OpenAI vs Anthropic vs Open-Source | Token Maxing, AI Hangovers & The Coming ROI Reckoning

1. 0:00 – 1:22

   Intro

   1. MGMatan Grinberg0:00

      The world going forward, there is going to be nothing that no one can build. Everyone is trying to commoditise the other. Value accrual is a time-dependent phenomenon

   2. HSHarry Stebbings0:08

      Meet Matan Grinberg, CEO and co-founder of Factory. Before Factory, he was a physicist. He spent 12 years trying to be one of the best string theorists in the world. Now he's changing the world of software development. He works with some of the biggest enterprises in the world. He does look like Matt Damon from Good Will Hunting, but he is one of the best founders I've met

   3. MGMatan Grinberg0:28

      So many of the tasks that we're doing, we don't need the very frontier to do it. We might see a short-term contraction of usage of the very frontier models. I think it's pretty embarrassing that we don't have frontier open models in the United States. Name a legendary company that has a shit sales or marketing team. You can't. The age of the polymath is back. We will see the best companies treat teams more and more like SEAL Team Six or like professional athletes. Ready to go? [upbeat music]

   4. HSHarry Stebbings1:06

      Matan, it is so good to have you in the studio. You've just, uh, insulted my continent with the suggestion that we've only come up with bottle caps while you came up with Transformers. Um, [laughs] not wildly untrue. [laughs] But this is gonna be a fun show, so thank you so much for joining me.

   5. MGMatan Grinberg1:21

      Thank you for having me, Harry. It's

2. 1:22 – 2:41

   Will AI actually increase GDP?

   1. MGMatan Grinberg1:22

      a pleasure to be here.

   2. HSHarry Stebbings1:22

      Now, I was just doing a show yesterday with Rory and Jason, and Rory was basically saying, you know, the fundamental question is will we see an increase in GDP, um, coming from AI and the coding developments that we're seeing, and will it lead to GDP increasing above the 2% average for the last 200 years? Do you think we will see meaningful productivity gains from the AI tooling that we're seeing, or is Uber's concerns validated?

   3. MGMatan Grinberg1:48

      So I think yes, absolutely we will. We will see tremendous growth from these tools. I think it takes time to permeate through. Um, because you can tell, like, on an individual basis, like, almost like on a problem-by-problem basis, we can solve problems faster with these tools. Now, companies generally org-organize around solving problems. Um, and if you're organized around solving problems and you have some set of personnel, you might say, "This is the number of problems we can solve at a given time based on how many people that we have." Everyone is now gonna be able to solve more problems with the same number of people or solve the same number of problems with fewer people, but it takes time for the resource allocation to adjust. A lot of businesses will have to ask, "Do we want to solve more problems now because of the increased leverage that we get, or do we want to solve the same problem but now we can do it in a more efficient

3. 2:41 – 5:05

   Smaller teams or bigger ambitions?

   1. MGMatan Grinberg2:41

      manner?" That's, I think, a question that a lot of businesses will be, will be, uh, grappling with.

   2. HSHarry Stebbings2:46

      Do you think we will have fundamentally smaller teams? Uh, which ultimately suggests that number two is the option that most people take. Or do you think we will have actually the same size teams and we'll just go after a more expansive area?

   3. MGMatan Grinberg2:56

      It's really not obvious because, um, there are dynamics that it's hard for me to predict. But what I will say is, again, bringing it back to problems, all of these companies are now going to have to think, "Okay. We have all this new leverage. Do we want to solve the same problem? Do we want to increase our ambition and solve a bigger problem, or do we want to solve more problems, uh, that might, you know, our users maybe have?"

   4. HSHarry Stebbings3:19

      I was watching Andrej Karpathy and he was talking recently about, you know, the 10X engineer actually is wildly misunderstood, and you won't see the 10X engineer. You'll actually see a smaller number of 100X engineers and kind of the rest, and this bifurcation of engineering talent. Do you think that is the right way to look at the future of engineering talent?

   5. MGMatan Grinberg3:40

      Directionally, yes. Because what is a 10X or 100X engineer? I, I don't necessarily agree with the language around it, but, like-

   6. HSHarry Stebbings3:47

      Why not?

   7. MGMatan Grinberg3:49

      I, I think, like, it just implies as if, like, 10X, 10X of what? Is it pure output? Like, if, if it, if you, when you say 10X it means, like, how much code they're writing. Yeah, now I can write a billion lines of code with these tools. It might be shit lines of code, though. Um, so maybe the way that I like to think about it is, like, load-bearing individuals in an org. It's kind of like if you remove this person, things fall. Where there, there, in some orgs there might be people where if you remove them, nothing happens and they're n- you know, not load-bearing in that case. And so, you know, the... Basically, these people who have very high leverage are now being handed a tool that gives them even more leverage. And so using the language of 10X or 100X, yes, they're levered up. They can have even more impact. Um, but, uh, a-and I think, you know, with that leverage language, those who know how to use leverage will be able to have even more impact, and those who don't will kind of on a comparative basis be that much less valuable to a business.

   8. HSHarry Stebbings4:47

      When we think about kind of the two different parts, you said that, hey, you have the option of you can do more with the same size teams or you can reduce teams and do what you already did. If I am thinking as a leader today, what would be your biggest advice to me on how I should think about resource allocation for tokens

4. 5:05 – 6:49

   The resource allocation problem: tokens, dollars, people

   1. HSHarry Stebbings5:05

      internally?

   2. MGMatan Grinberg5:06

      Yes. This, this is a, a great point. Um, this resource allocation problem of token-- It's not just tokens. It's like dollars, it's tokens, it's people. This is, I think, going to be the thing that over the next 24 months every C-suite is going to be thinking about. And I think the right way to go about it is what is the core competency for our business? What actually matters for the business that we are doing? Um, and then how do we allocate resources accordingly? In other words, if you're a logistics company, your core competency is probably not software development. Now, you might have had a lot of software engineers as a means to an end to deliver on your logistics goals, let's say. Um, but that might not be your core competency. And so what you should be thinking about is not how do we get more engineers to make more features because that's what engineers have in the past been judged by, like how many features do they ship in a quarter. Instead it's like what are the actual output metrics that matter for our business? And how do we now allocate resources, whether it's dollars, whether it's tokens, whether it's headcount, to, uh, more dramatically move the needle on that business outcome? And I think this is, this is great for the world because I think part of the reason why so many organizations got so bloated is 'cause we were in a period of time where everyone was focusing on intermediate metrics. If you're an engineering team, we wanted to ship three features this quarter. Did you ship three features? We shipped four. What a great quarter. Like, that doesn't necessarily matter for the business at all. And so now it's like finally coming back to what matters in the first place. Like, what are the business metrics that we wanna, you know, move the needle on? Is it customer satisfaction? Is it revenue? Is it market share? Um, and you can kind of tie back every individual's work to that, whether it's marketing, sales, engineering, all of it.

5. 6:49 – 10:01

   Kirkland's $500M AI bet and the build vs buy question

   1. HSHarry Stebbings6:50

      Kirkland announced a $500 million spend. You're, you're friends with, uh, Winston from Harvey. Uh, fantastic guy, um, who obviously I'm sure has-- I don't know if you guys have spoken about this actually, but, like, that's a, it's a big spend.

   2. MGMatan Grinberg7:04

      Yeah.

   3. HSHarry Stebbings7:04

      Five hundred million dollars across five years to internally build their own Harvey or Lagora.

   4. MGMatan Grinberg7:08

      Yeah.

   5. HSHarry Stebbings7:08

      Um, how did you think about that?

   6. MGMatan Grinberg7:11

      Um, I mean, it's fun. You know, talking about core competencies. Kirkland's spending half a billion dollars to build their own AI tools. Uh, my understanding is that building AI technology is not a core competency of that firm. I think I'm, I'm s- I was surprised to see it. Now, I actually think this is good for Harvey because it's nothing like trying to do something yourself to make you realize, oh, shit, this is actually really difficult. This doesn't actually matter for us to have the in-house ability to build this ourselves. Let's go and have someone who is an expert in this to go and build this for us. That is my sense.

   7. HSHarry Stebbings7:44

      My favorite is also the amount of people that like, "See, we told you how easy it was." And you're like, it's so easy they're committing half a billion dollars. [chuckles]

   8. MGMatan Grinberg7:52

      Yeah.

   9. HSHarry Stebbings7:52

      That would suggest the opposite.

   10. MGMatan Grinberg7:54

       Yes.

   11. HSHarry Stebbings7:55

       Um, I had Brandon on from McCool the other day, and he was fundamentally saying that the next 12 months would be the most value accruing 12 months for AI infrastructure companies. We would see that the models of the products and the AI application layer companies would be most at risk denigrated. Would you agree with that?

   12. MGMatan Grinberg8:16

       I would disagree. I'd pretty strongly disagree, um, for a couple things. One, actually sticking with the Kirkland thing, I think as an example, we're so used to a world where moat in software was, "I know how to do this and you don't, and so you're gonna pay me because I have the engineers who know how to build this, and you simply cannot." Now, the world going forward, there is going to be nothing that no one can build. Every single piece of software anyone will in theory be able to build. Now, back to the resource allocation, though. Is it worth your time and your energy to go and build it, or should you go to someone else who has already built it or can do it faster? To me, it kind of-- An example of this is, like, suppose we had a, a very busy day at work. I could probably go and pick up lunch for everyone on the team. I know how to do it. I know how to walk out the door, place an order, hold the bags, bring 'em in. Now, just because I know how to do it, is that an efficient use of my time? Probably not. I'm probably gonna say, "You know what? For my resource allocation, I'm gonna pay someone to go and do that for us," because at Factory, our core competency is not that the CEO goes and gets lunch for everyone. Um, and I think it's somewhat similar here, which is like, just because you can build a lot of these things does not mean you should. And in fact, oftentimes you wanna be really ruthless about what are the few things that you and your team own and do end to end, and then if it's not relevant to your, like, core business and your core competencies, outsource it.

   13. HSHarry Stebbings9:40

       What would you like to do but it's not core competency for you, and so you don't do it because of focus?

   14. MGMatan Grinberg9:46

       Oh, man, I, I enjoy, like, making breakfast. I haven't done it in like three years.

   15. HSHarry Stebbings9:51

       [laughs]

   16. MGMatan Grinberg9:52

       Um, and I li- like, there's nothing like w-- It's just, it's, it's not time efficient. It just doesn't make sense to spend my time doing that.

   17. HSHarry Stebbings9:57

       All right, okay. [laughs]

   18. MGMatan Grinberg9:59

       [laughs]

6. 10:01 – 11:58

   Models, apps and infra: who gets commoditised?

   1. MGMatan Grinberg10:01

      Um, but like I, I... It is, you know, it's something that I do enjoy. But, um, but to your point on model applications infrastructure-

   2. HSHarry Stebbings10:09

      Mm.

   3. MGMatan Grinberg10:09

      Um, I'm not sure if you've seen, there's a meme of the Microsoft org chart, and it shows like, you know, different segments and they all have guns pointed at each other. Um, just to show like in Microso- you know, there's a, there's a lot of bureaucracy and everyone's kind of fighting for who gets to do what. Um, I think that image is pretty accurate to what's happening right now with the models, the application companies, and the infrastructure companies, where everyone is trying to commoditize the other. Everyone is trying to say, "Oh no, this one is irrelevant. This-- All the value's gonna be here. All the value's gonna be there." The reality is value accrual is a time-dependent, uh, phenomenon. So like, it's not like there is one person whose steady state gets all of the value. That's not how it works. It's maybe for this next year, this person is who has the pricing power, who gets the value. This next period of time, these people get it. We are all, you know, whether overtly or not, and maybe I'm saying the quiet part out loud, like everyone is trying to commoditize the people that are not them. So for example, we're model agnostic. We want to give our customers the best pricing, the best performance, the best speed for whatever task they wanna do in their software development. Um, and we wanna make sure that OpenAI, Anthropic, Google, Microsoft are all under pressure to make sure they give the best models for as cheap, as quick as they can, and don't feel like they can just, you know, charge whatever they want. Now, similarly, they, you know, the model companies wanna make it such that the applications are all trivially easy to build, and really the, the product is the model. And then the infra companies have their own spin on this. But the reality is, like, everyone's trying to commoditize the one that's not them. And so, you know, from, uh, from Mercor's perspective, you know, it's very much in their interest that models that have access to proprietary data get differentiated value and capture a huge amount of value 'cause that

7. 11:58 – 13:57

   The bear case against Factory

   1. MGMatan Grinberg11:58

      validates their business model. It's a, it's a big push and pull of who can, who can get the leverage.

