Deedy Das: Menlo Ventures' Strategic Bets on Anthropic and AI's Future Stack

Basically very iconic in the AI world. Okay. Hi td welcome to Venture with Grace. We're live.

Hi, nice to be here. Thank you for having me, Grace.

Okay, to start the show I would love for the audience to learn more about you. So actually you grew up in the US for like nine years or like a few years. And then before you go back to India I guess like what are some like, I guess, I mean I almost asked what your parents do, but like I was so curious. But like I guess like what are some core lessons that you've learned early on in your career that you kind of took into to shape you into who we are today?

That's a great question. So I was born in India, but then I spent age 3 to 10 in America. Both my parents were working, they were in tech. My dad was at IBM, my mom was at Oracle and I actually when I reflect on my childhood it over time I've come to realize that my time in America was actually pretty critical for the person I am. Mostly because in those seven years I was going to this, the school in the Bay Area of called Challenger School. It was this private elementary school where I would go be dropped off at like 7:30 in the morning and be picked up after my parents work days were over at 6:30. So effectively had very little time with my parents. I mean most of my time was with school and my friends. And one of the things I learned very early on in life was think for yourself. The book is not always right. Sometimes you just have to reason about what the right way to proceed is based on your principles. And it turns out that I didn't even realize, I mean when you're, when you're in the journey you don't realize how important of a lesson that is. But over time I've begun to realize so many people were, did not have the fortune of being taught like that. So I think I organically became a bit of a rebel growing up because I was like why would I listen to somebody else's opinion on something? It's your opinion, but I have my own. And, and so even with my parents, with school, with authority, I always felt like I had my own reason or opinion or a belief based on the things I knew. And I think that was really solidified in the seven years of school that I was here because my teachers early on were very good friends of mine and you know, I'd invite them to my birthdays as a 9 year old, 8 year old and they would always encourage me to like think for yourself and come to your own conclusions. Don't just let anybody dictate how you should think about something. So I think that's the core lesson that stuck. Stuck with me for a very long time.

Wow, that's like actually really interesting. I actually feel like it's very hard to think for yourself because sometimes like since like your parents sounds like they're very successful too. So like basically like how do you kind of like have your own opinion or like how do you arrive at a conclusion? Like, do you, I guess like how much research is like well researched about like a subject? I know I sound like. I mean I have like a one year old. So like now I have to send them to Challenger school. So but anyway, so I don't know how you actually arrive at a decision. Even like I said, doubt. Like I constantly think about how we make decisions. But like, I feel like a lot of the authority like are because of their more successful or like, let's say like, oh, by the way, hi. Hi, Kumar. I wanted to shout out any audience who comments. So I wonder, so how do you think about basically making career or like life choices by yourself? Like, what is the thought process goes behind it?

I think with. Well, there's a couple of things to say. The first thing I'll say is I, over the course of the, you know, 30 plus years that I've been alive, have noticed that people pay a lot of time and attention thinking about things that doesn't matter, that don't matter at all. They spend a lot of time like debating these like quirky, like useless nuances on things without like thinking about big picture stuff. They often take the big picture stuff for granted. And then they debate small stuff that doesn't matter. So people do a lot of like, hey, you know, what should I major in? Which is still a big decision. But they'll like spend a lot of time thinking about that or they'll spend a lot of time thinking about should I take this class or should I take that class in college? You know, should I be doing this extracurricular or that extracurricular. But most people don't spend time thinking about what is the arc of my life and why. And I think at a very early age I was like always curious about that stuff. And I, and I think it comes down to this, and this is my second point is most of it comes down to first asking why, no matter what you're told, and then figuring out, okay, you want me to go to college. You know, when I was a kid, I would tell my parents this all the time. Like, you want me to go to college and, like, work some job? Why? Like, okay, I need money. Why? Like, why? Why is money useful? Why? Where does money come from? Where do I want to spend it on? Is there something I need from it? And I think very early on, I was, like, really concerned with those questions. I'm like, why do anything at all? Why do I go to school? Why does you know both things outside in, like, nature and in the realm of science and math, but also, like, just fundamentally, why does life work the way it does? And so the first part, I think, is just asking the questions. And then when it comes to finding answers, I mean, now more than ever, I think you should just do research. I mean, there's so many opinions of so many smart people that are on podcasts, on blog posts, online. You can find so many people who have a different answer to this. And then you start reasoning about, okay, maybe for some people, college doesn't matter. Then you asked another why? Like, why are some people more very successful without going to college? But also, some people are pretty successful when going to college. What is the difference between those things? And so I think when you let that curiosity drive you, you get to a pretty interesting worldview where then over a long period of time, you're like, I think I understand a little bit about how this thing works. And I think that's what I would encourage even younger kids or people to do is allow people the liberty for your. For your son, for example. Allow people the liberty to ask the question why? And then encourage them to figure out the answer themselves and don't tell them. So, like, this is what I think, but you figure out what you think. And I think that's very empowering because people don't realize this. Kids look up to parents. So if you have an opinion, whether or not you force it down their throats, eventually they will listen. But they also need to come to that conclusion themselves. They don't like, long term, it's not sustainable for you to just tell them and them to internalize it.

I think I actually have another question. Well, okay, maybe we could go into, like, your journey a little bit more on, like, you know, how do you pick Colleen? So, like, it seems like, was it from the KPCV engineering fellow? Or like, how do you come across glean? And then why do you join? Obviously it's a very successful company. Arvin was on the podcast. A bunch of, like, lean investor were on the pot as well. But curious, like, how do you find the company and how do you decide like which company to join? Because there's like a million startups.

So I was at Google at the time. I'd spent about four plus years there. And I had come to the realization that, hey, this is probably not the best place for me to be at this part of my career. I can go into wise. That was the first part. Then I thought, okay, I know I think I want to, ideally I want to start a company. And I had a couple of things that I wanted to figure out about starting companies that I just didn't know. So my first step was really to talk to a bunch of people in venture. I thought that was a pretty good step. I'm like, who has the most knowledge at a high level about a wide variety of startups? Probably the people who spend all day investing in them. So I talk to them and then you quickly realize, okay, some people seem to know what they're talking about, some people don't know what they're talking about. Some people are using proxy signals to judge. So you get some lay of the land of what startup good, a good startup looks like. Then I started actually having meetings. This was a long process. So I tell I had a meeting with a bunch of different companies and I would approach those meetings very much like I was interviewing them. I'm like, hey, why are you doing this? Can you tell me about it? What's your vision here? Why do you think it'll work? And I think I had more than maybe, I want to say 50 different startups that I met with. And you know, I started at that point to think maybe I'm. Something's off with me because I'm also calibrating to what a normal seed or different startups that were of different stages. But I was trying to calibrate myself to what good even looks like for a startup. And I didn't know. And so you know when, when someone comes and tells you, I know you're very smart, but you haven't figured it out. I didn't know at a seed stage you should you have figured it out. Like, is this normal? Is this not? After a lot of conversations, I was introduced through one of my bosses to Arvind. Then there was like, hey, here's a person I worked with at Google. I think you should talk to him. He is about to start a company. And then that conversation changed everything for me where I was. It was very clear from talking to Arvin that he did not. He set the bar at a very different benchmark from everybody else.

