that I lived through that because now when I go partner with companies, I'm like, if I see that craziness, if I see that they're all over the place, they're being pulled in all these different directions, I'm like, okay, well, they've stumbled onto something gold.

46:03Yeah. And I was talking with Bucky ahead of this, and he nudged me over some questions for you. One of them in particular, actually, most of them do come from your time at Scale.ai. So while you were there, you developed an eye for talent vortexes. Could you explain this? What does that mean? For sure. You know, I think this is kind of like this also the name of the game and mentor, I think. But yeah, I mean, what really matters when you're building a startup is hiring extremely talented people. um and you know it's a it's this flywheel where you hire talented people and they'll hire more talented people and if you stop hiring really talented people then um it's much much harder to hire talented people because the bar is lower and it kind of reflects on all these downstream effects it makes it harder to make good products to sell bad products etc so um i think you know at scale we were just obsessed with hiring extremely high quality people um even so much so as i remember i got so frustrated my first year at scale because it seemed like everything was i was so overwhelmed i was i was coding all day i was labeling tasks all night alex was labeling tasks all night we were all just like totally overworked um and then we would interview these candidates and you know like these are smart people um and then you know i was ultimately will say, well, they're not, you know, no, they're, I don't think they, they meet this bar.

47:34And I'm like, oh my God, if we could just have like two or three more people right now, it'd be like, it would make my life so much better. But ultimately what that did was a few things. Like it just kept the engineering bar so high that, you know, then we went and were able to develop these, these sorts of great, great future products. And then also kind of created this like aura around the company, which I think can be, it's like this weird amorphous thing, but they can be very useful down the road just with everything. Like his scale just became like an MIT, like this is the selective place to work.

48:04And that sort of helps us hire way better talent in the future and stuff like that. So, you know, it's like it's always tough when you spend so much of your time on hiring and then you just keep rejecting all these people. But then at the same time, you know, I've just seen it play out. And now multiple companies where if you do that and you're very intentional about hiring, you just create like a world class team and things become a lot easier for you down the line. And so then how this, you know, translates to investing. Right. Yeah. That was going to be my second question. How does this inform your decision making?

48:38I'm trying to mind the talent for Texas. Yeah. And there's like a few, I'd say like a few different ways I do it. I mean, the funniest way is, you know, I remember one of my first investments at FF was Persona that I worked on with a few other investors there, including Napoleon. And he's like, okay, well, you know, Lee Marie, like how strong is this engineering team? you were an engineer and i remember i looked through not saying you should always do this i remember i looked through every linkedin of every engineer just to like we want to understand like what types of like backgrounds are they hiring what types of people are they hiring um and then also you just like looking through like talking to people talking to people there like looking they're looking through company reviews and i think like persona in particular has just created this insane culture where not only do they have all these extremely experienced um yet like fast moving engineers but also their retention um like employee retention is just off the charts crazy like like no one ever leaves because they have such an amazing engineering culture and so it's like this is like a this is an unfair advantage even though it's a very crowded space i didn't need verification like they have a truly special team and it reflected in we obviously did a lot of product feedback calls too but like you know time and time again we're hearing that you know this is the best product on the market so there's that there's the vote out that way and then i say like that's a relatively skilled out company it was a like a series c i think um but you know when it comes to like very early stage companies one of my favorite things to do as an angel and now one of my favorite things to do as a vc is i have a lot of technical friends and they very you know occasionally they're looking for jobs and i love to make matches and so i'm always trying to if you just want to talk you know some ideas i'll tell you companies inside the portfolio i'll tell you companies outside the portfolio whatever location stage whatever like it's my job to have like a big map of all these companies in my head and what i'll do is like this isn't like i love to make matches because it makes me feel good because i mean i love to see my friends crushing it but then separately i get a lot of data on like what companies are able to recruit and retain really strong people and so i've very frequently been i've like updated my model of a company like oh my god vortex because you know i take somebody i used to work with a scale that i know is like a really strong engineer and then like you know he's asking like oh well you know maybe you should give me an open eye referral but then i'm like oh well you know yeah i will and i did but then i also was like hey you should also meet this like super early stage startup you know it's got you know you'd be their first engineer i was like like you know i was just kind of lobbing it lobbing it out um and then the startup managed to close him um and i was just like holy moly like this is this is like i i update on the founder I'm just like, wow, this person's like clearly one, a good identifying, good identifying talent.