   2. HSHarry Stebbings12:02

      What is the belief that would invalidate yours?

   3. MGMatan Grinberg12:05

      The bear case against Factory is if one model provider gets significantly better than all of the others. Um, so basically, I think a key thing for us is that all the models are gonna be roughly as good as each other. They'll be good at... Well, one is a little bit better at review, one is a little bit better at testing, one's better at Python, this and that. It all kind of fluctuates every week. Like, even already people have a, have a hard time keeping track what model is number one, what's the latest thing that came out? If one model ends up going way above all the others, that's a case where it's like, okay, we wanna put... You know, companies might wanna just completely go in with them, um, but then that's a monopoly for the entire economy to be worried about.

   4. HSHarry Stebbings12:45

      Is the rate of model development sustainable? And what I mean by that is, like, you know, I was, I was with the founder of Nabeus the other day, and he was talking about it on a, "Oh, every few weeks we see new models." And I said, "You're, you're, you're wrong. Every few days."

   5. MGMatan Grinberg12:55

      Yes.

   6. HSHarry Stebbings12:56

      Especially when we look at Chinese open source, it's like three or four a week.

   7. MGMatan Grinberg12:59

      Yeah.

   8. HSHarry Stebbings12:59

      Is that rate of model development a feature of the time that we're in, or is it an ongoing characteristic or trait of this environment?

   9. MGMatan Grinberg13:10

      I think eventually we'll stop seeing them as model releases, and they'll feel more continuous.

   10. HSHarry Stebbings13:14

       Hmm.

   11. MGMatan Grinberg13:15

       Like just how before it was, you know, GPT-2, then GPT-3, then GPT-3.5, then 4, 4.1. But like, and then you get more and more des- And like, you know, 4.523. Like, eventually they're just gonna not announce it, and it's just like, "Hey, look, here's our model that's continuously getting better." Um, because it- already people have fatigue. Like, engineers at the enterprises that we work with can't keep up with every single model that comes out, nor should they. And I think that's the, the, the whole, you know, the case for the application layer, whether it's us or like a Harvey or whoever else, is we're gonna figure out what model is best for what use case, where the trade-off is between cost, um, quality, speed, and we'll just deliver that to you based on the task that you have. Because it's hard

8. 13:57 – 17:08

   The rise of open-source models

   1. MGMatan Grinberg13:57

      to focus on what matters for your business and then also keep track of all these models that keep coming out.

   2. HSHarry Stebbings14:01

      A big question that people have is around the rise of open source and whether everyone is concerned by the amount they're spending on tokens just being so much-

   3. MGMatan Grinberg14:08

      Yeah

   4. HSHarry Stebbings14:08

      ... larger than they thought. "Hey, we spent our annual budget and it's May. Shit. Um, maybe we should move to open source."

   5. MGMatan Grinberg14:15

      Yeah.

   6. HSHarry Stebbings14:16

      And we're seeing more and more great companies use frontier models, see where they can get to, and then move to open source to get as close to that as possible.

   7. MGMatan Grinberg14:24

      Yeah.

   8. HSHarry Stebbings14:25

      How do you feel about that being a considerate threat to maiming the market for frontier models?

   9. MGMatan Grinberg14:30

      I think it's a really important counterbalance because it show... It basically allows you to make the trade-offs of what tasks do you want to put what level of intelligence on. Um, and I think it's a really important counterbalance because a lot of enterprises will realize so many of the tasks that we're doing, we don't need the very frontier to do it. Like, and we can do it much faster, much cheaper with these open models. Um, and again, it's part of the resource allocation. And to do good resource allocation, you wanna be able to be anywhere in that cost, quality, speed trade-off.

   10. HSHarry Stebbings15:03

       I, I love the, uh, the GIFs on Twitter or memes on Twitter when it's like, you know, "Me naming a file," and then it's like the massive cigar with the blowtorch.

   11. MGMatan Grinberg15:12

       Oh, yeah.

   12. HSHarry Stebbings15:13

       You've seen the...

   13. MGMatan Grinberg15:13

       Yeah.

   14. HSHarry Stebbings15:13

       Yeah, I love that.

   15. MGMatan Grinberg15:14

       No, 'cause it's such overkill. But also there's a funny dynamic that emerges, uh, which is there's kind of an ego thing where, "Oh, no, no, the work that I'm doing, only a frontier model could handle. Oh, this mere open model can't deal with the work that I'm dealing with." And this is like even admittedly, when I first started switching over, I'd be like, "Ah, I don't think an open model can handle this." And it's like, no, it probably can. Um, and it, it's kind of a funny thing to, to mentally deal with, of deciding, you know, manually or then having the router do it for you.

   16. HSHarry Stebbings15:42

       My question is, enterprises like security, they like reliability, they like ease.

   17. MGMatan Grinberg15:46

       Yes.

   18. HSHarry Stebbings15:47

       And when you have frontier models which are packaged perfectly, priced clearly, and it's secure, do they not just go for that, the easy option, over trying to be smart and intelligent routing to different open models?

   19. MGMatan Grinberg16:01

       So a couple things. Uh, one, it's easy when there's only one of them. But again, as we said, a new model comes out every week, and if you have to go through the full enterprise process to get every new model in, it's not very easy. Um, two is it's also really expensive. And if you're seeing your costs go up like crazy and not having an ROI case, it doesn't make as much sense. And I think something that's interesting is there's kind of like three phases that we're seeing happen in these enterprises. Um, so phase one, this was a, a couple months ago, was board yells at CEO, "Hey, Mr. CEO, what's your AI strategy?" CEO's like, "Shit, I don't know." Uh, "CTO, hey, what's our AI strategy? Let's make sure we adopt AI." And so then phase two was kind of AI at all costs, token maxing. "Part of your performance reviews, we're gonna measure how much you guys use AI. Everyone, you have to adopt." That was phase two, right? Get as, as many people to adopt as possible. Phase two happened a lot faster than people might have expected, and so now we're ending phase three. Like, phase two was kind of like the, the debauchery, the long night, you know, taking shots, having a great time,

9. 17:08 – 19:32

   The AI spending hangover

   1. MGMatan Grinberg17:08

      using all the AI. Phase three is the hangover, where you go and look at the bill and it's like, "Oh my God. We are spending so much. I have no idea what the ROI is. Does this... Like, is this helping our business?" Um, that's where a lot of these companies are at now, and I think this is why routing is so important. 'Cause they're realizing... And this is a true story. One of the CIOs I was speaking with realized, "We've been spending hundreds of thousands of dollars per month on people asking Opus 4.8 questions like, 'Hey, how's it going?' Like, 'What sh- what are my macros from the food I ate today?' Like, 'What's the weather like?'" And it's like, guys, like, we don't, we don't need the, the frontier of human intelligence to be doing this stuff for us. Let alone it's not even work-related in some cases, but...

   2. HSHarry Stebbings17:50

      Will we see a contraction then given the hangover period being realized?

   3. MGMatan Grinberg17:56

      Um, I think there are... We might see a short-term contraction of usage of the very frontier models, but I think that's healthy Um, I think it's healthi- healthier to do that than to, like, be blind to it and then have a real sudden kind of change there

   4. HSHarry Stebbings18:11

      Uber announced last night, I think it was, or yesterday, that they were having, like, a $1,500 budget-

   5. MGMatan Grinberg18:16

      Yes

   6. HSHarry Stebbings18:17

      ... per individual.

   7. MGMatan Grinberg18:18

      Yeah.

   8. HSHarry Stebbings18:19

      How do you respond or think about that?

   9. MGMatan Grinberg18:21

      We... I've literally seen this with dozens of our customers where initial- And this is a lesson on our post-sales team, where initially we came in, we were like, "Oh, by the way, we have these user limits, but here, these are the models go crazy." This was before we had routing. And it happened a couple times with customers where the usage would go crazy. They hadn't spent the time to actually determine what parts of the code base do we wanna dedicate these tokens to versus not. And then they were like, "Oh my God, we're spending so much. This is crazy. We need to put in token limits." And at first, when we weren't, like, the first time this happened, we were like, "Oh my God, their usage went down. What's going on?" But spending time with them, we realized, wait, we need to make sure with every customer we are having a very clear conversation with them of, you know, "It looks like you guys are spending a lot of tokens on some of these things. Have you thought about consciously, yes, we want to do this?" Sometimes we'll proactively set in those user limits just so it's better to be aware as you're going up as opposed to just going crazy and then kind of realizing. And so what's happened with Uber publicly has happened privately with a lot of customers of ours. And yeah, there's a little bit of shock where it's like, okay, wait, let's put in these user limits. But then you come into a question of, well, wait, this team is really important. They should have a different user limit than that

10. 19:32 – 24:14

    Token spend as a % of dev salary

    1. MGMatan Grinberg19:32

       team. And we're just getting towards this world where you have very nuanced resource allocation throughout your org.

    2. HSHarry Stebbings19:39

       To me, the biggest question that I ask myself, and I think we need to ask ourselves as an ecosystem today, is if Marc Benioff says that he spends 300 million on Anthropic, okay, for his devs, that is 3.8% of salaries. Okay, great. What will that number be in three years' time? Because if it's still 3.8, fuck. [laughs] If it's 20, fuck again, but fuck positive. [laughs]

    3. MGMatan Grinberg20:04

       Yeah.

    4. HSHarry Stebbings20:05

       Um, and if it's Brandon at McCaw who says that he's spending more on tokens than he is on headcount, fuck again, but even more positive. [laughs]

    5. MGMatan Grinberg20:14

       Yeah.

    6. HSHarry Stebbings20:15

       What do you think that percent of dev salary is in three years?

    7. MGMatan Grinberg20:19

       I think it's actually a more nuanced question than we might think. I actually think it can be as low as 0% for some individuals, and it can be as high as, like, thousands, tens of thousands of percent for some individuals.

    8. HSHarry Stebbings20:31

       And what's the dependence there?

    9. MGMatan Grinberg20:33

       It depends on what the unique skills of those individuals are. And I'm saying individuals and not devs in particular because I think the way we even organize roles is gonna be very different, where I'm not sure, like, dev as a word makes sense. Traditionally, it's like custodians of code, right? The people who do anything relating to code are engineers or developers. I think everyone is gonna be loosely interacting with code in your org, whether they're sales or marketing. Um, but I think the difference is there are gonna be certain people where, again, we're coming back to resource allocation, they get more leverage by using more tokens. And then there are gonna be certain people where actually they don't really need tokens at all, and that's not what... how they deliver value to the business. Like, for example, you know, maybe our best salesperson, the way we use them best is not by having them use tokens, but by going and meeting people face to face, right? That's an obvious example 'cause they don't write code in the first place. But I think similarly, maybe there are some engineers who they actually do their best work by spending time with users, spending time with our customers, maybe doing some data analysis that's not very token, uh, expensive. But then there are gonna be others who are delegating to dozens of droids in parallel, working on a ton of different crazy features and refactors and migrations. Um, but I don't think it's gonna be a consistent number across the board. In fact, I would argue that if your org has a standard number where it's like, we want every engineer to be at this percent of their salary in token use, you're probably, uh, painting with way too wide a brush.

    10. HSHarry Stebbings21:56

        If I were to say, "Give me an average number," what will that average be? Like, what will the median be?