Spoken to.

He was very clear about what he wanted to build. There was no question in his mind that this was a big market. It was. And it was a market that was actually a useful problem to solve for people. Number two, he had a very good story about why he thought we could win that market. It wasn't obvious. Enterprise search at the time was something that had been tried maybe hundreds of times before. It was the most common idea in the world. People have been trying it since the 90s. It's not new. So why would this one work and why haven't other ones worked before? So we did a little brainstorming but even he had a good thesis of why then. And then the third thing I think and by far the most important thing is I had come to understand at that point, I still believe this is that in the early days of a company the two biggest risk factors are ability to fundraise and ability to hire. Because those really change the dynamics of the. The set of opportunities you can do. If I have 15 million of Runway at a seed, I can hire a team of 12 and build product for about two years, maybe more and get to PMF. I give myself a lot of headroom. On the other hand, if I'm a company that can only raise 500k I don't have that headroom. I just can't cover that surface area of product. So he had that. He had both of those two things which I thought were completely de risked at least us getting to a series B. So that's how I was looking at it. I'm like look, you have a great idea that clearly you have a lot of good reason to, to, to believe it works. You have a reason why this one works. And three, you can hire and you can fundraise better than anybody else. Most other people and that is the biggest risk factor. And if there was a four, you have a distinct right to win in enterprise sales. You've literally done this before. You've built a multi billion dollar business with rubric so that those four to me at that point were reason enough to not go and start my own thing and to not consider anybody else. I'm like this makes so much sense. And it was, you know, several standard deviations above what normally I would see from other startups. And that's why I chose eventually to join glean in 2019.

It was called something else then I wonder. So like, I guess like another question is like I think it's a very separate topic. So like I guess not separate topic but let's I want to piggyback on some like operational question on like how do you select the 50 startup that you chat with? Is that just like anyone you stumble upon like or you see a job offer or you went through like the YC list like how do you pick the 50 companies? And you mentioned that like one of your mentor like someone that you know also knows Arwen and they made like a warm introduction but I feel like the thing is like okay, so how do you walk through that conversation to like be like hey, I'm actually looking for my next role. Here's my criteria which you like back in time you probably didn't really have one because since you travel with 50 coming their own not it. So I guess like how do you have the conversation and then the person, how do you like approach people who can introduce you? Because I think there since you're a VC today so like everyone will come to you for advice. Like your mentees that are like you know, I guess the next gen DD's are like coming over and they'll be like hey can you introduce me to blah blah. How do you think about the entire process of like finding a job in a very high potential startup and also join them at the very beginning stage Because I think that's a really big key as well. You are the founding engineer and like you know right now if like anyone trying to join cursor, it's like freaking impossible. So how do you think about that approach?

Okay, so two part answer number one is I would call it sourcing. How do you figure out who to chat to?

Right.

I got, I made a lot of these mistakes. So this is the polished version of the answer. After I know all the things I did wrong. I think perhaps the best way is to go and talk to good VCs. If you go too high in VC you might not get a conversation. If you go too low, you might not get anything of value. Somewhere in the middle of somebody who understands things in venture but still will have free time for you. They also have incentive to help you because they are trying to make their own company successful. But you do have to discount a couple of things. They say if they're going to try to feed you to their own portfolio companies, by and large you, you know, you'll average it out by talking to enough people. So that's where I would start. I think that you can do the angel list, the hacker news, the YC companies, all that is great too. I just, I find it very hard to distinguish what's good and bad just from looking at those lists. And it can get really hard because you just, you know, you can click into People's LinkedIn, you can see what they do, but you never really know if it's good or bad. So I think, and I think all of the early, early stage opportunities and that's where I think a lot, that's the only place you can actually create meaningful financial value for yourself. I think it's the best risk reward ratio is to go early. And if you want to join a less than 10 person company before it's made a name for itself, most of these guys don't have public job postings at all. You don't even know what the company is or whether it's good or not. And you assume because you haven't heard of it, it's not good. So going through VCs is probably the easiest way to get access to that essentially proprietary job opportunity. So that's how I would do it in terms of pure sourcing and then other friends and stuff of course. But usually warm referrals help the best people underestimate how valuable the warm referrals are in the sense people who, you know, usually have pretty good judgment so they, they will, you know, be able to show you good stuff. The second part is how do you then approach actually doing that conversation? Well, I think the biggest mistake that people make one is the inevitable look, if you are not technical, it is so, so hard. Like that is just facts. There is, I mean for most companies that are under 10 people in tech, they don't have a need to hire somebody non technical. They might have one non technical person. And if you are, if you want to be the first non technical person at a, at a startup, you know, be prepared to grind a lot. It not easy. You are essentially picking up every task of the company whether or not it's related to your job. There's no job description for that role. It is, you are the non technical person. We will see if you become a product person or a recruiting person or some marketing person or a salesperson. But right now you are chief non technical person. And so your job is to do everything that we cannot do as founders. Some call it chief of staff, whatever. That's a really hard job to gun for. So let me leave that aside for a bit. It's way harder if you are technical. It's, it's actually a lot simpler. Most people are just trying to find good talent that's interested in them. And most candidates I think do the wrong Strategy of spraying too broadly. They don't know anything about the company they're working for, they're talking to. And they, they're not doing basic diligence. I mean, 90 of candidates just don't do basic diligence on the company. Who, who are the people? What does the company do? What are the competitors? Like, just prepare. It takes half an hour. Now you have all these AI models. Just go ask one to do the research for you. If nothing else, come to the conversation somewhat prepared, with just a somewhat smart set of questions. And I think that already puts you in the top 10 percentile of candidates. There's too many people who especially come from these FAANG backgrounds, you know, who come in thinking that they are the gift to the company. And I'm like, that attitude is the problem. You might be great, but you can't come into a conversation being like, so I'm here to talk to you, when can I start? Or are you like, are you good enough for me? Is the kind of tone that I sense a lot of people have without doing any research into the company. Like, learn about the company, learn about a little bit about what they're trying to do, and try to add value. Those are the two things that make people stand out. Being thoughtful about the company and figuring out how you can add value to the company even before you start. So some things that have worked for me, even when I was recruiting at Glean in the past is I love when people build side projects and I love when people build something especially to prepare for a conversation. And I always tell people like, this is the most open book secret in the world. If I want to crush an interview with somebody that's not even a real interview, I will literally build something that that person cares about. And it takes zero time in today's day and age. And I will show them what I built in the beginning. I'll be like, hey man, could I show you something? This is something I've been working on. And even if it's an hour of work, you show them the thing you've built. You've won the interview in the first 10 minutes. The person will be like, nobody ever does this. You are crazy. I like you. You went, you researched stuff about me, you found out something I was interested in, you built something pretty cool. Just do that. But everybody else just sends a resume and writes this chatgpt3 paragraph thing. I'm not reading that, I'm not reading your resume. So it just doesn't work. So those are, that's how I would.

