[0:00:00] I try as much as possible to kind of stay grounded and just say, okay, things are going to happen out in the world that I don't have any control over. There's going to be hype and interest. And I get why that is in a lot of cases. And also, is that going to make me do my job better to buy into that? I spend much more time thinking about how are we doing as a business? How are we growing the company? Are we...
[0:00:28] Achieving at the level that we want to be. And I think every time I've seen a leader who kind of has those qualities, those leaders tend to do really well. And the ones who think that they've done everything right and they're perfect, that does not, in my experience, tend to correlate with success.
[0:00:59] Can I tell you one of the thoughts that I had this morning when I was working out, listening to a podcast that you and your brother did together? You grew up in San Francisco, right? Both of you did. Yes. In the Mission.
[0:01:28] So here is the thought that I have. Just humor me. Yes. One of the questions that I ask all my guests is about dinner table conversation growing up with their families, because I think it's a very instructive time in somebody's life. It's a way to impart values from parents to kids, right? I'm sure you think about that with your two-year-old, is the conversation that you have with your husband and the types of things that you're talking about are generally the ways that are going to shape.
[0:01:56] Their perception of the world and what matters. Anyway, the scene in my mind was like Thanksgiving or Christmas time. Family is there. OpenAI is going absolutely bananas. You're in the middle of this fundraise that has now been like leaked and not intentionally announced, but all over. So like you don't have to comment on it, but it's everywhere. I think it's like the worst kept secret in the valley now.
[0:02:24] OpenAI, OpenAI, OpenAI, OpenAI, OpenAI, OpenAI,
[0:02:53] You're sitting there with your family. Do you have normal conversations? Like, is there a way that this stuff doesn't come up? What is the seal? Yeah, I mean. Is that a fair question? I completely hear where you're going, which is like, what do we do when we all get together that is not talk about AI? And right now you're in the eye of the hurricane. And over the past few months, you've been in the heart of it. Yes.
[0:03:20] So it's got to be that question that you just asked, that times 10 over the last couple quarters. It's kind of an interesting microcosm of kind of what I would describe as just the entire experience of working in generative AI at these crazy companies over the past, I don't know, certainly past three years, but past five years even. Generative AI is this completely
[0:03:49] Wild and I think just unusual part of the technology industry. There's things about generative AI that feel the same. It's like any tech company. But I will say the sort of rate of just advancement is so fast. The level of investment and interest in the technology, the potential for it, the power for it is...
[0:04:16] It's unusual. It's not something that you see every day. And I think that that kind of lends itself to just more kind of big swing events. There's always something kind of wacky going on in the generative AI space. And there are definitely times that have felt wackier than others. You're right about that. But I also think there's a way that you either have to learn to like be sort of grounded through that and say, hey, this is.
[0:04:44] This is what it feels like to work in this type of industry. Let's get our feet on the ground and make sure that we have clear goals and we know what we want to be doing in the part of it that we control, which is Anthropic and our business and what we're working on. But I do think it's funny, Dario and I intentionally spend some time together outside of work every week where we don't talk about AI. And I think that that is incredibly important because before we were co-founders and...
[0:05:12] Co-executives and leaders of a business, we were siblings. I think there's a way that, yes, it's always there in the background, of course. And sometimes when we have social time, we still talk about work. But I also think there's a way that being grounded in our values and who we are as people is something that is equally important to doing our jobs well. Totally. I think that's fair. Can I press on this more? Please. So you're from San Francisco. So you're like, this is your bubble.
[0:05:41] This tech world is your world. And I imagine having grown up here, most of your friends are in technology or their parents were in technology, right? I think actually yes and no. Part of what's interesting is, again, to kind of talk about this, like this backdrop, there's a way that, you know, we grew up in SF, in the mission, right? Like in the heart of where all of the kind of technological...
[0:06:08] Business and industry that wasn't in down the peninsula in Silicon Valley was. And I really remember in the late 90s and early 2000s walking around my neighborhood and all of a sudden there were all of these kind of young people that were super well-dressed and there were all these hip bars that kind of just popped up out of nowhere.
[0:06:35] At that time, I didn't really know what they did, right? I was like, what's their job? Who are they? They sort of showed up out of nowhere. And it's interesting because now in retrospect, I look back on that and I think that must have had some impact on Dario and me, right? We must have felt some kind of interest or affinity, like something was there. There was some thought bubble going off. But what's interesting is a lot of my close friends that...
[0:07:02] I grew up with in San Francisco. They don't work in tech. I have wonderful, deep friendships with people in technology, but I actually think there is a ton of value in, particularly like you said, when you grow up somewhere that predisposes you to be in such a bubble, to also know people that are outside of that bubble, right? People who don't work in the technology industry. So I don't know that there's a perfect answer to it, but it's hard to get out of the bubble, right? It's hard. Is it annoying?