    11. MGMatan Grinberg22:02

        I would say order of magnitude, it'll probably be comparable to salary.

    12. HSHarry Stebbings22:06

        Comparable to salary.

    13. MGMatan Grinberg22:07

        Yeah, like, on the same order of magnitude.

    14. HSHarry Stebbings22:08

        Within the three-year timeline?

    15. MGMatan Grinberg22:10

        Yeah.

    16. HSHarry Stebbings22:11

        What percent of tasks today b- be... using frontier models could be done with open source models?

    17. MGMatan Grinberg22:19

        Probably, like, 80 to 90%. It's typically the planning that really needs the frontier models.

    18. HSHarry Stebbings22:26

        But is, is that not, like, the most... I'm, I'm sorry, I'm, I'm really, you know, um, sh- dim. [laughs]

    19. MGMatan Grinberg22:33

        [laughs] Harry, that's nuts. That's cool.

    20. HSHarry Stebbings22:36

        Uh, but, but, but if it's 80 to 90%, does that not just, just present the biggest bear case ever against code X or crawl code? 'Cause you're just taking away 80 to 90% of that time.

    21. MGMatan Grinberg22:47

        Well, it depends 'cause those... that 20, 10 to 20% could be the most important tokens. It's maybe, like, right? Like, 10 to 20% of the tokens, but those are really, really important 'cause it's kind of decision-making tokens perhaps.

    22. HSHarry Stebbings22:59

        Sure.

    23. MGMatan Grinberg22:59

        Um, but it's s- it's very similar to how we structure human orgs, which is, like, oftentime, oftentimes leadership makes very key decisions that determine the fate of the company.

    24. HSHarry Stebbings23:08

        Sure.

    25. MGMatan Grinberg23:08

        And they, they don't spend the most hours... Like, if you look at the human hours of a company, most human hours are not spent on making the decisions. They're m- m- they're on gathering data or implementing things, but then there's a select few hours where it's like, here is where we're gonna, you know, make this irreversible decision on the strategy. Um, and those people that make those decisions are also typically paid a lot in a company.

    26. HSHarry Stebbings23:30

        Yeah, but the only... Yeah, but the assumption there would be then that you'd have to increase that spend for that 10% even higher.

    27. MGMatan Grinberg23:37

        Poten- And that's what's happening already is, like, the frontier models are c- sometimes they're getting more expensive or you're using the ultra-high reasoning or, you know, this ty- this type of thing. Um, and so it's like, okay, if these, if this planning thing is the very key thing, we'll spend on it. But it doesn't necessarily mean that most of your tokens are going there. It's just for certain key steps, maybe you want to spend a lot, um, and it's worth that allocating the, the budget there. Um, but then once it comes to, okay, we have the plan, now let's implement, the open models are typically really good.

    28. HSHarry Stebbings24:06

        When we think about, um- What you're willing to spend on. I asked Brandon how much it costs to hire great AI researchers-

11. 24:14 – 27:30

    Factory's controversial culture: sales and engineering as one team

    1. MGMatan Grinberg24:14

       Yep

    2. HSHarry Stebbings24:15

       ... and he was like, "Tens of millions of dollars."

    3. MGMatan Grinberg24:16

       Yep.

    4. HSHarry Stebbings24:17

       Have you found the same, and is it impossible to hire great AI researchers in competition with Anthropic and OpenAI?

    5. MGMatan Grinberg24:24

       We are a very, uh, opinionated organization, and so, uh, it's very self- like the people who like the opinionated stances that we take are willing to not necessarily go and try to, you know, maximize the dollars that they can get out of, in the market. Um, that said, it is still, yeah, pretty competitive though.

    6. HSHarry Stebbings24:44

       What do you think is the strongest opinion that you have that most people disagree with?

    7. MGMatan Grinberg24:48

       I would say the opinion that we have that I think in the space that we are in is the most controversial, um, is, uh, the way that we treat what product is at Factory. Um, I think there's some very commonly held beliefs at the labs or at some of our competitors who are also doing kind of software development, and this is honestly growing up in the Bay Area. There's a very common Silicon Valley fallacy, which is there's like researchers, like the pinnacle, and then there's engineers who implement the research. You know, they're not quite there, but you know, they're still great. And then there's sales and marketing and all that dirty stuff. Ugh, if only we could build a better product and it would sell itself and we wouldn't need to deal with, you know, sales and marketing. Um, and it's just completely delusional. Like, the product at Factory is the entire journey from the very first time they hear our name till their 10th renewal after a decade of being a happy customer. Now, the software is a big part of that journey, but so too is the, the marketing that we do and the people that we have running that. Same with the sales pros, the people that present themselves in discovery calls or in demos or in solution engineering. Like, that entire thing is the product. Everyone is first class. It's not like we have engineers who are like wholly at the office and you're not allowed to speak to them unless you're an engi- it's like, no, no, no. We have engineers and salespeople sitting next to each other. There are no like engineer corner, sales corner, or any of that stuff. It's like everyone is completely intermixed. Um, when salespeople close a deal, engineers say, "We closed a deal." When engineers ship a feature, salespeople say, "We shipped a feature." It is entirely one team, entirely cohesive. Everyone-- There's no like first class or second class, and this is shockingly controversial, um, in particular in the Bay Area and in particular in coding or in AI. It's messed up. It's so messed up. And I think that the reality is, uh, it will come to haunt some of these companies one day because I think right now where there's a gold rush and everyone's like desperate to sign, you know, and, and get more tokens from these people, it's easy. They're kind of like... In my mind, it's kind of like they're astronauts in space where there's no gravity. Your muscles will atrophy. Gravity will come back, and if you don't have a good sales and marketing team 'cause you don't give it respect, the second gravity returns, all of your muscles will be atrophied and you won't be able to compete. And I would say this. Name a legendary company that has a shit sales or marketing team. You can't.

    8. HSHarry Stebbings27:10

       No, but I can name companies that have shit products but great sales and marketing teams. That's, that's the ironic thing on the flip side.

    9. MGMatan Grinberg27:16

       Exactly. And in fact, [laughs] it seems like-

    10. HSHarry Stebbings27:18

        Many more

    11. MGMatan Grinberg27:18

        ... most-

    12. HSHarry Stebbings27:18

        Many more

    13. MGMatan Grinberg27:19

        ... like most legendary-

    14. HSHarry Stebbings27:20

        Many more

    15. MGMatan Grinberg27:20

        Yeah. Yeah

    16. HSHarry Stebbings27:21

        Like-

    17. MGMatan Grinberg27:21

        Yeah

    18. HSHarry Stebbings27:22

        ... I'm not gonna name them because I'll get in trouble. [laughs]

    19. MGMatan Grinberg27:24

        [laughs] We're, I'm sure we're thinking of some of the same ones.

    20. HSHarry Stebbings27:27

        I mean-

    21. MGMatan Grinberg27:27

        Yeah

    22. HSHarry Stebbings27:27

        ... but I mean y- yes.

    23. MGMatan Grinberg27:28

        Yeah.

    24. HSHarry Stebbings27:29

        Ch- Cha- if Chad Peters were

12. 27:30 – 32:28

    Why agency matters more than credentials

    1. HSHarry Stebbings27:30

       here, he would say them.

    2. MGMatan Grinberg27:31

       Yeah.

    3. HSHarry Stebbings27:31

       Um-

    4. MGMatan Grinberg27:31

       Yeah, that's right.

    5. HSHarry Stebbings27:32

       Does what it takes to be a great engineer change when you essentially become prompter and manager of agents-

    6. MGMatan Grinberg27:38

       Yes

    7. HSHarry Stebbings27:39

       ... versus creator and doer of tasks?

    8. MGMatan Grinberg27:41

       Yes. It very seriously changes. And this is actually why we're selecting for very intentionally, like this culture that we just mentioned is really important because the best engineers are gonna be the ones that don't see sales and marketing as dirty work, but as, again, an important part of the product. Because as an engineer, you're no longer, you know, just your job is ship feature. It's no. You are owning full end-to-end outcomes of here's the way a customer is behaving. Here's how maybe we can change that behavior that makes them, uh, you know, uh, uh, a better user long term. It makes them more agent native. They get more out of our product. We can then follow them through that journey, enable the salespeople so they know how to talk about it or they know how to demo it. This like, this is like a full stack engineer that goes way beyond just engineering, but into sales, into marketing, into enablement and all that. Um, and those are the parts of engineering that really, really matter. Those are the parts that have made engineers typically good founders is when they have that. And the parts of engineering that become less important are funny enough, the things that the Silicon Valley has really bragged about a lot, which is like competition winning or like Olympiad type. Are you like as fast as possible at coding? Are you-- Do you memorize all the different nuances of these different languages? Those are the parts that don't matter. Like, if you memorized some coding lang- or some, you know, syntax of a coding language that someone else didn't, it doesn't matter.

    9. HSHarry Stebbings28:59

       People like not the credentialism, but I think they misunderstand VC mindset right now. And as a VC, I'm happy to share how you feel, w- how we feel, which is just like fundamentally there is intense uncertainty around what Anthropic and OpenAI will do and who they will kill at the application layer. And so in a world where we desperately seek certainty, we look for validators [laughs] and the validators of someone being a math Olympiad or, uh, you name it, whatever that is, that validation, in the wake of not having other certainty, that serves as a good crutch.

    10. MGMatan Grinberg29:34

        Yes. But it's a crutch. It's a crutch. It's helpful. It's a good indicator. Like generally, you can't win competitions if you're dumb, right?

    11. HSHarry Stebbings29:41

        [laughs]

    12. MGMatan Grinberg29:41

        Like it's pretty, it's pretty rare. Um, however, for these types of engineers that we're looking for, it's like that's cool, but that's kind of irrelevant. Like what have you built? How have you taken ownership and agency of things end to end? And this is not like we have people on our team that have won Olympiads-

    13. HSHarry Stebbings29:58

        Mm

    14. MGMatan Grinberg29:58

        ... and I think they're great and it's fantastic and a lot of my friends have. But there's also a certain like, especially there are some high schools that like really focus on like you must do the math Olympi- like you must do this, the AMIE to then go to the IMO and like this is the path to success, where actually that's kind of anti-signal. 'Cause there it's like you're not owning your fate or choosing your agency. You're kind of going through the funnel. But then there are people on our team who are like from the middle of nowhere where no one else in their high school ever did this stuff and they kind of took the agency of like- I think this is really fun. I'm really competitive. I wanna compete at this." And they kind of go and do those competitions on their own. Those are when the signal is still positive for this kind of engineer of the future.

    15. HSHarry Stebbings30:36

        I don't think I've told you this, but do you know who you always remind me of?

    16. MGMatan Grinberg30:39

        Who?

    17. HSHarry Stebbings30:40

        Matt Damon in Good Will Hunting. [laughs]

    18. MGMatan Grinberg30:42

        Oh, that's so... I mean, I'll take that to the bank.

    19. HSHarry Stebbings30:44

        Have you, have you, have you not been told that before?

    20. MGMatan Grinberg30:46

        Let me, let say that one more time into there.

    21. HSHarry Stebbings30:48

        You, you, you look identical.

    22. MGMatan Grinberg30:49

        That's very kind. I'll take that.

    23. HSHarry Stebbings30:50

        It's, it's... But we're gonna put up, like, an image here-

    24. MGMatan Grinberg30:52

        Yeah

    25. HSHarry Stebbings30:52

        ... and then you're gonna do a side by side.

    26. MGMatan Grinberg30:53

        Yeah. All right, we have a clip.

    27. HSHarry Stebbings30:54

        And, and-

    28. MGMatan Grinberg30:54

        This is the clip. This is the, this is the starter to the show

    29. HSHarry Stebbings30:57

        ... and, and they're gonna be like, "Aha, I totally get it."

    30. MGMatan Grinberg30:59

        Yeah. [laughs]
13. 32:28 – 35:06

    The age of the polymath is back

    1. HSHarry Stebbings32:28

       in every function?