Approach the conversation for the 50 conversation you had were it with the founders or was it just like a hiring manager?

It depends on the stage. So I think for earlier stage it was usually with the founder. As you go through series B and C, usually with somebody else, sometimes you do a quick call with the hiring manager before you talk to somebody else. Many ways to navigate that. I would usually say, hey, I would love to speak to the founder just because I want to understand their, their vision but I understand if they, they're not available. So. And I was always open with, you know, timing and stuff. Like whenever they want to call, I'm happy to have a conversation with them. So yeah, that's what I would say.

I'm also curious about the projects that you're building. So like you mentioned, like it doesn't take much time. First of all, I feel like as a non technical person I'm already dead in the first first round. But like right now there's a lot of tools that you technically can build. But I feel like the amount of research that you have to do also build something and then do that 50 times. It's kind of not that easy. It takes some times. How do you prioritize? Like do you talk to the most likely good people on paper first? Or like do you do, I don't know, like how do you kind of like actually go through the process though? I'm not like yeah, I feel like this is like a question that not applying to me but like I just felt like I'm fascinated by your answer.

I, I think I have an answer for non technical people too. I'm not saying it's always about go build something and you have to show code. Like that's not the point of it. I think building something is an easy way to show that you're clearly interested and you're not. Like you're not screwing around here. When I approach it the way the things that I would build are, you know, probably vibe code, something interesting right now, but probably build some small quick app. To your point. Yeah, I'm not doing that for 50 companies. So usually it's when I'm in like someone interested in the company and I want to make an impression, then I would go and gravitate to doing that. But there's always pared down versions that you can do from a non technical and technical side. From a technical side I would come in and say look Arvind, like I actually studied the five points that you said before about why you're differentiated and I actually think like, have you considered these things? Like there's other, other points that you. I don't think you've really considered. What do you think about these? Like, I just wanted to get your opinion on them and it serves two purposes. It's a genuine question for me. And two, he's like, wow, you've actually gone in and done your research. Like, this is crazy. You just can't treat startups like they're big companies companies. There's no hiring process, there's no bar, there's no like three, four rounds and then if you pass the lead code, then you're in. It's really a question of can I work with you and are you going to be all in on this thing that like all of us are and that requires you to show interest. And for non technical people, I say you, there's other ways you can show interest. Like there's so many other non technical parts of the business that you can think about. Do a competitive deep dive on. Like here's the four competitors. I went and tried all their products and I have some stuff to show you. You know, how useful that is. Like if you go to a seed company and just say, I know you all don't have time to do this. So I went and did a deep, deep dive on your three competitors and I think here's where the features are lacking. Here's what you have good, here's what you have bad. That might take a couple hours, but I guarantee you will nail the interview without them even asking a question, right? Like you've done the work, you clearly care. You show insight, you show being able to be proactive before being asked to do things. You've implicitly shown that I can succeed in this role and that's how I would.

One of the things that you could do, I wonder. So, separate question. So we talked about the Glean era and then maybe we could later on talk about how you join Melo. But I think like one of the things I find really interesting is you guys started like an anthropic relative fund, so maybe you could talk about that. And like, because I feel like that's such a. I think that's exactly if I'm like, I mean, obviously I'm not at your firm, but I feel like that's such a smart move for you guys. The anthology. Anthology fund. I cannot pronounce it, but anyway, so like how do you think about that approach and how do you. I guess, like, I guess that's just like creating an ecosystem fund. For Anthropic, I would assume that's like too like 25k per company or like maybe you guys share a little bit more about on that. And then I feel like that's such a great sourcing funnel for Melo. But like, I mean, I don't know what you want to talk about, but like I, I feel like that's such an interesting initiative. So I wanted to call right here.

Yeah, absolutely. Let me, let me talk about that for a bit. Anthology. The Anthology fund was one of the reasons that I wanted to join menlo. It was, it was a really fun idea. It was really interesting to me. It's a hundred million dollar vehicle that we run with Anthropic in partnership with Anthropic. MENLO is one of the biggest equity holders in Anthropic. And we always had this idea that this is clearly a new era of AI models that will have an ecosystem around it. And ecosystems don't come out of thin air. There's a lot of work that goes into building an ecosystem, both from a company side and from Anthropic, but also from a financial side and funding that you want to build on your ecosystem and that you can work more closely with. The idea was we have $100 million fund, but we didn't want to run it quite as a corporate venture fund. Corporate venture funds have a lot of subtle dynamics which we think are probably not the best way to run a fund. They are run almost explicitly with the intent of bolstering the ecosystem. So they're not really choosing, in many cases they don't choose the best companies. They choose companies only because they can help the ecosystem. And I don't think that's a healthy way to grow an ecosystem. So I'm somewhat self serving. So we spoke to Anthropic and agreed on this idea of we will help run the fund, we will run it with you and have you guys on board. We will still do a lot of the ecosystem work, but we also want to go and help a bunch of these companies that might be slightly even off thesis for you and generally help the broader AI ecosystem. So what we do is we offer companies a lot of benefits, right? So we have the, obviously a bunch of credits. We have 30k anthropic credits. It goes up if you, if we do more money into you. We have 100k and AWS credits, office space, all this other stuff. But that's probably one of the draws. Our anthology founders tell us the bigger, bigger draw is you have this pretty selective community of founders who are, you know, 25 companies that can learn from each other's experiences and how they are building with AI and also be very close to the Anthropic team, which is one of the best teams when it comes to figuring out how to do modern engineering on an AI stack. So they find those sessions super valuable. How do you think about product, how do you think about retention, how do you think about architecture? And then for us the benefit is we get to be into a variety of companies at various stages that may not be a part of the company. Core Menlo thesis and the core Menlo thesis would be hey, we want you know, 15 to 20% ownership in a company. We only going to do few deals a year and then we're going to help those companies through to, to an exit. Now we can do a lot more, we have a lot more flexibility in check size and we can say hey, we have a vehicle to work with all of you together. And then from those that pool, if we see people doing well, we can go and find a way to work with you and maybe even preempt your next round and bring you in as a official like proper Menlo bet. So that's kind of how we see the entire program. And then for Anthropic of course is the other side of the puzzle. Anthropic loves the fact that they can work directly with these companies that are on the cutting edge of how you to build on these tools and they can figure out hey, what does our ecosystem want? What should our product roadmap look like? And that's the three pronged approach here.