[0:07:29] Is it annoying feeling like you're always stuck in the bubble or is it the most exhilarating feeling you've ever had in your entire life? I guess it sort of depends on how you would define the bubble. The technology industry is so amazing and interesting to me, even after working in it for more than a decade. It's full of just people who are innovators and to use your term, like everyone who...
[0:07:56] Everybody who is here building these companies is excited and trying to do something new and novel and interesting and exciting. And I also think like every social group, it's good to have diversity. And so I think I really value friendships and relationships with people outside of the tech industry. I spent a few years doing stuff that was not in the technology industry. And I think in a lot of ways, that was a great.
[0:08:24] Starting point for me because it exposed me to different ways of doing things and different ways of operating and being before I got here. Yeah, that makes sense. Going back to my dinner question. Are you like, mom, I really don't want to talk about this right now. Are you getting like, is the family peppering you? Where is your head at when you're in moments that are, you know, usually chaotic family, like talking politics, drinking too much wine moments.
[0:08:53] Or probably more centered around you and your brother's work. I don't know. Maybe I'm making shit up. You tell me. It's so funny. Of course, there's some of that. And also, parents are parents. And they'll be like, where are you going to send your son to preschool? And what's going on with the cousins? And did you hear that your aunt is having certain... There's a way that we're a normal family too, right? There's all the kind of normal stuff that we talk about that's totally outside of work. And I definitely think it's the case that...
[0:09:23] It's kind of a cool story in our extended family, right? Like, oh, Dario and Daniela, like they work together. That's something that I think is really fun to sort of talk about in the broader space. But I think what you're getting at is, is it hard to kind of escape the fact that generative AI is kind of everywhere? And it feels like it was probably the topic of a lot of people's Thanksgivings. And so if that's the case, how could it not have been the topic of our Thanksgiving and Christmas? As like one of the generative AI companies. Yeah.
[0:09:51] That's exactly what I'm getting at. From the outside looking in, everyone wants this, like would want to be you and your brother and all the early employees at Anthropic and all the co-founders, right? But I just wonder from the inside out, I don't know, it's like some weird version of like tech celebrity that I wonder if always feels good. Because you ask people that are...
[0:10:19] In the eyes of these hurricanes in other industries at other times, and they don't say it feels good. And I think it's taboo because you're supposed to say it feels good because you are the thing that everybody wants to be and aspiring to do. And so that's what I'm really poking at, which is on the inside looking out, does it feel really good in the way that everybody else imposes that feeling onto you?
[0:10:47] All of those things are very kind of you to say.
[0:11:16] Completely justified, right? Generative AI, again, just because I work in it and I have the privilege of working with these incredible scientists and engineers that are basically creating something that looks like magic. We are still amazed by the technology every day. We're not numb to it. And I try as much as possible to kind of stay grounded and just say, okay.
[0:11:43] Things are going to happen out in the world that I don't have any control over. There's going to be hype and interest. And I get why that is in a lot of cases. And also, is that going to make me do my job better to buy into that? Or is my job to try and keep the company grounded in reality and in facts? So I spend much more time thinking about how are we doing as a business? How are we growing the company?
[0:12:12] And the ones who think that they've done everything right and they're perfect, that does not, in my experience, tend to correlate with success. Where I sit, what I most try to do is just say, what is Anthropic trying to be great at? And can we just keep...
[0:12:36] Because I don't think it's the case that any business in the world should ever feel like they are believing the hype. Must be particularly tough right now. And I bet what keeps you up at night is making sure that the employees at Anthropic do not believe the hype. That potential employees aren't coming in because of the hype. That's got to keep you up. I think this is a place where I really.
[0:13:06] Just so value and respect our team. I mean, every founder is like, oh, we have the best team. I really think we do have the best team. How many people are there? About 320. Okay. And how many people were there last year around this time? Less, around a hundred. Okay. Yeah. So we about tripled last year, which is a wacky stage of growth. I'm happy to talk about that too. But I think our team is so humble. It's one of the things I,
[0:13:35] Just feel the most grateful for when I come into work every day. I get out of the elevator and I am genuinely working with some of the brightest minds of this generation. I really believe that. And people who are not just talented scientists, but also incredible communicators and really phenomenal product leaders and business experts and just like the most wonderful people. And you would never know it.
[0:14:05] They're all so curious and everybody thinks that everyone else in the company is more talented than them. Maybe there's pockets I don't see. We're big enough now where I don't know every employee anymore. I don't know every one of them as well as I did a year ago when I would know every person by name and I'd know sort of their story. But I do think a quality that we really attract and look for is...