    2. MGMatan Grinberg32:29

       The age of the polymath is back.

    3. HSHarry Stebbings32:31

       [laughs]

    4. MGMatan Grinberg32:31

       Like, I g- growing up, I was so... Like, I was obsessed with math and physics, and I was so jealous that in, like, you know, hundreds of years ago, people like Da Vinci or Euler or Newton could be polymaths, and it was because their fields were relatively shallow. So, like, chemistry wasn't that built out, mathematics wasn't that built out, physics wasn't that built out. In Da Vinci's case, like, art and engineering and, um, sculpture. Um, and so you could get to the frontier of these disciplines withi- in multiple disciplines within your lifetime. And then growing up in, you know, the early 2000s and 2010s, pre-AI, fields were so deep, in my case, theoretical physics and string theory was so deep that you could spend literally 50 years catching up on all of the literature and academia that's existed before you contribute anything new. And so it was like this was infuriating to me because it was so frustrating. With AI, w- we're now completely the opposite. These tools can get you up to speed to the frontier. Obviously, with a lot of uncertainty about certain details. You won't have the depth of other people, but it'll get you to the frontier way faster than ever before. And so now if you're someone that's good at thinking around constraints, thinking about systems, holding uncertainty in your head and being okay with that, like know- knowing there are unknowns and knowing that you can still push the frontier forward despite that, you can be a polymath. You can push forward and create innovations on how to do developer marketing, while at the same time pushing forward the frontier of, like, token caching for software development agents, at the same time as, like, you know, being an incredible, um, solution engineer. Like, these are things that you can now do all at once. And so this is something that's very top of mind for me. And in our hiring process, we want to find the people that can be those polymaths. Um, the era is totally back. The polymaths are back.

    5. HSHarry Stebbings34:19

       I've had a lot of people say on the show that agent operations will be like the leading, uh, function that doesn't exist today that will be very common in three to five years. Do you agree with that?

    6. MGMatan Grinberg34:31

       What is the definition of agent operations?

    7. HSHarry Stebbings34:33

       Agent operations is the creation of agents and the maintenance of them. So to be able to go into different functions and say, "Ah, social media. I'm gonna create agents that allow you to create, distribute, share posts. Ah, marketing and design. I'm gonna create agents that allow you to create visuals, share them amongst each other, edit them, collaborate on them."

    8. MGMatan Grinberg34:52

       I think to some degree, everyone should be able to do that on their own. But I could imagine a world where there's kind of someone whose job it is, is to, like, find places that aren't as efficient and c- similar to operations now, like in organizations, but now it's just

14. 35:06 – 39:25

    What we'll look back on in disbelief

    1. MGMatan Grinberg35:06

       agentified, so they're using agents to make the organization more efficient wherever possible. But I think in general, if you have people that in certain functions that aren't proactively doing that, it's probably a bad sign.

    2. HSHarry Stebbings35:17

       What do we do today that we'll look back on and go, "Oh my God, I can't believe we did that"?

    3. MGMatan Grinberg35:21

       I mean, for an engineering team, like writing release notes. That's crazy that people used to spend hours of time writing release notes or, like, writing documentation.

    4. HSHarry Stebbings35:29

       So not everyone knows what release notes is. What is release notes?

    5. MGMatan Grinberg35:31

       So it's like, you know, uh, basically cataloging the changes that you've made in s- in the last, whatever, month or so, um, you know, and sending it out to either internal or external to your users. And generally, like, uh, this and, like, documentation. Like Stripe has a really great reputation. They had incredible documentation. Like, so many APIs had horrid documentation. Stripe was like the pinnacle. They were so good at this. Spent a lot of time doing it. Five years from now, it's gonna be like, oh my God, I cannot imagine, cannot believe that these people that get paid so much money spent hours of their time doing this. I think that's something that, you know, we definitely won't do.

    6. HSHarry Stebbings36:07

       Does that reduce the impact of Stripe's great documentation if everyone is equalized?

    7. MGMatan Grinberg36:12

       Uh, yes, but I think Stripe has plenty of places that they can differentiate. And I think it's a better world where everyone has documentation as good as Stripe's.

    8. HSHarry Stebbings36:19

       Totally agree with that.

    9. MGMatan Grinberg36:20

       Yeah.

    10. HSHarry Stebbings36:20

        How does the product review and especially, like, code review process change in the next few years?

    11. MGMatan Grinberg36:27

        Yeah, I think what's cool about, um, this agent native software development is review has been a big problem 'cause basically, uh, you know, first phase of rolling out AI coding tools was, "Oh my God, look how much code we can generate." You know, "It's incredible. I'm generating a ton of stuff."

    12. HSHarry Stebbings36:45

        [laughs]

    13. MGMatan Grinberg36:45

        And then phase, phase-

    14. HSHarry Stebbings36:46

        Fucking shit.

    15. MGMatan Grinberg36:47

        Yeah. Phase two was some poor staff engineer who has to review hundreds of these slop PRs that are like, don't adhere to your standards, are like completely misformatted and all this stuff. Um, but what's great about having, uh, this kind of like full end-to-end software factory as it were, is it's now very clear the ROI of investing in things that make your agents more kind of ready for production. So examples of this are, um, making sure your agents have access to up-to-date documentation, making sure agents can spin up a remote machine so that they're not just generating the code, but they can actually run it and see what the outputs are and iterate based on that to make sure that it's actually good. Um, setting up things like CI/CD or good linters or good pre-commit hooks, these are all things that, uh, the best organizations at like developer experience would invest a lot of resources in, but they would do it because it makes it easier for engineers to work, easier for them to onboard. But the, the impact of doing that well is just like one-to-one kind of correlated to how many engineers you have. With agents, though, the impact of that is now like 10X or 100X depending on how many agents you're using. Because the better your DevX, the better your agent ends up adhering to your standards, which means there's less time that that poor staff engineer has to go through reviewing your PR, which means you're kind of faster throughput in your software development.

    16. HSHarry Stebbings38:12

        When agents are the buyers and you're selling to agents, how does the world change, and does the value of great API increase?

    17. MGMatan Grinberg38:21

        I think, um, that value is increasing, especially because the thing that makes it easier for agents tends to be the same as the things that make it easier for humans. Um, at some point in theory, that could change, where, you know, if you're actually training models or training agents to be as efficient as possible communicating to each other. Um, but then the downside there is it's not as human readable. And so that-

    18. HSHarry Stebbings38:41

        But if you think about agent to agent, agent to agent doesn't give a shit about UI or design, but it does fundamentally care about data structures, um, potential integrations, uh, documentation.

    19. MGMatan Grinberg38:52

        Sure.

    20. HSHarry Stebbings38:52

        Do you know what I mean?

    21. MGMatan Grinberg38:53

        Yeah, yeah. Yeah. So I think one thing that if you don't have careful, um, standards in place, it can get bloated pretty quickly. But I think the best organizations who are the most agent native actually put in a lot of guidance on like, here's like the UI side of things and how things need to be. Um, being very aggressive about like pruning anything that's unnecessary, making sure there's not like, you know, bloated, like, I don't know, comments in all of your code that's like kind of gratuitous or, um... There are ways around it, but that's kind of where the human's job changes a little bit, where their job

15. 39:25 – 40:18

    Why the company is called Factory

    1. MGMatan Grinberg39:25

       goes from... I mean, part of our name, our name is Factory. Um, part of why it's called Factory is because the future of software development is where these organizations, instead of having engineers that build the software, they're gonna have engineers that build the factories that build their software. Visually, whenever I say this, I always think of Tesla's factories. I don't know if you've ever seen videos of the inside of Tesla's factory.

    2. HSHarry Stebbings39:46

       Yeah.

    3. MGMatan Grinberg39:47

       It's all these like robotic arms going, and you have the assembly line going through. Um, and there might not be as many humans in that assembly line, but you know damn well that humans, uh, designed this process to optimize the throughput to, you know, produce more Teslas in this case. Um, and so in this new world of software development, humans, human engineers are not gonna be involved as much in like writing the actual code, but they're the ones that are gonna be involved in how do we make sure it's not just creating all this bloat or it's technically getting the job done and passing tests, but doing it in a way that is really dramatically increasing

16. 40:18 – 44:21

    Labour displacement and the problems AI will finally solve

    1. MGMatan Grinberg40:18

       debt. Um, so there's, they're kind of like building the scaffolding around this factory that produces their software.

    2. HSHarry Stebbings40:24

       Do you worry about labor displacement when we move from working in the factory to working on the factory?

    3. MGMatan Grinberg40:29

       Short-term, yes. Long-term, no. Short-term, yes, because it's just a shock to the system where, you know, there are all these big layoffs that are happening that are pretty aggressive and, you know, these are thousands, tens of thousands of people that had a job that no longer do. And so I think that does worry me. Um, long-term though, I am very not worried because the reality is there is a huge number of problems in the world, ridiculous number of problems in the world, and a large percent of them can be solved or can be helped with software. And very few of those problems that can be solved with software are we currently solving with software. And so if we're going to be flooding the job market with tons of engineers, that means that we can now allocate them on the broader economy to solve more of these problems in the world. And if we have more engineers who are going and solving more problems in the world, that is a net good.

    4. HSHarry Stebbings41:21

       What problem is not currently being solved with software that will be enabled by this new technology? 'Cause everyone's like climate change, and I'm like, great.

    5. MGMatan Grinberg41:30

       Yeah.

    6. HSHarry Stebbings41:30

       You know how many people I've found doing climate change technology? Wow.

    7. MGMatan Grinberg41:34

       Yeah.

    8. HSHarry Stebbings41:34

       None.

    9. MGMatan Grinberg41:35

       Yeah. Well, and maybe part of that is because all, like the Googles have, have been hiring all these engineers. So distributing great engineering talent to more problems I think is gonna be a good thing. The economy has to match though and properly incentivize them, and that's something that I think will take a little bit of time, which is like the intermediate period. But like so many health problems, like so much of pharmaceutical research can be advanced with better engineering. And like the thing that really upsets me with some of the people who are talking about, you know, pausing AI development or any of this like, oh, it's, you know, it, it's a bad thing and it's gonna, you know, harm society. Dementia is kind of a go-to example where everyone understands how big of a deal that is. That is something that can be solved with better AI and better software. Like it's a matter of time. Like we will solve it and we can solve it. And by saying you wanna slow down AI, that's saying like these people who have relationships with loved ones who have dementia, you're like, "No, no, no, sorry, you guys, you gotta maintain that relationship for a little bit longer. We're scared. We don't know about AI." I think it's like, it's pretty, it's pretty harmful and it's pretty selfish to say that it's something that... To me, it doesn't make sense. It doesn't make sense.

    10. HSHarry Stebbings42:36

        Do you agree with government intervention?

    11. MGMatan Grinberg42:39

        Uh, in what capacity?

    12. HSHarry Stebbings42:40

        I- in free markets, when you think about, like, the allocation of resources, there are times when it is suboptimal from a human morality societal standpoint in a lot of cases to see, uh, engineers at Anthropic working on optimizing core code when they could be working on optimizing healthcare systems-

    13. MGMatan Grinberg42:58

        Yeah

    14. HSHarry Stebbings42:58

        ... or optimizing more critical or mission-critical things immediately.

    15. MGMatan Grinberg43:02

        Yeah.

    16. HSHarry Stebbings43:03

        Governments can intervene, offer subsidies-

    17. MGMatan Grinberg43:06

        Sure

    18. HSHarry Stebbings43:07

        ... offer economic incentives. Do you agree with that or do you believe in Adam Smith's invisible hand?