I have a question on this. So let's say you are I guess like a GPU provider company or. Well, I'm not talking about Nvidia, I'm talking about like you know, like a mini together AI or something like, or how do you think about they could build some sort of like joint partnership with the fund and what kind of ecosystem people should be involved. Because like right now you're talking about like you know, Anthropic, there's aws like they're offering credits and then Menlo come in and then be the selector and then bringing the deal flow because I'm sure you guys have like a ton of earlier stage deal flows and like maybe or like even just on anthropic ecosystem people are building on top of it and like you know, just having this ecosystem found makes sense. But like I guess like if you're running like a cloud GPU company, how would you kind of like go about this approach.

I mean we have very close. Well first let me say we have very close partnerships with Nvidia and AWS for all of their, the computing that they offer. And so our. For anthology companies specifically. Yeah, they go in the cases that they need to. They get a lot of their GPU needs met by those partners. If your question is how would I think about that if you were trying to run a similar fund with a GPU partner? Is that the question?

Yeah, kind of.

I would select for companies. I think the focus of the fund would be a little bit different because you know, people building on, I mean the, the layers are a bit different. So people building on Anthropic in my opinion is a little bit of a broad set. People building on top of GPUs directly is a slightly smaller set of companies, but they're also very, very interesting companies. And so I would go and pick the companies who have these GPU needs and there's so many of them and, and then work with those. And so for example, you know, there's a lot of people doing drug development and building a model to do that. That would be a great candidate for a company in say if you were running this fictional Nvidia fund, then it would be a great candidate for that. Obviously Nvidia has adventures. They have a fund. There's companies that are doing cutting edge material science research with foundation models that's pretty interesting for, for a fund like that. So really deep foundational model work that that requires GPUs I think are the best fit for funds like that. And I would just focus on that set.

How do you think about like, do you think like Airbnb for GPU makes sense or not? Like serving the non core. Like you know, I feel like a regular. If you're like a venture backed startup in a later stage, you already have like the core partners like Nvidia or something else. But like for, for the people that are like trying to build something like you mentioned about like you know, material science model or like just like people who are like working a lab like on the side they're building a model that's the need actually exists or like do you feel like it's kind of like a fake assumption of like people will actually use Airbnb for GPUs?

I think that there is a need. I don't know how big the need, how distributed the need is. I know there's a big need for GPU. GPUs. The real question when it comes to A business like that, There are businesses like that. I mean a lot of cloud businesses are essentially that the hyperscalers offer GPUs that you can rent the. On the top end of the market, there are people who are going to require, I mean, I call it on the order of 10,000 plus GPUs. When it comes to an order of that scale, Airbnb for GPUs. If you want to make any margin, it's not going to make sense for them. They're probably going to go direct and figure out a way to build their own stuff. If they need GPUs beyond a certain capacity, you know, like Xai has got it done and a bunch of other companies are considering to do so. There's that and then there's the long tail of the market where you're like, okay, you might need a GPU, but you're like using it once a month. So I'm, even if I run this business, it's just not a good business. I'm like, I'm getting $4 off of you. And there's not many of these customers and they're very low volume. So the way I would think about that business is, okay, how big is the middle? Like you really need a very chunky middle for this to work out. So a bunch of people spending, I don't know, on the order of, call it 10 to 50k a month on GPU spend. So they're at a weird middle ground where they need GPUs. They're not really, really big where they're going to get their own and then you serve that need. So that's kind of like from a more first person principles level how I would think about that market. But at the same time, that market clearly exists. There's, there are hyperscalers doing that. There are other companies doing that on a business level. I would say the hard part about doing that is you are, it's, it's essentially going to be if you own the GPUs yourself, it's going to be a low margin business and you're going to have to underwrite demand predictability. So if you essentially buy a bunch of GPUs and no one's using it, then you're screwed. And if you don't buy enough, then you're still screwed. So it's really a good question of can you get the supply demand on that marketplace to work correctly. Even though it's a low margin business, you can still get insane volume just by making that happen. There's A bunch of companies, I don't know, I think Lambda Labs, Crusoe, right. Everyone, etc. Who have a bunch of GPUs that they're renting out like this. Totally.

I wonder. Okay, so another question is on like brand. So like the two days ago like on our podcast we had the Cluli founders. Yeah, they're like very interesting founders. And I wonder like, so they mentioned about like, you know, OpenAI is like a lot higher valuation because of they are doing many. Just like they're doing marketing. Right. And compared to. Okay, another model. I wouldn't name names but I wonder how do you think about like is marketing everything in the AI era or like how do you think about like prioritizing product versus like you know, hiring 50 interns to marketing your company? Do you think this have longevity or you feel like it's a, it's just like a backtrack on the Internet?

I, I first want to say I. Roy is one of the boldest people I've ever met. I know it's not a very popular opinion. He's a controversial character. But I, I just think like he has, he has a lot of that core rebel DNA and I just like seeing it. Like he's, you know, doesn't do everything right in my opinion. But he's curious and he's, he's bold. So I, I genuinely think he's a great, he has the, the ingredients to be a very special founder at some point in his life, if not now. So that aside, talking about marketing and distribution, essentially. Yeah. So to start it off, it is absolutely a huge problem that so many companies that are even very, very big Fortune 500 scaled companies are beginning to realize distribution does not come for free anymore. It doesn't even come if you run ads anymore. You have to be creative if you want to get distribution. The audience has shifted. People don't fall for traditional kinds of marketing anymore. Back in the day you could run some Google Ads, some Facebook ads, put a billboard up and you'd be fine because it was. That space wasn't crowded. Now that is insanely crowded. And users are like, I'm not falling for this. It's a really bad way to get users because those users also don't care. They're not very organic users that are very invested in your product. True word of mouth style growth comes from certain places where there you get people on the beginning of the S curve who are extremely willing to try things and then they espouse those products because they love them and that's how great organic growth happens, which is kind of the dream for every company in various shapes or forms. Now, coming to his question about models, OpenAI very quickly realized that they had a business in consumer and they, they could make so much money with these subscriptions. And I think it became a natural focus for them to optimize for that. And I think, you know, Sam kind of said it himself in this interview. I don't think he meant to say this, but he was like, well, do I want the. Would I want the smartest model in the world or a billion users? And he's like, I'd want a billion users. And that kind of goes to show what the focus of that company is. And he's right. I mean, if you have a billion users on a product, it is an insanely good business. You will find ways to monetize that. And if you are going for a consumer business where you want a million, a billion users, we all know that growth is everything for that kind of business. On the other hand, for enterprise, growth doesn't always look like that. You can be a very fantastic enterprise business and have very few people even know that you exist. So the first thing I'd say is it depends on what you want to do. Generally it matters, but it really depends on where the business focus is. The second thing I'll say is, yeah, it's absolutely important. It's extremely important to crack marketing. I tell a lot of the founders I work with this is think what you will of Roy. He is a marketing genius. He just knows what people resonate with. And I used to think that was easy. But most people have no clue what people resonate with. They're too stuck in their own head or their own set of friends or their own bubble of people. They don't know how to translate what's important to them to what matters to people. And I can tell you this firsthand is I told Roy this also at Glean, I interviewed like four or five CMOs of fairly reputable organizations and I was extremely, extremely disappointed with the how unaware they were about their users. They don't know anything. I mean, you know how to run an ad campaign, great. Anyone can learn that. But do you know what the user wants to hear? Do you know what you're how to sell to the user, how to market to the user? And they don't. And my question to them is, I'm like, look, if you can't get 100 people to read your blog post, how are you a cmo? You can't market yourself. So what have you done? So I, I'm, I, I feel like there's a moment like the traditional era of marketers aren't always very good at this modern era of what you need to succeed and it's extremely important for companies. Attention is kind of saturated. You need to get people's attention and to get people's attention, you need to understand what resonates with them. So yeah, that's a long winded answer to say. I agree, marketing is extremely important. Different companies have different focuses on the kinds of marketing they do.