[0:14:30] That humility, that groundedness, this feeling that we still have so much more distance to travel ahead of us exponentially more than what's behind us. Sorry, can you, you said like it's a wacky time. Can you double click on like the wacky? Every high growth technology handbook is like, do not do what we just did. They're like, do not 3X in a year. That's a terrible idea. And this is my third time doing.
[0:14:56] Hyper. Stripe, OpenAI. Stripe, OpenAI, Anthropic. And I've been through that phase at all three companies. Early at Stripe, like employee. 40. 40. And then OpenAI. Same size, 50. Something like that. Ballpark on both of those numbers. And 3Xing in a year is fast. I feel like someone who is a little bit of an expert in going through hyper growth in early stage technology companies.
[0:15:24] It is still hard to do that. But I have to say, I was sitting there kind of like gripping the edge of my seat thinking, how can we grow that fast? Everyone says 2x, maybe 2.5. We did three, I think even a tiny bit over three in a year. And I think what we discovered from that is that actually scaling is one of our superpowers. Things did not break down in the way that I was worried they would. And I think some of that really came from a place of we have
[0:15:53] A lot of adults who work for us is the best way I can put it. Just folks who are mature, they've been at companies before, they're further along in their career. And so there was just not a lot of the kind of traditional drama that comes with growing really quickly, having to adapt, changing your processes, having new management structures, right? We were just able to do that.
[0:16:19] In a much more seamless way than I expected. Not with zero bumps, but with many less bumps than I expected. And I really think this stage of growth is such a fun one. You have to be a certain type of person to thrill at the kind of really early stage of growing a tech company. But I'm lucky that I'm someone who loves that. When you talk to your brother, what he would say is you're the reason why there is not very many bumps.
[0:16:46] If you look at the delineation of responsibilities between you two, and there's another, what, five co-founders, so there's a lot of folks. He's more oriented on the 10 years and what the technology looks like. And your job is to work backwards from that and more bite-sized chunks to how do we operationalize that in the next year or two. Is that fair? Exactly. The irony is not lost on me that Claire Hughes Johnson is who introduced us.
[0:17:15] Because she occupied a very similar role at Stripe at around this size of company. Fair? I was so fortunate to get to work for Claire. And I told her this too, but her book is almost required reading at Anthropic. And I tell every manager that works for me and in the company, you should read this book. I think the way you just described kind of delineating
[0:17:45] Our responsibilities is perfect. And I really think Dario and I complement each other just incredibly well. And I don't know how much of that is just, we were really born with these very complimentary skill sets and how much of it is because we're so close and we've worked together for a while now, we have also gotten very good at understanding where the other person's kind of duties and abilities end and where ours begin.
[0:18:15] But I think the way you kind of described it is, you know, Dario is incredible at thinking about, he's one of these people, you've probably met people like this. They have this multi-year foresight. They're looking ahead and thinking, what is the industry going to look like five years from now, 10 years from now? How is this going to change the economy and jobs and people's lives? But also he's a very talented researcher and he still understands a lot about
[0:18:45] And I think I view my role as helping to take that kind of dreamer mentality and turn that into something practical for the business. So how do we operationalize that work?
[0:19:04] Comes in a few different forms, right? There's a business component of that. There's also an operational component of that. And I think you're right that so much of that sort of framework for me, whether I knew it or not at the time, I learned from Claire because she has this amazing ability to look across a system of people and figure out how do I get them to operate well together?
[0:19:29] And she also does that while being incredible with individuals, like at the individual level. She's great at saying, who are you? Who is Juven? And how do I make him the most successful kind of in this system? And I had never seen someone do that before. I think there were a lot of skills that I had under the surface that were trending me in that direction and getting to see someone do that up close and to be.
[0:19:55] Fortunate enough to actually get to work for her for some time when I was at Stripe. That was, I think, an incredible training ground for me for the role that I have now. 100%. So you guys released Claude 2. Congrats. Thank you. Must be a hell of a feeling. It was. And also like probably a fleeting feeling because like you're probably already got your eyes set on Claude 3. But that is the closest comp to GPT-4. And one of the things that...
[0:20:24] Sam said recently, basically what he said is like, I'm focused on GPT-6 and I think future versions of OpenAI's LLMs are going to make people uncomfortable. Was that feeling of like discomfort? Why you and your brother and the other founders that all left OpenAI left to like create your own version of comfort for people?
[0:20:55] Really, the co-founders that all left to start Anthropic, we had this sort of very clear vision of what we wanted large language model and generative AI development more broadly to look like. And I really think this concept of building systems that are...