    19. MGMatan Grinberg43:11

        Um, I think it's a, it's u- it's certainly useful in some cases. Like, I don't think anyone would argue that the government should never intervene ever in the economy because there are some things, especially as it relates to, like, military uses or safety or things like, like weapons, like you're definitely gonna need, you know, some involvement there. Um, I think there's some incentivization that can be helpful just because, uh, there's, there might be some problems for a society that maybe capitalism doesn't see the immediate feedback loop of, and so you might wanna juice the incentives a little bit to get an outcome that you're looking for. But I think generally I'm pretty reluctant. I think you need to have a very good case for why you need to do that. Even like the example of climate change, um, you know, talking about that one. Um, it's obviously a very sensitive subject or a very important subject for a lot of people. And, you know, you could make the case that the faster we develop AI, the sooner we solve climate change, 'cause AI, you know, can s- help us solve a ton of these problems. But to develop AI faster, you might need to consume fossil fuels and emit them and, you know, that emits CO2 into the atmosphere. And so the question is, like, you know, short term it might be slightly worse, but it ends up getting us to solve the problem way sooner instead of dragging it out over 50 years or 100

17. 44:21 – 45:51

    Are we in an AI bubble?

    1. MGMatan Grinberg44:21

       years. And so there's some of these cases where the natural kind of free market will incentivize it the right way, and there's some cases where it won't, but I think you would need to be very ve- very careful about the cases where you do want the government to say, "Hey, we wanna step in here."

    2. HSHarry Stebbings44:33

       Do you think we are in an AI infrastructure bubble?

    3. MGMatan Grinberg44:36

       M- maybe there's, like, some short-term blips, but, like, long-term, absolutely not. Like, not even close. I think there might be s- there might be similar corrections to, like, this thing at Uber where, oh, we were going a little haywire. We weren't allocating it appropriately. And then it's like, okay, let's l- yo- lower consumption a little bit, but, like, on the net, absolutely not.

    4. HSHarry Stebbings44:53

       What bottleneck do we have today that will be completely solved within a few years?

    5. MGMatan Grinberg44:58

       I think the biggest bottleneck by far, working with all these organizations, is the human side of it. It's just like behavior change. Like, especially if you're-

    6. HSHarry Stebbings45:05

       And what you're saying there is like, is selling to large enterprises and how they do change management?

    7. MGMatan Grinberg45:09

       Yeah, or even y- on an individual level. Like, if you're an engineer who's been an engineer for 30 years, it's hard to change those patter- like you're stuck in your ways to, to a certain degree. Um, but there's also a funny thing where some of these engineers who've been engineers for a very long time or who have been engineering managers, they might be more reluctant to use these tools, but sometimes they're better because they know how to delegate. They know how, they know how to deal with some of the, the junior engineers where if you tell them the wrong thing, they're off in the cave doing the wrong thing for seven days. They come back with something completely useless. And then on the other end of the spectrum, there are people earlier in career who don't have as much of a standardized workflow that they're used to, so they're more eager to adopt these new workflows, but they don't know how to manage people. They don't know how to delegate as well. So there's kind of an interesting

18. 45:51 – 47:46

    Lessons from selling to enterprises

    1. MGMatan Grinberg45:51

       balance there.

    2. HSHarry Stebbings45:51

       When you look at now, you sell to some of the largest enterprises in the world in some cases. Um, what do you know now about selling to large, large enterprise that you wish you could tell young Matan two years ago?

    3. MGMatan Grinberg46:04

       So this is the first job I've ever had, which I think is always a funny thing to say. Um, 'cause prior to this I was a theoretical physicist. Literally never, never like coffee shop, any of that, literally never have had a job. Like, never have been paid to do anything aside from physics until this, which is, it's a whole separate thing. But-

    4. HSHarry Stebbings46:20

       All right, Matt Damon. [laughs]

    5. MGMatan Grinberg46:22

       [laughs] But I will say the thing that has been the craziest learning, and this is obvious to anyone who's in sales or like Chad and Chris, you know, to them it's obvious. To me, the thing that was the most kind of visceral altering thing was, um, meeting people face to face makes such a big difference if you're trying to sell them something. But also you should never try to sell something. You should only try to understand their problems and see if the solution that you might have can actually help them solve that problem. That's another thing. Like if you go in a conversation trying to sell something, especially to engineers, like don't waste your time. If you go in trying to have genuine curiosity about... And it, it's really easy because these organizations do their engineering so differently, and I find it fascinating how like all of these different banks, you know, consulting firms, pharmaceutical companies, they have the most different ways of building software and it's really interesting to go talk to them and to understand it. And the best way of talking about it with them is face to face, and people love talking about their problems and they love talking about all of the bureaucratic nightmares that they have to deal with. And then by understanding all of that, you can actually get a sense, you know, is our software a good fit for them? Will it help solve their problems? Um, it's also just so fun to then like meet up with them a year later and be like, "Ha, remember when you had to deal with that bullshit, and now you don't have to?" And that's just such a rewarding feeling of, you know, making their lives better in that way.

    6. HSHarry Stebbings47:42

       In terms of like being there in person and the sales process,

19. 47:46 – 50:46

    From string theory to Factory: the origin story

    1. HSHarry Stebbings47:47

       you got Sequoia very, very early on. Sequoia obviously one of the best and most prominent investors. Can you just tell me the story of how you got Sequoia having never had a job and only being paid to do physics?

    2. MGMatan Grinberg47:57

       Yeah. So I was obsessed with physics basically since I was 12 because, um, I was a bad student, uh, and my geometry teacher told me that I had to retake geometry in high school. And like, I never tried in school, but I always prided myself on being good at math. And when she told me that, I was like, "Are you kidding me? She thinks I need to retake geometry. Like, I'll show her." And so my first order on Amazon ever was textbooks for algebra two, trigonometry, pre-calc, calculus one, two, and three, uh, differential equations, um, and maybe a linear algebra textbook So I bought those textbooks, and then the summer between middle school and high school, I studied all of those, like did all the problems in all of them. Um, and then in high school took exams to place out of all of those classes. Um, and then I asked my dad what the hardest math was. He said string theory, which is technically physics not math, but I was like, "Okay, I'm gonna be a string theorist." And that was literally all I cared about for basically the next 12 years of my life. All I cared about was math and physics. Um, ended up going to, uh, to Princeton because they had a great physics professor I wanted to work with. Um, he's, uh, this famous professor named Juan Maldacena, and I was like the first undergrad to work with him and write a paper with him. Then I ended up coming to Berkeley to do my PhD and, you know, work with a great advisor there, and then only at Berkeley I realized, like it kind of all comes crashing, like, "Holy shit, I've just been doing this because it's hard and because someone said I couldn't do it. Like, what the hell do I do with the rest of my life?" Like, this is crazy. Like, everything came crashing-

    3. HSHarry Stebbings49:24

       What caused that crashing down moment? And, and why did it take so fucking long? [laughs]

    4. MGMatan Grinberg49:30

       Um, partly, partly-

    5. HSHarry Stebbings49:30

       12 years. 12 years. You're slow. [laughs]

    6. MGMatan Grinberg49:32

       I have, I have tunnel vision. When I get obsessed with a problem, it is all I think about.

    7. HSHarry Stebbings49:36

       I was at law school for two weeks. [laughs] It was a quick, quick realization.

    8. MGMatan Grinberg49:39

       See, some people are faster. You know, I wasn't as... I wasn't quite as quick. Honestly, part of it was being a, a, as part of a grad student at Berkeley, you have to teach classes. Um, and I was teaching a class to like, whatever, 18-year-olds who didn't give a shit about physics. And I was like, "Oh my God, this would literally be the rest of my life." Just like sitting and doing lectures and doing these classes. On Rate My Professor, I think I had a one out of five. I was like horrible. Like, I didn't... Yeah, it wasn't, it wasn't a good fit. Um, but it was a kind of this existential crisis, like, what, what do I do? And so, um, you know, kind of realized it was probably gonna be either quant finance, which is what a lot of math and physics people do, um, big tech or startups. Um, I ended up doing the quant finance interviews like every good physicist does, and, you know, almost took it, almost went to New York to do it, and then last second I had a, a advisor that I spoke to who was like, "You know what? Stay at Berkeley for a bit. Don't do it. You're always gonna be good at math. You could always go and do quant finance. Stay at Berkeley, explore, learn some stuff, whatever." So I was like, "Okay, fine. You know, I'll, I'll do that," you know. So ended up taking my first CS classes at Berkeley. I'd learned, like, to code for physics for like simulations and all this stuff,

20. 50:46 – 52:30

    Discovering code that writes itself

    1. MGMatan Grinberg50:46

       but never in a formal class. And I'm very competitive. And I found that in these classes I was doing better than some of the CS students, which was very competitively satisfying. It was like, "Oh, okay. I'm gonna take more of these." And then it wasn't until I took a seminar, um, in what was called program synthesis at the time, now we call it code generation, and it just completely nerd sniped me. 'Cause the idea here is not machine learning for video or audio or images, but it's code with the explicit purpose of creating itself. And there's something just so fundamental about that. And a decade of physics, like physicists and mathematicians, they're never interested in the case of like N equals three, or like N equals four, four-dimension. It's always like what is the N-dimensional solution? What is the arbitrary, the fundamental, you know, solution to things? And there was something so fundamental about this idea of like code generating itself, and it just got me obsessed. So I stayed at Berkeley, and for the next year that was kind of what I spent my time on. My advisor was very chill and just allowed me to just, you know, take AI courses. Um, and eventually I realized that the way to actually solve this problem was not in academia, um, but in the industry. Um, and to properly solve it in the industry, you would have to start a company. Um, but I knew nothing about starting companies. So, 'cause again, all I cared about was math and physics, didn't know anything about this. So what, uh, does someone who wants to learn about starting companies do? Well, they order on Amazon Peter Thiel's Zero to One, and they look up on YouTube how to start a company. [laughs] And so, and so, uh, you know, read Zero to One. Incredible book. I know it's so cliche, but like to someone who didn't kn- growing up in the Bay Area, shockingly, I just like did not care about any of that. And reading this, it was like so concise, beautifully written, all that. You know, loved that. And then, you know, was watching these videos, a lot of them like Y Combinator, you know, videos

21. 52:30 – 55:30

    The cold email and 3-hour walk with Sequoia

    1. MGMatan Grinberg52:30

       and all this stuff. And then I stumbled upon this, uh, I think it was like a Stanford VC club podcast, um, with this guy whose name I recognized because at Princeton when I wrote that paper with Juan Maldacena, I had cited one of his papers. So it was a, a theoretical physicist. Like, I remember this guy's name, but he was on this podcast talking about how he sold a company for a billion dollars and was a investor at this place called Sequoia. And he also in this video seemed like pretty sociable and normal, which I don't know if you've interacted with-

    2. HSHarry Stebbings53:01

       I'm not sure, dude. [laughs]

    3. MGMatan Grinberg53:02

       Theoretical phys- yeah. You know, compared to theoretical physicists though, like he can maintain eye contact, you know, he was somewhat normal.

    4. HSHarry Stebbings53:08

       [laughs]

    5. MGMatan Grinberg53:09

       Very rare.

    6. HSHarry Stebbings53:09

       That's such a low bar. [laughs] He can, he can maybe hold himself in a social setting.

    7. MGMatan Grinberg53:14

       Yeah. And so I was like, "Okay, who is this guy?" You know, "I gotta talk to him." So I ended up writing him an email being like, "Hey, I'm Matan. I also used to be a physicist. I wrote a paper with Juan." Didn't say the last name 'cause it's like if you know, you know. Um, "Would love to get your advice." And, you know, he responded that day and invited me down to Sand Hill, and it was supposed to be a 30-minute meeting. Um, but we end up going on this walk and ends up being a 3-hour walk. Uh, and on this walk it turns out we had very similar reasons for getting interested in physics, very similar reasons for leaving physics, and at the end of it he was basically like, you know, "It was great to meet you, Matan. You absolutely need to drop out of your PhD, and you should either join Twitter right now because Elon just took over and it's hardcore, like for your resume if you like voluntarily go there, or you should start a company." And I was like, "Okay, thank you so much. I appreciate you taking the time. Like, I'm gonna, you know, I'm gonna go think about it." But in the meantime, I like had already known about Factory. I was just-- I didn't wanna ruin the meeting with like a pitch. Um-

    8. HSHarry Stebbings54:09

       You didn't wanna transactionalize this.