I wonder as an investor, what's your thought on like 150k on AD?

What do you mean?

Yeah, so like the clue people I asked them like how much does the ad cost? They mentioned is 150k. Unless I'm remember wrong. I'm sorry if that's the case, but I think that's around the ballpark. So yeah, as an investor, if you are an investor of their company, like how do you think about that kind of approach?

Approach, there's no one answer to this. I think it's, it's one of those things where if the 150k ad does 25 million impressions like Roy's did, then it's a good idea. But if it didn't, then you can always go back and say, yeah, maybe that wasn't a good idea. Generally the way investors or the way I would think about it, most investors would think about it is, look, when I have made the decision to fund a company, I have given trust to the founder in how to spend this money. It is not my job to call these shots. I can have an opinion, but this is not my job. If you decide it's a good idea, then you do it. This is, it's your company. You have all the right to go and do that. And I have like basically entrusted you with the judgment to make these calls and my, my personal opinion would really, really vary based on who's doing it for Roy. Yeah, I mean the dude is a genius at marketing. Like for him spending 150k, I wouldn't be, I would be surprised if it didn't work. But then if, if you had a very traditional founder who's basically never done anything organically in marketing, does never posted once on like a platform, has never written a blog post, is very out of touch with what people feel and they come and tell me, hey man, I want to do a 500k ad campaign. I'm like, I don't know man, I feel like you should probably try other things first and work your way up to a point where you can do that instead of throwing money at a problem that's not going to solve that problem. So that's kind of the rationale I would have when. When I. When I think about things like this.

How do you think about, like, audience build on your own? Because you're like very popular on x, on like LinkedIn. And like, when did you start it, like, intentionally doing so? And how do you just. In general, how do you think about building an audience?

I. People hate my answer to this because I think they think it's not true to some extent. But this is genuinely how it happened. I joined Glean in 2019 at Glean was the first time I felt like my employer wasn't breathing down my neck, trying to make me say or not say certain things online. Google and Facebook are very hard. Those are the two other companies I was at, are very hard places to actually be able to write online because there's a hundred different legal policies on what you can and can't do and your job is at risk. So that first thing step was, okay, now I can actually write. And I always enjoyed, enjoyed writing in general. I enjoy writing, I enjoy reading, I enjoy things like that. It is important to me to generally express things I find cool more than anything else. So I started like, kind of posting things that I thought were cool for no real reason. I think most of those have zero likes. Like, it's not for any purpose whatsoever. It's just like, hey, I found this thing cool. Then occasionally some things would get 10 likes. And I was like, oh, people, people think this is cool. This is awesome. Around early 2020, I told. I was. I had this conversation with my friend. He told me, hey, a lot of the ideas you have are, are really bad if you thought they were really good. In fact, you would write, write about them online. And I'm like, that's an interesting point. You've played this on me. And it got me thinking. I'm like, he's right. A lot of the things people say out loud are actually really bad ideas. And then we get into this fight because we're in the moment. Why not just go and post online? Because then if it. Then if it meets my own bar for what's publishable under my name, that's. That's a way to vet the idea. And I found that as I started doing a random thing, a random rabbit hole of interest that I would go down when I would post them, it would force me to do the extra 20% work at the end to refine the idea into something good. And a lot of times, you know, you have a, you think you're what you, what you, you have a thought, you're like, oh, this is a great thought. And then you're about to press send and you're like, it's not a great thought. I should do some more work on this. And, and, and, and refining that process happened around early 2020. And then I just told myself, well, look, it's the pandemic. I might as well do one of these a day. That's my goal. One of my, my personal goals was I'm gonna do a thought a day. And that's all I did for, I mean, it's not all I did, but that I, I kept those principles. And then a couple of others, like, no, no politics usually, no jokes, no talking about sports, only like, tech and a couple of areas that I really cared about. And over the last four years, it just grew because I was consistently doing things that interested me and that, that's kind of the story there. It was never really to build a following or anything like that. I always enjoyed sharing stuff. Like, my first viral thing was actually. Well, one of my first viral things was a resume template that I wrote for in sophomore year that's now been used over a million times. And I thought that was cool. And then before that, I wrote a blog about hacking the Indian education system and finding anomalies, statistical anomalies, and those results. And then that became very viral and had millions and millions of views and it got all this media press in India. So, you know, I was always used to writing and this was like a forcing function for me to do it.