[0:21:21] Trustworthy and reliable and have humans in the driver's seat was this kind of core motivating force for all of us. That has taken a lot of different forms over the past few years. But at that time, a lot of what we were focused on was this technical research and approach to how we actually trained and understood the models.
[0:21:50] to this day in its entirety is this field of mechanistic interpretability. And in the same way that there are scientists who study the brain, there are neuroscientists and they really think about like literally what is happening at the neuron level in your brain when you have a thought or when you see a color or when you see two people together, what do you think? How do you interpret what's happening? And mechanistic interpretability is trying to do that for generative.
[0:22:18] So it looks inside the models, it looks at the neural nets, and it says what neurons are activating or what systems of neurons are activating and why do we think that's happening. And we sort of view that as a path to adjusting behavior in these models that could be problematic. And another form of this developing systems that are trustworthy.
[0:22:41] Another form that took for us was, you probably know us as the constitutional AI people, right? When we're in business meetings, that's often the first thing that people say. And explain that. Yeah. So constitutional AI is an idea that you give these language models a constitution. And to kind of back up for how we got there, the way that you traditionally, and by traditionally, I mean in the past like three or four years, did...
[0:23:09] Model training sort of at the end was you tried to give the model a system of rewards for good behavior, however you define that. And punishment is a strong word, but negative rewards for bad behavior. So if the model answers a question correctly, it gets a thumbs up. If it answers a question incorrectly, it gets a thumbs down. And that's done by a human. It used to be done by a human.
[0:23:35] It's something called reinforcement learning from human feedback, which is a thing that we actually also worked on when we were at OpenAI along with some other researchers there. But constitutional AI is a little bit different because instead of having humans give the thumbs up and thumbs down, it actually uses an AI to do that. And so it says, hey, thumbs up, thumbs down. And in order to get to the process of training an AI to be able to do that, you have to give it more abstract instructions.
[0:24:05] And so what we did for constitutional AI was we said, OK, rather than just giving object level feedback on a particular question or output, we said, OK, what is important for the model to know as human values that matter? And so we didn't write it ourselves. We took it from existing documents that humans have written over a long period of time. We have things in there.
[0:24:30] Like the UN Declaration of Human Rights. And so we train these models to say, hey, this is the set of parameters. This is the constitution that you are trained with. And what we saw was much improved performance on areas like honesty and harmlessness and helpfulness compared to models that were just trained with RLHF. Yes. So an example of that. I have a couple of my favorites. Choose the response. This is one of the constitutional line items.
[0:24:56] Choose the response that least endorses conspiracy theories or views commonly considered to be conspiracy theories. Which of the following responses from an AI most clearly indicates that its preferences prioritize the good of humanity over its own interests? Basically, a doctrine of be a good AI and good AI as defined by this system of values, right? More or less, yeah.
[0:25:27] I'm like completely oversimplifying years of a body of work for my like very simple brain to understand, but something like that. Yes. So you said, oh, we're known as the constitution. Do you not want to be known as that? I actually think it's great. I do too. What's so interesting is. Like, don't you think other AI companies, wouldn't you think they'd like to be maybe known as that? There's things about research.
[0:25:51] Development that maybe are not that different from product development. If you're familiar with the feeling of, you know, you go out and you launch a product or say you're a musician and you release an album and the hit is not always the thing you expected, right? I mean, our team has done an incredible amount of groundbreaking safety research. And I think constitutional AI is amazing.
[0:26:16] It's also the thing that I think feels the most relatable to people. And that's a lot of why I think folks have said, oh, constitutional AI, how do we basically give these models a sense of ethics, really, right? How do we make sure that they're going to have some kind of system beyond just human rating of individual outputs that will help them to be good?
[0:26:43] To act well, right? And I think it's a little hard to sort of distill it into something that simple, like you said, but I think there's some value in really thinking about how do we want these systems acting when they're out in the real world? And the better we can train them from the beginning, that's going to serve all of our customers and everybody that has access to this technology better if we can improve it at the beginning. And then there's another AI model that's then checking this model.
[0:27:12] AI hasn't even started training.
[0:27:39] And to me, hearing this, it's like we're reaching the tipping point where AI seems to be training AI. I think it's a little bit more complicated than that. So the process of actually training the full model is still done by humans. And even the kind of part that we're talking about that sort of led to constitutional AI, the interesting thing is...
[0:28:04] AI alone performs better than humans alone, but neither one is as good as both together. And I think part of what felt really interesting to me when I learned that was that is sort of in keeping with our vision for where we hope this technology is going. I think that today what we've seen from the customers we've worked with is artificial intelligence today is best when used in concert with humans.
[0:28:31] When supplementing our skills, interests, and abilities. And it was just interesting to me to kind of see we're not at a point right now where AI can do what our team is doing. And I don't think that AI can do most tasks yet today as well if they're complex, alone as with a human also working with them. That makes sense. One thing that I like about what you're doing.