    9. MGMatan Grinberg54:10

       Yeah.

    10. HSHarry Stebbings54:10

        Yeah.

    11. MGMatan Grinberg54:10

        'Cause it was so, like it was incred- like we had the exact same reasons for getting interested in physics and all that.

    12. HSHarry Stebbings54:14

        I totally get it. Yeah, yeah.

    13. MGMatan Grinberg54:16

        So like didn't wanna, didn't wanna dirty it with any of that.

    14. HSHarry Stebbings54:17

        No, it's like kinda like an LP where at the end you're like, "I don't wanna ask for a track." [laughs]

    15. MGMatan Grinberg54:21

        Yeah, exact- exactly.

    16. HSHarry Stebbings54:22

        Yeah. [laughs]

    17. MGMatan Grinberg54:22

        And, and then the crazy thing, the next day I go to a hackathon in San Francisco and see across the room this guy who also went to Princeton who I, like, recognized but I didn't, like, know super well. End up talking to him. He's also interested in this problem. We like, we joke that it was like intellectual love at first sight. This is my co-founder, Eno. Um, and basically that day forward, we spend like every day talking nonstop. I had some shitty demo that I built. Eno is 1,000X better of an engineer than I ever will be. Um, and so he and I for the next like 72 hours like put together this better demo, and then I call up this investor and I'm like, "Hey, I have something cool I wanna show you." So we hop on a call and I show him this demo, and I'm like, "What do you think?" He's like, "Eh, it's okay." I'm like, "Are you fucking kidding me? This is gonna change the world. What are you talking about?" He's like, "Okay. Well, would you work on it full time?" And I was like, "Yeah, absolutely." He was like, "Okay, drop out of your PhD and send me a screenshot." And keep in mind, like my parents immigrated from the Soviet Union to the United States with basically nothing. [laughs] The, the fact, the fact that I was doing a PhD to them was like their pride and joy, like it was the thing they were the most proud of. Um,

22. 55:30 – 1:01:19

    Dropping out and the $1M check

    1. MGMatan Grinberg55:30

       [sighs] but there was so much momentum, [laughs] so much momentum, I'm, you know, I'm-- he answered my email, we got along well, I met the co-founder the next day, and I was like, you know what? Fuck it. Dropped out, sent him a screenshot, and he was like, "All right. You have a meeting with the Sequoia partnership tomorrow morning. Be ready to present."

    2. HSHarry Stebbings55:47

       You've, you've never presented to a, a venture partnership before.

    3. MGMatan Grinberg55:50

       No.

    4. HSHarry Stebbings55:50

       So, so what happens?

    5. MGMatan Grinberg55:52

       I made a shitty deck, put some slides together.

    6. HSHarry Stebbings55:53

       We, we, we, we go to the Sequoia HQ-

    7. MGMatan Grinberg55:55

       Yeah

    8. HSHarry Stebbings55:55

       ... put some slides together.

    9. MGMatan Grinberg55:56

       Keep in mind, I didn't even know who the hell they were. Like I didn't-- it was just like, oh, who are these random people? Like, okay, whatever. Yeah, I'll go talk to them. I wish it was recorded. I wish it was recorded because I'm sure I came across as so arrogant.

    10. HSHarry Stebbings56:06

        How did it go?

    11. MGMatan Grinberg56:08

        I thought it went fine. They asked some questions. I think I was pretty, uh... Again, I didn't know anything about VC land or startup land or any of that stuff. So like retrospectively, I know like Alfred and Pat and Roloff, they were all like in there. They're all asking questions, and I was probably dismissing some of them. "Oh yeah, we'd solve that easily. We'd do this, we'd do that." Um, keep in mind, this was in, uh, April of 2023, so this was like way before anyone was thinking about agents, way before people were even using Copilot. We were talking about fully autonomous software development agents. Um, and, uh, it was kind of a blur. Uh, and, uh, you know, the next day Sean calls me and he's like, "Hey, we wanna give you a check."

    12. HSHarry Stebbings56:45

        How big was the check?

    13. MGMatan Grinberg56:46

        A million dollars.

    14. HSHarry Stebbings56:47

        A million dollars?

    15. MGMatan Grinberg56:49

        And you know what he gives me shit for? You know what the terms?

    16. HSHarry Stebbings56:52

        Five post?

    17. MGMatan Grinberg56:53

        Five post. [laughs]

    18. HSHarry Stebbings56:57

        I mean-

    19. MGMatan Grinberg56:59

        Yeah

    20. HSHarry Stebbings56:59

        ... why-- I'm not being rude. Why did they bother doing a partnership meeting? [laughs] Like in the nicest way, that's like a coffee. Like I know it's a dick comment-

    21. MGMatan Grinberg57:07

        Yeah

    22. HSHarry Stebbings57:07

        ... but like when you manage, like $7 billion-

    23. MGMatan Grinberg57:09

        It was a different time. It was a different time. Early 2023 was a different time. Things got crazy.

    24. HSHarry Stebbings57:13

        I think you bought 20% post.

    25. MGMatan Grinberg57:15

        Yeah.

    26. HSHarry Stebbings57:16

        Just for, for, you know, listeners, I mean, on, on last funding round, that'd be like a $300 million position, not including dilution.

    27. MGMatan Grinberg57:24

        And it was one of those things, a lot of my, a lot of people I spoke to were like, "You should go shop that around. You could get better terms 'cause it's Sequoia." And it's just like when you have a connection like that, there's a certain thing to me where like obviously you want to maximize, you know, the, the, uh, position for the business, but like no one else would've believed in me except him. No one else would've understood. Like f- I literally had never had a job before. No, like no other partner I would've met... Retrospectively, it's like, oh yeah, whatever. No one else would've done it, and it's one of those things where like trust and loyalty and like belief, to me that matters so much more than like the price tag you get or whatever. I wanna make sure that the people that I have in my corner, because we're building a legendary company, it's not just gonna be 10 years. This is like a lifetime.

    28. HSHarry Stebbings58:03

        Would you tell founders to take a discount for Sequoia?

    29. MGMatan Grinberg58:06

        So generally, yes. I mean, they're the best firm. In particular, if there's like a special connection with you and the partner or there's a special reason why them in particular. But I think what really matters is you wanna have people that are there for you when the days are tough and when it's not obvious. Because when you're a hot company raising a hot round, everyone's your best friend. Every-- It is their job to make you feel special, and they are really good at it.

    30. HSHarry Stebbings58:28

        What's the best way someone's tried to woo you?
23. 1:01:19 – 1:02:39

    Does Ivanka Trump add value as an investor?

    1. MGMatan Grinberg1:01:19

       days. Then, you know, we were introduced, we got along really well. Um, and so then that was how the connection was made there.

    2. HSHarry Stebbings1:01:25

       Does Ivanka Trump provide value? People will look at it and be like, "Oh, branding, just a name, whatever." And I didn't mean that disparagingly at all.

    3. MGMatan Grinberg1:01:32

       Yeah.

    4. HSHarry Stebbings1:01:33

       I think people often think that with kind of famous celebrity names.

    5. MGMatan Grinberg1:01:36

       Fair.

    6. HSHarry Stebbings1:01:36

       Does she actually provide value?

    7. MGMatan Grinberg1:01:37

       Yes. She is-- First of all, she's one of the kindest and smartest people that I've met. And like, there are people that you meet that, you know, are famous that are kind of like a letdown or like, "Oh, they're different than I expect." She is genuinely so kind, so intelligent, and like people just in-- throughout tech, throughout the world really love her, and for good reason. And she has an incredible network. She's so generous with her time. Like, there is like kind of dirty work investor help that she helps out with that some other investors who are more known as investors do not do. And so sh-she and the firm more broadly really earned that right, um, on the cap table.

    8. HSHarry Stebbings1:02:18

       That's really good to hear. I, I hate the statement I, I, I'm, I'm not sure Ivanka was quite your hero. Uh, but like, like, you know, people say never meet your heroes.

    9. MGMatan Grinberg1:02:26

       Yeah.

    10. HSHarry Stebbings1:02:26

        They'll always disappoint. And I think that's just total bullshit.

    11. MGMatan Grinberg1:02:29

        Yeah.

    12. HSHarry Stebbings1:02:29

        I remember meeting Doug Leone, who was one of my heroes, did not fucking disappoint.

    13. MGMatan Grinberg1:02:33

        Yes. Oh, yeah.

    14. HSHarry Stebbings1:02:33

        Like, I left being more like, God, he should have been even more of like a, a, a poster boy for me 'cause

24. 1:02:39 – 1:07:45

    How the coding market matures

    1. HSHarry Stebbings1:02:39

       he was-

    2. MGMatan Grinberg1:02:39

       Oh, yeah

    3. HSHarry Stebbings1:02:39

       ... so great.

    4. MGMatan Grinberg1:02:40

       Oh, yeah.

    5. HSHarry Stebbings1:02:40

       So I, I, I totally agree with you that. That's very funny. I would love just your thoughts on some market composition that I'm struggling with.

    6. MGMatan Grinberg1:02:46

       Yeah.

    7. HSHarry Stebbings1:02:46

       Which is like when you look at cognition, you look at Claude Code, you look at Codex, you look at Curse, now with Groq, how does this market evolve and mature? Is this an AWS, Azure, you know, GCP? Is this an Uber, Lyft? What is the mature state of this market?

    8. MGMatan Grinberg1:03:01

       Yeah. So I think what is necessary for the best outcome for the consumers is going to be models that are separate from the applications. You as a consumer do not want to, um, use applications that are provided for you by the same people that are giving you the model because the incentives are misaligned. We get often-

    9. HSHarry Stebbings1:03:18

       And the incentives are misaligned why? Just so people-

    10. MGMatan Grinberg1:03:21

        Because if, let's say, uh, the example of coding, like if I'm a model provider and I'm working with a large enterprise and I'm giving you a coding tool, I want you to use as many tokens as possible 'cause I'm an API business and I get more money the more tokens you use. Um, and I don't have a huge incentive to be more token efficient, um, other than like, you know, yeah, I wanna give a good product experience, but not strong incentive. Um, versus if you have model providers and you have an application layer that allows that enterprise to decide between the different providers, if you're a model provider, you better damn well be the best or the cheapest or the fastest, or else you'll never get tokens through to you. So it puts the best incentives on the model providers. There's that independent, um, agent in our case there. Um, and then that gives the best prices to the enterprise. It also gives them the best in terms of like if one model is really good at this language or that language, it allows them to kind of adjust between them. Um, and in a world where they're-- you're like vendor locked in, then you can slowly get like laziness and slower shipping and, uh, and as a, you know, consumer, you end up getting a worse experience.

    11. HSHarry Stebbings1:04:26

        Okay, so it's not good for the consumer if the model is tied to the application. Okay, cool. But bluntly, we are seeing Codex and Claude Code eat a huge part of the market. What does the market look like in three years in terms of market maturation?

    12. MGMatan Grinberg1:04:42

        This is gonna be different from cloud. I think cloud, a lot of people suffered because, you know, the cloud providers came and said, "Hey look, sign this three-year deal. We're gonna give you a big discount. We'll get everything good for you. It'll be all right. Come on in." And then they would do that, and then they would jack up the prices, and once you're standardized on one, it's gonna take you two years to switch to something else. So good luck, you're stuck with us, and we're gonna charge you more. Everyone has scars from that. So now every CIO I speak to is really keenly aware of we cannot, you know, throw our lot in with just one model provider. We're gonna need to be agnostic. And so, you know, you could be agnostic by saying, "Hey, every engineer, we're gonna give you Claude Code and Codex and Gemini CLI and all these other tools." Um, but then the problem is now you're asking your engineers to use 10 different tools, or you can use someone like Factory where you can use one tool and you can kind of decide, um, kind of like in an auction on a task-by-task basis which model provider do we wanna use. Do we wanna use an open model? Do we wanna use Frontier? You know, which one of those?