More often is I have like three question on this segment. So, like, I think like, one part is about posting ideas. Like, like your friend mentioned, like, I mean, honestly, I think I have some alpha in the most basic ideas. Like, I couldn't really polish them into like a viral tweet because some of the ideas are like, so dumb. Like, most people will think I'm a freaking idiot. But I think in the long run, in 10 years, it's like a really good idea that I actually think it's an alpha that I would not actually want to explain to the public. And so I actually don't want people to steal my idea if that's the one idea that I have. The stuff that I posted are like, that's fine. It's like detail our ideas, right? But like, I wonder how do you think about if you actually have an alpha that you don't want to share. And another problem is like, not problem, but I guess like questions. Like, I think of stuff that go viral. So for example, I feel like Roy is like an organic viral character on the Internet, right? So like it's like, it's thought provocative. It's like things that like people have very like different opinion. It's a kind of black and white vibe, things. So like that will actually help you go viral. But that's not actually a lot of people's general idea. Like, you know, I mean, I have ideas, but like they're mostly like in the gray area all the time. Like, it's not gray area, but it's like more of like it's not black and white. It's kind of like complicated. So like it's not like, you know, you double think things, right? But like it's like for his idea, like when you articulate out, it's like a very viral idea, but like that doesn't reflect your actual opinion about things. So that will help you attract a lot of audience. But on the other hand, are they quality audience or are they just like there for, for the, for the sake of being there? But after coming out of like the crypto era, I feel like marketing is literally everything. I don't know if you have any cadence. So you mentioned that's like, I'm sure like the algo change since like 2020. I feel like the LinkedIn algo also change. Like how do you navigate the algo changes that like. And then back in time, you, you're posting once a day. So like recently I saw your. You were tweeting literally four times a day. So like, I guess how. Not today. But like, I mean recently, every day you have multiple posts. That's what I'm trying to say. So how do you think about. Are the algorithm gonna only promote one piece of content a day or like, do you feel like. I mean, I heard about Twitter, people say you have to post eight of them and then like one of them will hit or something. So like is quantity better or quality quality better for you?

Okay, a lot of questions. Let me get into all of them. So on alpha, this is very clear for me. I, I usually think that nothing really has alpha. So if you have uncovered an interesting insight and you share it, odds are even if you shared it, no one else will have the same passion for the insight that you do. So I would not be concerned about it. I. I generally think that is over. Overrated And I think sharing things that you find interesting and that you might have alpha over is good. And, and it has this effect of people just thinking that you are the expert in that. So. And you know, it's a good feedback loop. So if you share something with alpha and you get people to engage, it kind of forces you to dig deeper into the idea and then it kind of is a feedback loop where you actually kind of become pretty much of an expert in some, some stuff like that. And it's happened to me too, like the random rabbit holes where I posted something I knew like 5% about, people reacted to it so well. So I'm like, okay, maybe I should learn some more. And then I became better and better and better over time. So generally my view is post your alpha. There's nothing really that has that much alpha on virality and thought provoke, I mean, provocative content. This is a hard one because you, I realize very quickly, and most people will realize when they, if they post often enough, that there are hacks to virality. You can, you can be outrageous, you can be flagrant, you can say crazy things, you can be extremely funny, you can crack jokes. These are all great hacks to violence, to, to virality. Talking about, you know, divisive news events, talking about generally divisive stuff and having a side controversial stuff. These are all great ways to go viral. I understand that. And honestly, it is very hard for me not to do it because I like that kind of stuff, so. But I tell myself it is not worth two things. Number one, just because something gets likes does not mean it's useful. And you learn that very quickly just by looking at numbers. So likes and follows are not correlated. Sometimes you get 300 likes on something and you get 200 follows. That's an important post. Sometimes you get 10,000 likes on something in two followers follows. This is actual numbers. They're not correlated. What you really want is for people to care about the things you say, not for people to like the one odd thing you say. It's kind of like a person at a party. They can crack a hundred jokes and they're very funny and fun to be around, but you're like, are you gonna think that person is the coolest or smartest person there? No, it's like, you're just funny. People like being around you, but that's not the point. So similarly, like, you can have engaging people at parties that are not, you know, are not the smartest people at the party or the most important people or useful people at the Party. So I think virality, for me, it was very important to draw my lanes. I'm like, here are four or five topics I care about and I have a set of principles that I will follow when I post about these things. I'm not going to post divisive content. I'm trying to kind of try not to do anything that's too provocative and stay within the bounds of reason that I think I would like to see in somebody else else. Like, I don't want people to be like, always like, hey, look at how I screwed this thing up. Or like, oh my God, I look how shitty this thing is. That's crazy. Like, you can do that and it's great. And I'm friends with people like that and I do that in my personal life. I just don't think it builds a good online audience. You want to be seen or I want to be seen as somebody who is useful, factual, trustworthy, doesn't take sides on most things that you don't need to take sides on something like someone like that, that. And if that's the person I want to be in order to be useful, that's what I'm going to try to do. And if, if I have to cut back on my virality for that, I will. That's okay. So that's on virality. When it comes to. To. You talked a little bit about being useful or another way to think about how to post content, what kind of content to post. I think the way the principle I use is it's not easy to do. It's way, way harder to do than it is to say is I write, I read my con, I read what I'm gonna write, and I'm like, if I'm trying to imagine myself as the average person reading this post that probably follows me and think, would I gain value out of this? It is extremely hard to do because most people don't have calibration to what they find interesting and what their audience will find interesting. Because, and I see this on with all of my very, very smart friends. I have so many, like, genius friends who are way smarter than me who don't post online. Right? They have many, many reasons to not do that. And often the main issue is their bar for interesting. Everyone's bar for interesting is based on their context. So everything they already know for like two years they don't find interesting. And they say it out loud. And I have to tell them, dude, the thing you just said, fascinating. And they're like, this is obvious. And I'm like, well, it's not obvious to everybody. And, and, and it's a, it's a constant dance of figuring out what is obvious to you and useful to other people that you can share. And so I try to do that as much as I can. It's like what is useful for people to share, not like what do I find cool? Because often what you find cool is a very ill formed, ambiguous idea that it's not interesting or useful. And so it's a lot like college essay writing. It's like what you think that the college admissions officer wants to hear is like a really ill formed, not good depiction of yourself. The real good depiction of yourself is so obvious that you don't even want to write it down. You're like, oh, this is not that interesting anyway. So that's on usefulness then. The last thing you asked was cadence. There's times I do four times a day and I'm not proud of it. It's usually when I just have a lot of free time between my meetings and I'm like, I have half an hour between a meeting. I don't know if I can get any real work work done. Let me, let me post something interesting. Generally I try to do two one to two a day. So I have my morning slot, evening slot and my morning slot is usually around 9 to 10, sometimes earlier. My evening slot is usually around 6 to 8 and those are the two ones I try to hit tactically. I use this app called Typefully to schedule. I don't do post different posts for different platforms and usually my metric is I will definitely post on Twitter and if it's LinkedIn appropriate I will check the LinkedIn box and it will also go to LinkedIn. That's kind of generally what my, what my any tactical stuff around it is. And a lot of that just came organically over time. I tell myself minimum one a day and if you can too, that's about it.

Do you engage on the content? Like do you comment on other people's stuff for the sake of comment on other people's stuff?