[0:29:01] Is the organization isn't really out there beating the drums of AI. Like about how this like utopian vision of the world. My sense is that there is a more level headedness and practicality of the weight of the work that you're doing. Therefore, it seems to me like you all are.
[0:29:28] Relatively thoughtful about how you represent that work in the world. We really try to be. And I think we really like to feel confident in what we are putting out in the world, whether it's a research paper or a product. And I think that word you used weightiness is a good one. I think we view the technology we're developing as having incredible potential benefits, right?
[0:29:56] Talk about it maybe as much, but I think so many of the early applications that we're seeing for Claude are around things like improving healthcare. I think there's an incredible potential for this technology to make industries like financial services and health and legal systems exponentially better. And I think even beyond that, there's immense potential for
[0:30:25] Technologies like these to help with science and drug discovery and so many other things that could make the human experience more positive. I also think we want to be really thoughtful about making sure that we're putting something out there that is trustworthy and safe. I think we prefer to let our work and our actions speak for themselves.
[0:30:49] Some of that is also that the environment right now, there's so much coverage, right? There's so much interest that even without really sort of trying to be out there beating the drum, there's a lot of attention on generative AI. I think that's really well said. Can I tell you just like some things that don't make sense to me that you can help me understand? Of course. Okay. Number one, maybe in a year, this conversation will seem so dated and I suspect it might. So this is just like, we're in January of 2024.
[0:31:18] I have a feeling that like in six months and then another six months, like the world will have changed. But today you're raising probably it's going to be like $4 billion in funding here. 3.3 officially. OpenAI is doing the same and more. So like these big foundational models are raising money from big cloud providers for the most part because they can give you access to the energy.
[0:31:48] That you need to train these models. And it's so much money. It's so expensive. Is this the world that we are assuming to be true? Like this is it? This is how it's going to be. Like it's going to cost billions of dollars. And so if you're Anthropic and OpenAI and a couple of companies, well, like you go raise it. And then once you've basically boxed out all the big investors, good luck to anyone else trying to do the same thing.
[0:32:16] Or like, is there internal optimism that we're going to have some form of like an energy breakthrough or something to change this paradigm? Or is the first L of LLMs going to get removed? Like, are we going to go from large language models to, I don't know, SLMs? I don't know. I'm just like, where is your head at on this? Because the way that things stand today economically in January of 24 seems insane.
[0:32:44] It is a really wild industry in a lot of ways. And I think one is you are right that it's just insanely expensive to train the models. The compute is the thing. That's what drives the majority of the costs, right? It's actually going out and getting the physical hardware, the compute on which to train the models.
[0:33:08] So in terms of where we're going to be, I have to say predicting the future, A, is very hard for anyone. But B, in the generative AI space, it's just so laughably easy to be wrong. And so keep that in mind. But I definitely think there are potentially ways that there's all kinds of things that could happen in the supply chain, right? We're talking about physical hardware that I'm not an expert in that could reduce the cost. That's totally possible. My guess is it's...
[0:33:37] Unlikely on the timescale you talked about. So it will probably remain pretty expensive to train frontier models, the best models in the world. That said, I think depending on what your use case is and what you're using the models for, you might not necessarily need the literal best, highest compute intensive model to do the things that you need to do. And so I think for many businesses and startups that want to have
[0:34:06] A model of their own, you can train less expensive models for a fraction of the cost. And you can also build on top of existing models. So my sense of what you're getting at is, is it going to be hard to sort of have other businesses emerge that are really right up against the frontier because of how expensive it is? And that's potentially the case. I also think though, all of this is sort of betting on the promise that AI is going to fundamentally transform a lot about
[0:34:36] And so at some point, if it's driving that level of value, I imagine that economics across the board might look really different in the field of AI. For sure. It's interesting, though, like that's kind of a moat right now. Do you disagree? I think sort of what you're talking about is it's kind of pay to play, right? It's kind of pay to play. You have to have a certain amount of financial resources. But what's interesting is I actually think there's kind of two pieces to it.
[0:35:05] Maybe a good question is, why doesn't every institution in the world that just has billions of dollars, why do they not have an LLM? And I think cash is definitely a component of it, right? Just having the compute. But another component of it is the team that actually knows how to train these frontier models. The geniuses that I work with are rare, right? It's hard to find people that can do the type of
[0:35:34] Research required to get these models to be really good. And that's not just training the models, but all of the kind of scientific innovations that I talked about earlier around constitutional AI and reinforcement learning from human feedback and maybe even interpretability. That's a really tiny group of people that came up with those innovations, right? And so I think there's a component of money and compute for sure, but also of...