    13. HSHarry Stebbings1:05:37

        How-- can you help me understand the paradox of, hey, we need to be more cost efficient with Replit? We're gonna run the same prompt on three models at the same time.

    14. MGMatan Grinberg1:05:45

        Yeah.

    15. HSHarry Stebbings1:05:45

        And I don't mean that di-- there's no diminishment to Replit.

    16. MGMatan Grinberg1:05:47

        Yeah.

    17. HSHarry Stebbings1:05:47

        That's like them providing a great product, but...

    18. MGMatan Grinberg1:05:50

        Well, so I haven't seen them or that use case as much in the enterprise. I could see for maybe consumer use cases where you're not as cost sensitive 'cause you're not doing things at crazy scale, where it's kind of fun to see, oh, I wonder what Gemini does versus OpenAI versus Anthropic. Um, and you know, for some enterprises, if there are things that are like very sensitive or very secure, you might wanna do that. But for a lot of-- like if you're a non-technical person building an internal dashboard, you probably don't need 10 different models to generate different iterations of it.

    19. HSHarry Stebbings1:06:15

        Totally get that. In terms of the market maturation, what happens to the Lovable and Replit market? [lip smack] We saw OpenAI, you know, kind of release a competitive product last night. I just don't know what happens there. Can you help me understand it?

    20. MGMatan Grinberg1:06:31

        It's not obvious to me. Um, and part of it is because not too many people that are close to me use those tools frequently. Like most of the people that I know either don't use like AI tools or they're like technical and using Factory. Also, like I'm not gonna be f- no one, none of my friends don't use Factory. Like what do we th- come on, we wouldn't be friends.

    21. HSHarry Stebbings1:06:48

        [laughs]

    22. MGMatan Grinberg1:06:49

        Um, uh, so I need to understand a little bit more about that user. My sense is they're probab- and we're still in the early innings, so I'm sure they're quite agile to figure out what is the exact niche that they wanna occupy. Um, but it's not super obvious to me what the kind of focus is. 'Cause my understanding is some of them have been pivoting towards the enterprise a little bit. Um, but I think from the enterprise perspective, the like-

    23. HSHarry Stebbings1:07:10

        I think they've been pivoting towards the enterprise in non-developer centric functions. So like, hey, if I'm, uh, Lovable of the world, I'm gonna sell to sales teams, marketing teams, uh, customer support teams to allow you to create amazing, uh, materials-

    24. MGMatan Grinberg1:07:23

        Yeah

    25. HSHarry Stebbings1:07:23

        ... uh, with no experience developing.

    26. MGMatan Grinberg1:07:25

        Sure. I mean, in that case, I think that, that niche does make sense a little bit more. I think the, if the-- I think it would be ill-advised if they were to try and go to the niche of non-technical people writing code for code sake, 'cause I think that is gonna be run by like... If you're gonna need enterprise controls over who has access to what databases and what code and all that stuff, that's

25. 1:07:45 – 1:08:50

    The coming security danger zone

    1. MGMatan Grinberg1:07:45

       gonna be run by the engineers. That's gonna be where Factory goes. If it's things like, you know, if a salesperson wants to build a customized demo app or customized website for something, I could see in some cases that, that having some value there.

    2. HSHarry Stebbings1:07:56

       Are we entering a danger zone for security? A huge amount of net new code created that may not be as secure as previous, and we're seeing just the worst hacks, security leaks, and this is just the start?

    3. MGMatan Grinberg1:08:10

       Yes. Yeah, it's gonna be crazy. Um-

    4. HSHarry Stebbings1:08:12

       When you say it's gonna be crazy, like what, what does that actually mean?

    5. MGMatan Grinberg1:08:14

       Like the amount of code that's being generated, the, uh... I think code generated is growing exponentially. The security efforts aren't growing in kind, and I, so I think there's kind of a lag there. Um, I think there are probably going to be, in the next couple years, some pretty big incidents that occur because of AI gener- there honestly probably have been. I just-- Whatever incidents that have occurred, no one's gonna admit, or typically they'll be reluctant to admit, um, if it was like AI involved or not. Um, but also, I think we haven't even seen the most adversarial behavior yet.

26. 1:08:50 – 1:11:43

    Should US startups use Chinese models?

    1. MGMatan Grinberg1:08:50

       Like, I think people can use these tools to be quite adversarial, and so I think security, like the higher the stakes, it's gonna grow in importance, and so I think the security part of the market is really important.

    2. HSHarry Stebbings1:08:59

       Do you think US startups should be allowed to operate so extensively on Chinese open-source models?

    3. MGMatan Grinberg1:09:06

       Yes. Um, uh, using an open model is fine. I think the... Like there are kind of two concerns. There's one concern is if you're sending your data externally to like a different nation, which is one concern, and I think that the concern is there about like we don't wanna send our data, um, to China or g- generally, I mean, you should probably wanna keep your data to yourself regardless, or like within country regardless. But I think the, the separate concern is like, oh, the model itself. Like even if we host it in the US, is there a concern with the model itself? And to explain some of the concern there, I think the idea that some people have is like, I don't know if you've seen in like those spy movies where there's like a code word where suddenly someone starts acting like... You say the right word, and then they're like in robot mode where they're gonna go act adversarially. I think the concern is that some of these models might secretly have that ingrained within where you say a trigger word and then suddenly, even if it's hosted in the US, it's gonna like send data somewhere else, or it's gonna start, you know, trying to intentionally kind of break whatever is it is that you're doing. Um, suppose any nation were to try and do that. Suppose they wanted to make a model that had one of these trigger words that's gonna go and act adversarially. Theoretically, you would want to do that as late as possible because if you do that in an early model and someone discovers it, they're literally never gonna use your models ever again. I don't see that as a big concern, and also if you're deploying correctly, like not as a consumer, but in the enterprise, if you're deploying correctly, data exfiltration or like kind of some of this, um, adversarial stuff, generally you can fight against. I do think just from a, you know, I'm, I'm quite patriotic. I think it's pretty embarrassing that we don't have frontier open models in the United States. Um, so I do hope to see us, you know, reclaim superiority there.

    4. HSHarry Stebbings1:10:41

       Europe is significantly behind, especially on the model development side. Do you think Europe is too far behind to catch up?

    5. MGMatan Grinberg1:10:47

       Probably on the like frontier model lab side. On the like-- There's so much to do on the like infra build-out and energy side of things. Um, but again, the thing that's very difficult, uh, in the different parts of the world is you have like democratic countries where things generally are slower. Suppose you say, "We need to do this thing." You need to get a lot of support. You need to convince certain people to do things. You need to pass legislation. It takes a long time. Um, but the benefit though is, you know, theoretically, we get this balancing act where we don't go too crazy in any which direction. Um, you know, other parts of the world where it's more authoritarian is like, "This is the thing we're doing. We are doing it. We're acting now." You get to move quickly. Now, there's less kind of correction 'cause what if you're going on the wrong course? But in cases like AI where it's pretty clear like for build-out, you need to build data centers. You need, uh, energy, um, and energy requires a lot of build-out as well. That has a huge amount

27. 1:11:43 – 1:14:22

    Data centres and the public backlash

    1. MGMatan Grinberg1:11:43

       of lead time. You can act faster. In the West, things are slower. So that's one thing that kind of goes against us. It's a little bit slower to get this stuff done, especially when there's all the politics that you have to deal with it.

    2. HSHarry Stebbings1:11:53

       Do you worry about the public backlash to data center development that we've seen? I think it's like 40 out of 100 data centers post-approval don't actually get built out in the end. Do you think data centers will be seen as a symbol of wealth concentration and technology superiority?

    3. MGMatan Grinberg1:12:08

       Yes, but I think that's, at least in the United States, the beauty of having states, uh, is we get some like selection where we can have different experiments of like what's it like for a state that says no to all data centers? Well, okay, there won't be as many jobs that get created there. Whereas the states that do allow for data centers to be g- to be created, you know, people will prosper. They're gonna have great jobs. They'll, you know, see the, the downstream benefits of it. Um, but it's nice. It's like we have little Petri dishes to test out and see how things work. That is the beauty of the United States. Um, and I think in Europe, I mean, it's a, it's tough. I think m- th- there was some w- there was some good positioning that Europe had You know, a few years ago, a few decades ago with nuclear that I think hasn't been, you know, delivered on as much as of late, but that would've been a world in which Europe would have a way to bounce back a lot in AI on the, on the energy side

    4. HSHarry Stebbings1:12:56

       100%. I blame the Germans [laughs]

    5. MGMatan Grinberg1:13:00

       [laughs]

    6. HSHarry Stebbings1:13:01

       And that's our German audience gone. Um, uh, dude, I wanna do a quick fire round with you.

    7. MGMatan Grinberg1:13:05

       All right.

    8. HSHarry Stebbings1:13:05

       So I say a short statement and you give me your immediate thoughts. Nebius versus Coolweave, who has a larger market cap in five years' time and why?

    9. MGMatan Grinberg1:13:13

       To me, and this is m- this is speaking from strongly biased as an application, uh, person, like I'll take the grab bag. It doesn't matter as... I actually hope for a world in which I don't even k- our users don't even know which one is, is under the hood.

    10. HSHarry Stebbings1:13:27

        For you, I would want Coolweave to be bigger.

    11. MGMatan Grinberg1:13:29

        Why?

    12. HSHarry Stebbings1:13:30

        Because Nebius, I think, have more ambitious plans to be full stack, which will eat into some of your plans in a way that Coolweave don't.

    13. MGMatan Grinberg1:13:38

        Ambitions. What are ambitions? I don't think it makes sense for them to do that.

    14. HSHarry Stebbings1:13:41

        [laughs] Okay. Um-

    15. MGMatan Grinberg1:13:45

        Businesses need to think about their core competencies, and like if people are trying to expand beyond their core competencies, Kirkland & Ellis, great. Look, have fun. It's not your core competency. I don't think it makes sense.

    16. HSHarry Stebbings1:13:56

        Uh, yeah, I get you. I think you could argue that it's a lot more adjacent there, but I get you. Do we have a series of businesses like a Nebius, like a McCool, where customer concentration is like 90% of revenues?

    17. MGMatan Grinberg1:14:12

        Will we see more of that?

    18. HSHarry Stebbings1:14:13

        Yeah.

    19. MGMatan Grinberg1:14:14

        Yeah, probably.

    20. HSHarry Stebbings1:14:15

        Is that a bad thing or a good thing?

    21. MGMatan Grinberg1:14:17

        Um, it's bad if you're an investor in one of those companies 'cause it's a little riskier, but I think

28. 1:14:22 – 1:15:32

    Selling without forward deployed engineers

    1. MGMatan Grinberg1:14:22

       you can find a steady state. It's just scary. You just know there's kind of a sword of Damocles above your head of like if they ever, you know... It, it's just risky. It's risky.

    2. HSHarry Stebbings1:14:31

       A sword of Damocles. It's the first time it's ever been said on the show.

    3. MGMatan Grinberg1:14:34

       [laughs]

    4. HSHarry Stebbings1:14:34

       Um, tell me, can you sell to enterprises today without an FDE model?

    5. MGMatan Grinberg1:14:39

       Yes. Have a good product. The, this, the thing about the FDE thing blows my mind is like y- like for us, when we do FDE, like the way I think about it is their goal should be acceleration. It basically, if there's a customer where if we just give them our product, they'll scale to like, uh, a million in six months, I'll throw in FDEs if they're gonna scale them to a million dollars worth in three months. Great. They accelerated that. If I'm sending in FDEs as services, like I'm not Accenture here. Like I'm not trying to be like, or Infosys or Cognizant or whatever. Like we are not a services company. If we need FDEs to make the product work, we have a shit product. Like the point of FDEs should be accelerate and get them consuming faster. If you're putting in FDEs 'cause that's the only way you'll get a deal done, I'm sorry my friend, you have a shit product.