Not for the sake of it. I, I only do it when I'm honestly this is pretty embarrassing, but some it's really fun. Sometimes you just have an evening free and you're like, oh my God, a viral tweet. Some people are commenting on it, let me engage. And then you just spend like an hour just reponding, responding to people about random stuff. So there's days when that happens. That's probably once in two weeks maybe or once a week. Generally I only respond if I happen to check my phone and there's something interesting that I have to say something to. I don't do it to boost the algorithm or anything like that. And you also asked about algorithm stuff in general. I generally try to not care too much about that. There's some things you learn organically over time that the algorithm rewards and doesn't reward. But again, it's like I, I think it's a fool's errand to try to reverse engineer the details of that and it's honestly too distracting. Like I have my day job, I already spend a bunch of time doing this. Like as long as I can get interesting stuff out, I'm not bothered about how the algorithm treats it.

Can you. Since like I want to talk about one investing thing before we like, I know it's supposed to be an investing up though, but like anyway, so how do you. Can you walk us through like one company that you have, how you meet the founder basically sourcing deal and then like how do you diligence the founder founder and winning the deal to thinking about exiting?

I, I think venture today is extremely crowded and there's a lot of capital chasing a very few great companies. The way the most valuable two things in venture probably right now are sourcing and winning, it's not diligence and, and I mean even though helping the company later on is useful, it's just non determinant of your success in some way. So I think those two are the hardest ones. Now when there's a deal process where a founder comes to me and says hey, I'm raising X amount, I'm trying to close this round in two weeks, I'm talking to XYZ, other VCs and I'd love to pitch to you as well. I take those meetings, I do all of, all of the stuff. I generally don't like that dynamic. This is something common terms I guess in venture, but this is what they call a shotgun wedding. I don't think that if I'm investing at a seed in series A in a company that in two weeks I can underwrite this business. Not from a business point of view, it's from a team point of view. It's the same way of saying I can't be best friends with someone knowing them for two weeks. And in our style of investing at Menlo, each partner here does two deals a year. Those two deals are essentially like, it's like a love marriage, right? Like you can't. If I'm Going to work with this company for the next seven to 10 years, two weeks is not enough time for me to assess whether I can get along with this founder. So usually my approach and our approach broadly at Menlo is you form relationships with companies that you're very interested in. And I'll give you an example of that. So the anthology fund also goes, serves that purpose as well for us. We did the series, I led the series A in a company called Good Fire. I met Goodfire and Eric at the Seed and I remember he pitched to me. I loved the pitch. I loved the team's background. I thought this was an extremely important problem. I'll get into what they do. And I said, look, I, I'm interested. But eventually I was too late to the process where they had just signed a term sheet with somebody, you know, he'd worked with before. And I respect a lot from Lightspeed Namdi, so great. I was like, look, I still want to be a part of this. And what I'm going to do is I'm going to write a small check. If you allow me to. I would love to write a small check from the anthology program. Let me show you how valuable I can be in the next coming months. And let's like, let's keep talking. Let me build that relationship. I did that over time. We spent a lot of time just emails, chatting about the future of the company, direction, how to do a bunch of things, hiring. And I was like, I care about you guys enough to do all of that. It's not to win the deal even. I mean, for me it was like, I just care about this company. And so eventually when it came to a point where they were like, look, you know, I asked them how they're doing and they're like, look, we might be gearing up for a fundraise in two months. And this is how we're thinking, thinking about it. I was like, this is going to be us. Like, I, I want to do it and I want to do it before you even go out and have a process. And let's, let's figure out a way to make that happen. And because, and in, you know, this is just how crazy Venture is. In about three days, he didn't run a process. He just emailed his investor lift serve of a bunch of email people on email that he was in, in contact with. And in three days they got seven term sheets and he chose us because he knew us. I mean, this is not, again, it's not rocket science. It's the same way like, you know, like, like dating works. Like you don't you, you know and you trust a person you've, you've, you've like worked with for six months. Like I've, I, I had that kind of trust and credibility versus maybe some other people, didn't they, they're great too, but they just came in a little bit late and they didn't know that person for long enough. So that's how I would go and, and then go win that, that deal. Yeah. So, I mean there's not much to say. So what Goodfire does is they do mechanistic interpretability on, on large language models. So it helps you understand. I call it brain surgery. For LLMs and all models, even outside of LLMs, we think it's really important for things like recently GPT4O had a sycophancy problem where it was glazing all of their users. And you know, those are kinds of problems you can catch on a safety interpretability level that you can tell that the weights are aligned with something that is not appropriate. We think that that's one approach but it's a very nascent field and this is a field that's going to be more and more important over time as AI applications are deployed and more and more important critical decisions. So it's a longer term bet on, on something like that. It's not a bet where you're going to see revenue go from 0 to 100 million in three years, not, not in the next three. So that's kind of the premise of the company.

I wanna, I don't know if you have like one more minute.

I have time.

Okay, great. Maybe we do like one more question. Although I have like literally Some roll down 75 questions. Okay, can we talk about seat strapping? Because like obviously yesterday I was speaking at a conference. So basically we're talking about C strapping. So like you know, people just like.

Give me one second.

Yeah, okay.

I think I pressed a button and now you're muted. Yeah, this is better.

Yeah, I can hear everything.

Okay. Okay. Yeah. Can you repeat that?

Oh yeah. Seat strapping. So basically I was speaking about Wednesday. Yesterday we're discussing about this on stage. So I was thinking because of like right now we only have some, I guess like seed precede emerging manager fund and then the later stage fund of, I don't know like the AUM game funds and then like the sea. Strapping is kind of perfectly fitting into that scenario but probably not for a long term because these like tiny funds want to raise the Next funds and the next one and the next one. And then if people only raise seed and they never race again until ipo, what will happen? Do you think that's actual effect that like currently AI will have on startups or do you feel like it's a fake scenario?

I think there are some companies that may be able to seed strap themselves. I have seen a variety of companies where that happens and I do think that broadly there's a good case to be made that that that could happen for a larger segment of, of companies. Not all companies, but there are some companies who might never need to raise. After 5 million, you know, you can be profitable, you can keep the team very lean and you can grow into a place where you don't have to dilute yourself as a founder anymore and go through to IPO that way. A couple things about that. Number one is one reason people don't want to do that often is because competitive risk. If you seed strap and somebody else is trying to compete with you and has a hundred million dollars, it's still hard. You know, they, they can out drown you in budget in marketing, they can drown you in in hiring and resources, they can build features faster than you. So what would have been a great business in isolation now suddenly looks like you're competing with somebody with more resources. So there is that element of seed strapping where this is why I say it works for some things, it doesn't work for everything. Now if you have something so unique and so differentiated and such a moat where doesn't matter how much money you put into your competitors, you're still going to win, absolutely go do that. So that's what I would say. And there's also this effect in venture at large where you know, probably not the most advisable for me to talk about it, but look, there are a lot of funds who are raising a lot of money. And if you've raised a lot of money in your fund, the math of venture is that you have to deploy that money into essentially growth investments. So because if you raised say a five billion dollar fund, how are you going to deploy five billion dollars in call it three years unless you write huge 200 million dollar checks into growth stage companies. So in many ways you also rely on there being a growth round in order to deploy that fund that, the big fund that you, you just raised. So anyway, there's the, the venture reason why it exists. There's a reason why companies want to not do it. Don't dilute yourself, build a great business. Perfectly acceptable there's some competitive risk. When other companies in your, in your segment decide to not do that, can you continue to beat them? Don't know. And so that's how I would sort of paint, paint the, the whole picture. And then the other thing that people miss is I think, think the numbers here are pretty important. It's not. I, if I recall correctly from the data I've seen, there are more people seed strapping and bootstrapping their companies, but that is not at the expense of fewer people building traditional startup businesses. So both of those things can exist simultaneously too. You can just expand the pie of things that are around to be built.