51:21And then two, good at like convincing Cal that has many options to go work there. So yeah, it's kind of like, how do I, you know, and I try to do these poker nights and that's another way of like getting people together and getting insights and things like that. But do, how do I like find all these signals of like these brilliant people in my network? Where are they going? And like what companies are managing to, to, to snag them? It's a much more thorough process than what is memed about quite frequently. And it's like LinkedIn turnover from open AI. So you're deep in the weeds. Yeah, no, I try to, I mean, ultimately I think adventure is like, you know, you want to figure out like people, people that you trust and really smart people.

52:03And, you know, it's not like a super scalable thing, but I think if you, like, work hard enough, try hard enough, and you'll start to hear, like, the same names come over and over again. And you'll, like, be like, okay, these are the places where, like, they're just really sneakily good at recruiting all the best people. And you should probably just, like, invest in those companies. So I want to pivot like a little bit, but go back to our conversation on like the AI-ification of companies. From your perspective, have you been seeing more bootstrap startups in this era? Like at what point? Let's just start there.

52:44Yeah, I don't see like, I mean, and to be fair, I'm, you know, usually I'm looking like seed and after. So like the might just because I look a little bit later than like maybe a pre-seed investor. but um i don't see a ton of bootstraps i do see this there's this new term called seed strapped that i've been seeing more of this is no this is a real thing molly you gotta look it up um this is new this is strapped seed strap company basically what this is is a company where like they raise a bit of venture funding like a few million dollars to like get out the gate um and they are so wildly successful with their tiny team and they're like crazy 10x 10x revenue thing that like they're just hand over fist of money like they don't need they don't need more money they don't need to raise a series a like maybe they don't like maybe they don't need to raise a series b yeah you know like maybe maybe they live they're like first first round of funding after after the seed is like a series c i've been hearing this time and again it's like companies founders are like yeah i want to raise this funding but then no more after that you know this is the only round we're gonna raise and i'm like are you okay sure usually that's totally like total like you know that's that never that's like never what happens but then like for these companies at least a few companies that like i don't pretty well that is kind of what's happening like do i think they'll ever raise yes like you know i think or ultimately they want to build like a huge business and they got a higher sales team like we were talking about before like there's probably and it's a competitive market and then you've got these people that are like you know potentially like trying to compete with you and you need more resources like yes you'll probably have to raise again but it'll just be interesting i think we'll see multiple examples of like a company raised a seed and then a company raised like a series c and it's just gonna be weird but that's just yeah kind of kind of how this this world is because you are so close to the ground to the efficiencies on the early stage side of how to rebuild and build a company from the ground up with the most efficient team possible with all these tools and all these cool things that are going on how does this inform the legacy portfolio of growth companies i know you don't do much on the growth side but i'm i'm really interested to know like you have companies like brex plaid robin hood rippling square these are all large companies how how are you or the team helping inform them restructure if they are to meet this new environment i mean a lot of these companies are just being like i mean very proactive about all this i mean i think they all like are yeah the vast majority of them realize that like hey like we need to make sure that like we're using the most cutting edge ai tools when it comes to like cutting costs or making the user experience better or things like that and you you've been seeing a lot of companies like reinvigorate it like because they're able to to really like create a create a great product with ai um they have like today um and so like yeah i mean in general i think a lot of these growth growth stage companies are like you know doing doing a good job and you know whether it's like we do a lot of dinners like executives at these companies and like talk through like okay best practices sharing best practices of like what tools they're using what they're doing um whether it's like things like that um or just like i mean i mean a lot of these companies are just incredible incredible companies that figure stuff out.

56:06Cut overhead, insert the robots, make the product tax again. Yeah, I mean, it's just incredible how much your day-to-day changes with these tools, right? Now we've got deep research to write memos internally. And as you said, there's all these tools that we can use to glean insights of who's going where. Was that a little plug for glean? oh yeah i uh this this is a high plug you know like i describe it and you say it and then uh this is not an advertisement but glean if if you want to advertise for sorcery let me know i use glean a lot yeah yeah they're yeah incredible incredible tool um definitely we do this thing portfolio review and it's like every i think every quarter we go through all of our or most of our investments and like gleaned as the thing where it will like summarize a lot of our board decks into like okay here's like the key metrics and commentary and it is a lifesaver yeah especially for my partners i have a lot more boards than me i couldn't imagine doing that non-automated we got to check that out check it out by the way um how many boards are you on uh let's see i mean i am on i have one board director seat and that is windsurf um and then I have a few observer seats.

57:31I still observe Dion's board, which is an investment fund. We did part of their series A. I'm an observer on Nooks's board. So they're a AI tool for sales, salespeople. They do a lot of different things. They have a power dialer, which helps you cold call really efficiently. A training product where you can like train before the call. What are you going to say? And then also a new prospecting product, which is really great. But I'm an observer on that board. I met Dan, the founder, when I was at scale, actually. It's another scale talent vortex outcome. And then also observer on a stealth investment.