    6. HSHarry Stebbings1:15:29

       What do you think of the whole grindslop, uh,

29. 1:15:32 – 1:20:32

    Grindslop, sleep and treating teams like athletes

    1. HSHarry Stebbings1:15:32

       element? We, we, we talked a little bit before about the show with Niko at Corgi, which generated-

    2. MGMatan Grinberg1:15:36

       Yeah

    3. HSHarry Stebbings1:15:36

       ... a little bit of discussion online.

    4. MGMatan Grinberg1:15:37

       Yeah, just a little bit. [laughs]

    5. HSHarry Stebbings1:15:38

       Just a little bit of discussion online. Um-

    6. MGMatan Grinberg1:15:40

       Harry, you're ruffling feathers as always. [laughs]

    7. HSHarry Stebbings1:15:43

       Dude, I said nothing. [laughs] This is honestly, it's like, it's like someone comes to your party and does something wild, and it's like, "It wasn't me."

    8. MGMatan Grinberg1:15:50

       Yeah.

    9. HSHarry Stebbings1:15:51

       But, um, what do you think of the grindslop?

    10. MGMatan Grinberg1:15:53

        I feel like a lot of the things we've talked about actually is like something everyone needs to be wary of is intermediate metrics, and grindslop comes from intermediate metrics. Like, oh, you know, generally to do things you need to spend time on it, so let's focus on how much time do we spend instead of like are we doing the thing, right? Like the analogies I use is, uh, imagine trying to measure who won a basketball game by who sweat the most. Like you could sweat a ton, but look at the scoreboard. Like are you doing what actually needs to be done or not? And I think for us, we wanna focus on like getting the best players. I don't care if you sweat a ton or if you sweat very little. If you're scoring a lot, great. We want you on our team. Um, now generally for most people, you have to sweat if you wanna, you know, get things done. Um, but I think you are doing a bad job on hiring if you need to like mandate certain crazy hours or you need a bed in the office. It's like, dude, get a good night's sleep. Like you don't need a bed in the office. Like just go get a, get a apartment nearby that's nice and cozy. Get eight hours of sleep. If you as an important member of your team at your company can get your job done on two hours of sleep, you're not doing very high leverage work.

    11. HSHarry Stebbings1:17:01

        But do you know, I did think it was an amazing opportunity for Eight Sleep to do like an amazing social campaign.

    12. MGMatan Grinberg1:17:07

        [laughs]

    13. HSHarry Stebbings1:17:07

        I would've delivered it. I would've got the founders-

    14. MGMatan Grinberg1:17:09

        Yeah

    15. HSHarry Stebbings1:17:09

        ... outside being like, "We got you covered."

    16. MGMatan Grinberg1:17:11

        Well, it's what, like that's literally, like we-

    17. HSHarry Stebbings1:17:12

        Hey, wouldn't that be funny and viral?

    18. MGMatan Grinberg1:17:13

        When we were, when we were 30, we had like a, uh, 30 people, we had like a, a what we call a surge, like a pretty aggressive like two-week sprint. Um, and as part of it, I got everyone on the team Eight Sleeps, like fully free, whatever, $3,000 per person. Like, you know, the decadence of startups, right?

    19. HSHarry Stebbings1:17:29

        Yeah.

    20. MGMatan Grinberg1:17:29

        But I think the, the idea there is like we are optimizing for output, and the people that we are bringing onto the team, it's like SEAL Team Six, like the NBA All-Stars. Like it is worth every dollar to make them more productive and, you know, to deliver on these ambitious goals that we have. And so if we can do that and, you know, this, at least for me, I think Eight Sleep helps with my sleep. Engineer, like great, let's do it. Let's, they're gonna be better. They're gonna have more of their wits about them. They'll be sharper. And the type of engineering work that we do is not just like grunt work, how can we spend as many hours to do it? We have droids for that. The work that we do is like might require like really deep thought, really kind of like every ounce of brain power that you have, in which case, if you didn't sleep well, like you're not gonna make as good of a decision.

    21. HSHarry Stebbings1:18:15

        If I gave you unlimited money, what would you spend on today that you're not spending on?

    22. MGMatan Grinberg1:18:20

        I think gen- generally, um, we will see the best companies treat teams more and more like, whatever, SEAL Team Six or NBA, like professional athletes. Not in the way that Google did it with like, oh, you get like a bounce castle and all this like weird shit, but like where like athletes, it is kind of, like it seems like they're getting pampered, but it's kind of a burden. Like your diet is monitored. You get like, you have to do your like hour-long massage after a game to make sure your muscles are recovered for the next game. You have to do like an ice bath and all this stuff. Like it seems glamorous, but sometimes it's not. I think spending on that type of stuff, but obviously in the more like intellectual domain, I think that's what more and more companies will do. If I could spend an incremental dollar to, uh, make every person ... sleep that much better, recover that much better, be that much better at making decisions, it's probably worth it

    23. HSHarry Stebbings1:19:10

        You're such an American. Do you know what I like? I like Limoncello. Do you know what I like? I like smoking

    24. MGMatan Grinberg1:19:15

        [laughs]

    25. HSHarry Stebbings1:19:16

        Do you know what I wanna do? I wanna sit in the sun under the intense vitamin D rays, and I wanna take in life with my friends.

    26. MGMatan Grinberg1:19:22

        Yeah. For what it's worth-

    27. HSHarry Stebbings1:19:22

        And you guys are, like, optimized to the extreme. Did you see the Steven Bartlett video the other day?

    28. MGMatan Grinberg1:19:27

        No.

    29. HSHarry Stebbings1:19:27

        Do you know, you might not know this guy, Steven Bartlett. He said, "I drank three gla- two glasses of wine-

    30. MGMatan Grinberg1:19:32

        Yeah
30. 1:20:32 – 1:21:19

    Anthropic vs OpenAI

    1. HSHarry Stebbings1:20:32

       [laughs]

    2. MGMatan Grinberg1:20:32

       If you can recover after, you know, to each their own.

    3. HSHarry Stebbings1:20:35

       Oh, God, that would be the funniest thing ever. Um-

    4. MGMatan Grinberg1:20:37

       Work hard, play hard.

    5. HSHarry Stebbings1:20:38

       Yeah. Uh, okay. You can invest in one company, uh, on IPO day. Sorry, dude. Anthropic or OpenAI?

    6. MGMatan Grinberg1:20:45

       In my mind, the answer here is, I think they're approximately equivalent. Like, to me it doesn't really matter. The biggest reason that affects, like, the EV is, like, volatility of the company. Um, that's the only th- like, 'cause I think from the business perspective they're both very well-suited and they're, like, kind of well-positioned there. Um-

    7. HSHarry Stebbings1:21:05

       So you're saying Anthropic?

    8. MGMatan Grinberg1:21:07

       Probably. Just i- uh, past is an indicator of the future and, like, there's just been more, like, random chaotic turbulent events at OpenAI. Um, but, like, from a business perspective, to me that's like they're

31. 1:21:19 – 1:23:53

    Did Dario do AI a disservice?

    1. MGMatan Grinberg1:21:19

       both great choices.

    2. HSHarry Stebbings1:21:20

       But Anthropic. Okay, we got that. Uh, good. Uh, can I ask, has Dario done a misservice or disservice to the ecosystem by saying, "We're gonna take your jobs. We're gonna take your jobs. We're gonna take your jobs"?

    3. MGMatan Grinberg1:21:30

       Yes. It actually, this, like, really upsets me. So-

    4. HSHarry Stebbings1:21:33

       Mm-hmm

    5. MGMatan Grinberg1:21:33

       ... on one hand I maybe just implied Anthropic there, but on the other hand, I think that has been not only, like, disingenuine and wrong, but it's, like, really hurt the psychology of a lot of people, developers, like, just people in the world. Does AI a, a disservice? Does the world a disservice? 'Cause this is, again, talking about the use cases that are gonna, and the problems that b- will be solved for society. This feeds fuel of, like, we should slow down AI. We should stop doing it. And honestly, it's for selfish reasons that they did that. Because if you're trying to raise unprecedented amounts of money, you know, hundreds of billions of dollars, whatever, the best way to convince people to do that is to say, "All of capitalism is gone. The only company that's left will be me. So you better give us your dollars." And then suddenly when it comes to IPO, when now suddenly all the humans and the people that you might be replacing now have money that you want them to put in your IPO, then suddenly it's, "Whoa, whoa, whoa. Oh, no, humans are pretty important. There are gonna be jobs again. You know, we like you guys." That, that pisses me off.

    6. HSHarry Stebbings1:22:32

       I totally agree, and what's ironic is the ones who've never said it are the ones who've never needed the money. When you look at a Zuck or a Damas, they've always had a very different stance to Sam and Dario when it comes to labor displacement and jobs.

    7. MGMatan Grinberg1:22:46

       Yes.

    8. HSHarry Stebbings1:22:47

       It's really interesting. The ones who need it and the ones who don't.

    9. MGMatan Grinberg1:22:48

       I mean, it's, it's just, it's a shame because, like, for all the philosophizing about AI and intelligence and all this stuff, it's like incentive is driving the outcome and the incentive is, "I wanna raise a lot of money."

    10. HSHarry Stebbings1:22:58

        Which legacy company do you think has most embraced AI well?

    11. MGMatan Grinberg1:23:02

        Honestly, EY, the accounting firms. It's one of our largest customer- I know. That's, you said shocking.

    12. HSHarry Stebbings1:23:09

        Nah, that's fucking shocking. You win an award.

    13. MGMatan Grinberg1:23:11

        They, they are so agent native it's crazy. They're one of our largest customers. They are-

    14. HSHarry Stebbings1:23:16

        How? They just push it down the organization?

    15. MGMatan Grinberg1:23:18

        They're just like-- Basically, they saw what happened with the cloud. They saw scars of being, like, late and kind of not jumping onto it aggressively, and they have some great engineering leaders there who are like, "Look, this is gonna be scary. Some people are gonna get upset. You know, it's not gonna be the easiest thing, but we are going to make our org agent native it- if it's the last thing we do." And they were honestly pretty early to it as well. To me, I think that's one of the most interesting things seeing is, like, they're more agent native than some, like, startups, which is wild. Brave New World.

    16. HSHarry Stebbings1:23:52

        Brave New World.

32. 1:23:53 – 1:24:50

    What he's changed his mind on

    1. HSHarry Stebbings1:23:53

       Uh, final one. What have you changed your mind on most in the last 12 months?

    2. MGMatan Grinberg1:23:57

       What I've changed my mind on the most in the last 12 months is there was a brief period of time where I thought it might be just one or two companies that run away with being kind of the frontier and the best. What seems pretty clear to me is it's probably gonna be at least four that are gonna probably be approximately as good, and that is a win. Like, that is the win for humanity. Like, the bad case for humanity is when there's one that's really, really good. Like, I think there's probably gonna be at least four, if not many others. Um, and that's something that it seems there's, like, kind of growing evidence of, which kind of, uh, my sense is it's a hot take 'cause I think right now people are a little bit enamored with maybe one or two, but...

    3. HSHarry Stebbings1:24:35

       Listen, Matt Damon, it's been so wonderful to have you on the show. Uh, I'm gonna let you go back to Robin Williams. Uh [laughs]

    4. MGMatan Grinberg1:24:39

       [laughs]

    5. HSHarry Stebbings1:24:42

       Um-

    6. MGMatan Grinberg1:24:42

       Oh

    7. HSHarry Stebbings1:24:42

       ... and this show was brought to you by Eight Sleep. Uh [laughs]

    8. MGMatan Grinberg1:24:45

       [laughs]

    9. HSHarry Stebbings1:24:45

       I'm kidding. Uh, dude, it's been so much fun. Thank you so much.

    10. MGMatan Grinberg1:24:49

        Thank you for having me

Episode duration: 1:25:00