What do you think about like, the future of hiring? Do people like. So right now everyone's vibe coding, but I try to vibe code. It's like essentially you still have to understand a lot of the coding concepts. So it's not really like I write human language and the app just showed up and I realized you have to use ChatGPT. You actually write a prompt to actually prompt that thing. So, but maybe I'm wrong, but like, I guess, like, how do you think about like the future of work? Like what kind of people will actually be hired? Is that like the engineers who have ability to write prompt or like, I guess, like how is like the hiring perspective kind of like shift in the AI era?

I've never been a fan of leetcode, so I, I thought that was always a terrible way to hire anyway. So I thought, I don't think, I think hiring was always broken. I don't think that's a new AI phenomena. AI made it more obvious. But man, like, I know so many people who are absolutely trash at building products who have great, great jobs because they know how to leak code. This is just facts. Everyone knows that in tech. There are people who are directors and VPs at big companies who, I kid you not, can barely write code, barely build anything of use. This is always the case. There were after a point, it was very easy to get a job because you could lead code, get a job and sort of grind your way through and then get up the ladder. Today it's very obvious that that doesn't work. And startups have always had a kind of different bar even before. The way I think about hiring is, and especially I glean to a large extent is there are some baseline skills that I want you to have. If that means coding, if that means vibe coding something great, I'll give you a complex vibe coding task. Hopefully it won't one shot it and you have to figure out the bugs and debug it. That's great. As long as you've accomplished that final goal. I'm indifferent of what tools you used. The second thing is, it's also very clear that there are a large part of engineering is not even coding. I mean, not even Vibe coding. It's like debugging. It's the most annoying thing in the world. And so I like to focus a lot on hey, can you understand complex systems and debug them? So one classic kind of interview that I like is there's a GitHub repository of, of like a, a blogging framework. There's one bug or one feature request that I want you to fix. You've never seen this framework before ideally. So if this question leaks, it's really hard to ask it. If you've never seen this framework before, go get your ide, bring whatever you want to use and figure it out. Solve, Solve the problem. It's my job to choose a problem that is sufficiently hard that I don't think Vibe coding would work. And it's a real world problem that it wouldn't work. If you can solve that problem and Vibe code it or prompt enough to get it, great. I don't care as long as you solve it. So that's another technique I really liked. And then the last three and four thing I look for is number three. I think I look for niche expertise a lot. It's always mattered in hiring. Vibe coding is really good at doing stuff that is generically useful. It's really bad at doing extremely fine grained stuff. And a lot of the real engineering, quote unquote, real engineering of a company is like one of the things we had to do at Glean is absolutely tear apart elastic from the ground up and rewrite it. Vibe coding isn't going to do that for you. There's not enough data. There's. There was literally nobody who had that we could find on GitHub who'd ever done that. Otherwise we would have just used that. So there's not a single other source of data for something like this. For that you have to think from the ground up. And I'm sure Vibe can. Coding can help you with certain parts of it, but not the whole thing. So how do you reason from first principles or have extremely niche knowledge of general computer systems and architecture and memory and how they work is pretty important. And there's various other ways to test for that. That's not lead code. So that's number three, number four and my most Important factor is I think interviews just do not optimize at all for you as a person and your motivations for this job, which is like singularly the most important thing for me when I'm hiring personally is like my, my entire belief is like, look, everyone is like a line. They have a Y intercept and they have a slope. I only care about people with high slope. I think even if your intercept is low, if your slope is high enough, you're, you're gonna be great. So I, I look for people really who are really, really hungry to solve problems, but at the same time like able to collaborate well and, and work with people. People like that I think generally can be very, very successful at at most things. It would be. I would much rather have like a, a bad coder who would, who would, who's ready to work their ass off because they really, really care about this versus like a great coder who's like nothing interests me, nothing is this problem is just not that interesting. And then they keep whining all the time about that and I'm like okay look, you're barely, barely going to give me 20 hours of your week and you don't really care enough to do anything. I don't want someone like that. I want someone who's really, really invested in this. So that's kind of the two, the four things that I would look for.

Love this. Okay, so I want to end with a 1 minute fire round because though I really want to keep chatting with you for the next two hours but I feel like you still have to new word but okay, let's do. What's your favorite book?

My favorite book is probably the Namesake by Jumpalahiri.

Who would you invite to your dinner party?

Who would I invite you at dinner party? I would probably in this is funny. At this moment in time I would love to invite R for a rock on Twitter. But in, in actually all of history I would love to invite. You know, I don't know. Great answer to this.

Probably Michael Jackson who made the biggest impact in your career.

The biggest impact on my career was. Was probably my first boss at Google. This guy's name is Al Carmi. He was a, he was an Israeli guy. He made a huge impact. The one thing he would always tell me is, um, if you can do technical stuff, never stop doing technical stuff. And it stayed with me ever, ever since. I still code all the debt all the time even though I'm in venture. Oh my God.

Where can we find yourself? Work.

I spend a lot of time reading, but if I'm outside the house, I just like, you know, pretty basic stuff. I like to. Going on hikes. I like to. I like to run, I like to play sports. I. I play cricket on Sundays and I like doing triathlons and stuff like that. So. Pretty basic Bay Area type activities.

Love. Oh, like last. Last question. Only for you. How do you be so articulate? You are listed code the whole way. Like, I love it. I was like, how do you even think about. Like, I feel like even I have thoughts, I cannot be so articulate about them.

But I. I'm glad that you think that. I don't know. I think. I think in terms of how I tweet.

Well, I'm there. But anyway, Didi, thank you so much. It was such a fun conversation and you are so. I feel like. Wow, you're so living up to your hype.

Oh, thank you so much. I thank you so much for having me on here. I love your podcast. I've seen a bunch of these episodes and. And you're doing an awesome job, so really appreciate it.

Thank you. Okay, let me quickly add stream. Sorry, let me. What? One second.