58:07So TBD. A stealth investment. I'm going to guess no comment, but I'll take it. No comment. Yeah. Hopefully we're coming to announce soon. But that and then, I mean, I don't really see companies. They only have boards. But yeah, I work with like three seed companies. We have to talk about poker. Okay. okay lemurie can you please explain your poker situation i heard the buy-in is like a million dollars per hand is that right it's actually 10 million uh we are we are playing for snacks yeah no no it is not um it is is uh very very modest i like tend to keep it my mile like the whole point of the poker nights and they're very like they're like like it looks small poker night like it's it's in my apartment in the marina like we have a round dinner table and i put felt on top of it to make it look more like a poker table yeah i got on amazon it's like molly's game that's that's the goal um without the fraud but um yeah so it's like you know usually like a few hundred dollar buy-in um or maybe less i want it to be like if there's startup founders even that are very early stage that like to play poker i want them to be able to play poker and i want you know like also personally you know just i i'd much rather keep it friendly than people really focused on on um you know just betting giants amounts of money i think it's still like if you're playing with a hundred dollars you care you care about what the outcome is that's the point of the money um at all like i think if you play for no money people don't care um so you got to like put a little bit of skin in the game to just encourage good gameplay but ultimately it's so that i you know i get to get to know people like string them on a relationship with them a A lot of times it's like founders, engineers.

59:51I've had people get hired from the poker nights. I've had people sell their software to other people at the poker nights. I've literally built my relationship with founders who I later invested in, effectively starting with a poker night or maybe I'd met them once before. And then like we got to know each other through poker. So it's just like, and then at scale, we hired so many people. That was our strategy. We would bring people in the office to play poker. And then, you know, like they would get to know us and things like that. so a lot of the most important relationships in my life have been like formed through poker and it's something that i love to to like continue as a as a vc and um yeah there's not there's there's so many positive things i can say about it um but it's definitely uh it's not the meme where it's like i'm i'm not you know like making investment decisions based on how people play poker but i am like you know it's just a great opportunity to like get to know get to know somebody well there might be some bias in there some subconscious yeah you're really really uh you're really really bad or really really good maybe you know um yeah you're really really good definitely positive so what kind of tips would you have for someone who's just starting a poker like a youtube channel do you have your own youtube channel i am i uh let's see i like uh there's a few things that i recommend there's this app called poker power i think poker power app um that's like for beginners and like they do like courses and things like that um that that i've like you know tried before that i think is really great um and then there's like um i mean was on the poker club at mit and the book that we would always reference is called the grinder's manual it's like a pdf online um but in terms of like the theory theory of poker and like you know really learning how yeah to do to like the math behind a lot of it um it's great um but yeah also just like playing and finding people that play at the stakes that you kind of feel comfortable with.

1:01:46I got very lucky. I did not have played poker before college, but I was a Jane Street internee a few times. And instead of going out on the town in New York, we would just stay in the office and we would play poker after work every single day. And we played only$20 buy-ins, thank goodness. So I learned how to play by just like, I'd play, I'd make terrible decisions. And then my buddies would be like, hey, Lee Marie, do this instead of that. Or think about this and that. So, you know, just like figuring out places where you can like go on Mars. It's an amazing way to have your own community building while doing something that you love.

1:02:19Totally. So do I. And score some deals out of it too, because that's always important. Of course. Of course. Well, to wrap it up, we've covered a lot. We went through windsurf as much as I could possibly get out of you. And I appreciate that. we talked about the succession planning of Kleiner and how you guys have retained your legendary track record and a little bit more on yourself and your background especially with scale AI I mean it's really tremendous to end it out on a positive note I'd love to know what are you most looking forward to this year all right looking looking forward to this year I mean And I just think there's so much opportunity in AI in particular, but also in a few other things too.

1:03:06But I'm just very excited to continue to meet incredible founders and really partner with them closely and help build their businesses. But yeah, I mean, ultimately, that's what we're looking for. And I just never experienced, even back in 2021, when there was a very high deal volume, Like, I've never experienced so many smart people, like, wanting to become founders. And, like, the sort of, like, why now and, like, the tailwinds being so, like, real. And I just think that, you know, there's so much opportunity right now to build a massive company. And so, yeah, I'm just, I'm really excited to continue to meet founders.

1:03:48And, yeah, continue to make investments. Fantastic. Well, Leemarie, it was a pleasure to have you on. And thank you so much for your time. All right, Molly. Thanks so much for having me. Really appreciate it. Great questions.