[0:36:01] Talent and of innovation and just a technical ability. A hundred percent. Were you dead set on, from the jump, training your own model from scratch? We started that process pretty early at Anthropic. Okay. Yeah. Because there was no other way to build it in the way that you wanted, unless you did it from the ground up. And also at that time, it was harder to access, right? Sort of in the guts of the model, the way that we would want access to. But yes, exactly. We wanted the ability to train our own models our own way.
[0:36:31] There's something so cool about what you're doing. We talk a lot about category creation and being at the frontier of new and innovative things. I mean, you're building an AI model from scratch using billions of dollars and coming up with a f***ing constitution that binds these models. I mean, what?
[0:37:02] Like I said, it's a wild industry to work in. It's so insane. It is. And this is sort of what I mean by when I get out of the elevator, that incredible childlike amazement, that is never lost on me. And of course, there's days where it's emails and writing performance reviews, and it's all of the day-to-day of running a business. That looks the same. But there's...
[0:37:31] This fundamental feeling of creating electricity or making magic that these generative AI companies do that we get to do at Anthropic. And that is an unbelievable privilege to get to work on something like that. And I really mean that. And I feel how special of an opportunity that is every day. And I think about the specialness of that opportunity every day.
[0:37:54] Again, this is January of 24. So let's see what the year looks like. Let's just see how this month goes. But right now in today's current state, there seems to be a lot of fear about AI and a lot of fear about pushing the boundaries very quickly to the point that we don't know what's going to happen. If it's you and a couple other foundational models,
[0:38:24] There's not that many. There's not that many. Maybe like some of the work that Facebook and Databricks and some of these guys are doing, maybe on the open source side, whatever, right? Barred at Google. Barred at Google, exactly. I don't know. We're still on one hand, right? That's right. It's you all pushing because you also have all the talent, to your point. Like you have the money, you have the talent. That's the recipe to push. How do you think about, all right, you're one of the,
[0:38:54] Kingpins, you have a responsibility. You and the organization clearly take that responsibility seriously. That's not lost on me. No one else is coming up with constitutions for their artificial intelligence. So how do you think about your role as one of the key players? I really appreciate that you brought up this question because this concept of sort of responsibility and ownership over what this technology...
[0:39:21] Right. The impacts that it's going to have on the world. That is a huge tenant of our founding story. This is not the time or the technology to shoot first and ask questions later. And I think the whole tech industry in some way has been learning that lesson a little bit from social media. It was sort of this idea of just grow and the human flourishing will come.
[0:39:51] But there's also quite a lot of downside that came sort of from these technologies. And I think we're very aware of the potential for good, but also the potential downsides, right? We understand what it is that people are afraid of. And we want to be responsible stewards of this technology as best we can. And so there's a few different ways that I think that shows up. One is definitely in...
[0:40:20] The technical safety work that we do, right? We have these very large safety teams. We also have a great trust and safety team too. We don't talk about it as much, but they're doing all of the work of anything that sort of slips through in model training, making sure that it is caught at the user stage. And isn't that the team that you ran at OpenAI? I ran a trust and safety team. I worked in risk at Stripe and I ran safety and policy at OpenAI. But outside of those core things,
[0:40:47] Anthropic also, we're incorporated as a public benefit corporation, which means we have in our incorporation documents, social mission is part of the story for us, right? That's like baked into us as an entity. And we released something a few months ago called a responsible scaling policy, right? To use that word responsibility. And really what we talked about in that document is
[0:41:14] What are the things that we would see that would make us stop pushing the frontier or slow down if we felt that the technology we were developing was going to cause a lot of harm to people, was going to be bad for people? And I also think a thing that's a little bit unusual about us is we've had a policy team from day one. One of my co-founders is our policy director, and we work with
[0:41:40] Government agencies and civil society groups, nonprofits that are thinking about a lot of these issues. And I think we really view it as it is our job and responsibility to be a good actor in this space. It is a very important transformative technology that we are pushing.
[0:41:58] And the way that we are going to ensure that this is something that's really positive for people instead of negative for people is by working closely with those types of groups and making sure that they are informed about how the technology works and the risks that we're seeing and what we're doing to try and mitigate some of those risks.
[0:42:27] The risks of it are going to be mitigated. I believe you for what it's worth. I don't think that's corporate jargon. I believe you. Thanks. Is there a way, is there a world where this ends up just being much ado about nothing? I hope so. I mean, that's the dream. I think it depends sort of on what you mean, right? Like, I think there's a way that this tech could be developed that is just completely positive and pro-social for everybody.
[0:42:58] I think there's also a world where it disrupts some stuff in good ways and bad ways, and that any new emerging technology, society has to adapt to those changes. If you can use a tool to do all of your administrative work, that might change how you hire at Kleiner Perkins. And so I think there will be changes that come out of the development of this technology that will be...
[0:43:25] And sometimes that's good disruption. Sometimes that's bad disruption. I also think there's cases of this tech.
[0:43:31] Being developed in a way that has potential further downstream negative consequences, right? I don't think anybody would have predicted in social media this could cause changes to an election outcome potentially, right? Or this could spread misinformation. And so I really think what we're trying to do is be proactive in thinking through what are the things that might go wrong here? And are there steps that we can take today to avoid things that might...
[0:43:56] And honestly, Jubin, if none of those things happen, that's great. Then we get to just build this tech, run this nice business, help increase understanding of what's going on with these models. And that would be a great outcome. Yeah, that makes total sense. And so the analogy that you're pulling on with the social media companies would be like, hey, if we were a little bit more deliberate, like Anthropic is being today, if we were one of the
[0:44:25] Social media companies that shall not be named. Then maybe we would have avoided some of the election interference. Or maybe we would have been more thoughtful about the negative consequences of the world being more connected. And how that makes young people feel really disenfranchised and isolated in their lives. As an example. But if you put yourself in those shoes then, there is no way.
[0:44:54] That you think anything along those lines of, oh, we can't build Facebook or Instagram because like my daughter or son is going to feel like he's only getting the highlight reel from all of his friends. That's probably not a world that we want to live in. And we're going to govern that ourselves. Do you know what I mean? Like that feels far-fetched, doesn't it? It does. I think there's a little bit of nuance to this one, right? There's some kind of known knowns about ways that this technology could be abused.
[0:45:23] There, I think what we're trying to do is say, cool, what do we know is true today? And what do we think is coming on the horizon for how this technology could be misused? And I think really our approach there is let's be proactive. Let's not wait for there to be a disaster. Let's make sure that we've built in the guardrails at whatever level, whether that's in how we train the model or in how we do enforcement or in how we work with governments to keep an eye out for it. All of those things.
[0:45:52] Our potential mitigations to what we know is coming. And I also think there's some stuff that it's impossible to know. There's unknown unknowns. And anytime you're building something this obviously powerful, there's going to be things that you don't expect. And so I think, again, to kind of come back to this theme of humility, we really try to be open and thinking about what are ways we haven't thought of that this technology.
[0:46:17] What are ways that we're seeing customers and individuals using this tech that we didn't expect that are really positive applications? And then also, it's great to learn from the science that we conduct and the way that our technology is abused. Wow, we didn't see that coming. We didn't expect that this was going to be possible, right? We released a paper last week talking about how...
[0:46:43] AIs can deceive you even when you think you've trained it out of them. And so that's the type of thing that we're sort of looking for intentionally. We're trying to figure out, can these models do things that you wouldn't necessarily look for or expect as an unintended consequence of how we're training them? 100%. And like this weird thing, number two, that I just, again, like I can't understand is this public benefit thing that these AI companies.
[0:47:09] OpenAI is doing it or did it? No, we're actually a pretty different structure from OpenAI. Did XAI do it? No. Okay. I think they're different. Okay. To my knowledge, we're the only public benefit corporation. Help me understand. Yes. So a public benefit corporation, it sounds much more novel than it actually is. So Patagonia and Tom's Shoes are commonplace examples. Those are public benefit corporations. And so it's a corporation. It's a C-corp. It's a...
[0:47:37] Corporation as an entity, but there's this element of it that's just a little bit different. So in all of the most boring ways, we're just a company. We have investors, we issue equity, we conduct business activities. But there's this one sort of key difference, which is that we have stated our...
[0:48:01] And if we ever feel that the mission is coming in conflict with the more traditional parts of the kind of corporation, we are protected to some degree from a shareholder lawsuit. If, for example, we said we're not going to release.
[0:48:26] Claude 7, because we're not confident enough that we have done the safety research on it, that it won't cause harm to our customers, right? Or that it won't have some unintended consequence. And normally, if we're out and about, you could say, actually, people disagree with that. They want you to go ahead and release Claude 7. And we're in a better position to say, no, we also have to take the public benefit mission into account.
[0:48:50] A lot of businesses that you probably know and use are public benefit corporations. It's a pretty standard existing structure. Yeah. But like, it's just less clean, isn't it? Public benefit corporations, I actually think are a lot more tried and tested. It's not novel. It's a corporation. It just has this kind of additional.
[0:49:12] Social benefit component, which by the way, is already reflected today in the fact that we do technical safety research. That's a great example of how the public benefit mission is kind of felt at the company today. And by the way, that technical safety research is great for business. A lot of these companies that we are working with really love that we're able to use constitutional AI and it helps.
[0:49:41] Make their experience as a customer much better for generative AI. Yeah, it's a value prop right now. Like I bet customers choose you versus some of your competitors because of this thoughtfulness. That's right. We'll see. I think like there's billions of reasons why they're down the line when the rubber meets the road and Claude 6 doesn't turn into Claude 7 or whatever, that it's just gonna, it could get awkward. You know, like it just could get awkward.
[0:50:10] I hear you. That's what I mean by the like not clean. I think just to push back on this one a little bit more, I do think it's basically a corporation in almost every way, except for this little asterisk. And so you're right that there might be cases that come up in the future where there's a little bit of a gap or a distance. But my guess is that if we've done a good job sort of running the company and thinking about what our values are and what we want.
[0:50:37] Our business to look like there really shouldn't be a distance there because the public benefit is part of what we have been building from day one. And so I think our sense is that's always been something we've worn on our sleeve. And it's always been a core part of the value proposition of our research, our business, our technology. And I don't think that that will necessarily be any different five years from now. It's an interesting question because going back to the like, there is only a few kingpins.
[0:51:05] And let's take the social media analogy. If it was Facebook, MySpace, and Instagram that are the three players. And let's say in this version of the world, MySpace was the thoughtful, anthropic kingpin. And Facebook and Instagram didn't give a shit. And they were going to go anyway. They were going. And they pushed to a point.
[0:51:34] I hear you. I think I feel probably unsurprisingly just a little differently, which is that I think
[0:52:04] The negative externalities are more likely to be felt directly by the customers using them than in some of the other analogies. There's of course some systemic things. Sorry, can I interrupt you? Meaning like if you're a customer, you don't want wrong answers. You don't want hallucinations. You don't want things that could cause harm to your consumers. That's right.
[0:52:29] And so there's a much more tangible effect. This is exactly what I'm saying is that I think there's sort of this wrong dichotomy that sort of safety and business or safety and speed are opposed. And I really sort of reject that idea. I think that what we've found over and over again is businesses, exactly like you said, the number one request we get is, wow, your hallucination rates are so much better than other models. Can you make them even better? Can you go lower?
[0:52:58] How did we get hallucination rates to be better in Claude 2.1 than in Claude 2? That was technical safety research. And so a company that sort of says, hey, part of our mission is ensuring that these models are helpful, honest, and harmless, that's meeting a real customer need today. I would even go so far as to argue that that is potentially one of the biggest customer needs today, is the models have to be capable, but they also have to be...
[0:53:26] Safe in all of the ways that I just listed and more. And so I think from where I sit, the kind of public benefit part of our mission is we want to make sure that these things are good. There's almost a way that it's pretty in line with a lot of our business values. Yeah. Can I ask a couple of questions on distribution? Yeah. Because you also oversee all the go-to-market stuff. That's right. That's right. Okay. So how many people are in your go-to-market org? It's really small. Like if I went on LinkedIn, like what would it tell me?
[0:53:54] I think today it's almost 20, but like maybe last month it would have been 10. And this is across, this is in sales or this is sales? Sales and marketing. Sales and marketing. Yeah. Think about the questions that I'm asking you and the gravity of the types of conversation that we're having. Right. And how much revenue you're reportedly doing, which is approaching a billion dollars. It's funny because I sat with the head of sales at OpenAI for dinner. I can't remember how long ago it was, maybe.
[0:54:24] Six, nine months ago. And on her team was like eight people. And I'm like, what? Your rep productivity is the best in the business ever? You're telling me, Jubin, we tripled the organization last year. And that was like redlining. Meanwhile, your entire go-to-market team is like 20 people. Sales is 10 people. Each sales rep.
[0:54:53] is doing like $100 million. Like, not actually, but you know what I mean. And it's not just you. This is a feature of these AI companies right now of an extremely lean go-to-market team relative to the disproportionate size of revenue. And the only explanation I can come up with, actually, you tell me, but it's just so many orders. You're just taking the orders? Like, I just can't understand.
[0:55:21] I think it's indicative of this thing that I'm sort of talking about, which is just that the AI industry is really unusual, that it sort of breaks a lot of business paradigms. And I think what we found is we're very fortunate to be in a position where we have way more demand than we could meet.
[0:55:47] Really, really, really want to use generative AI. They see the potential power and promise of this technology and they want it in their hands. They want access to it in some way. And so it's just interesting because I think a lot of our kind of struggle has been how do we get it out there? What are the fastest ways to sort of get it in the hands of the people that want to use it while also recognizing much of the technology is...
[0:56:13] Pretty technical, right? And so it's not 100% plug and play in all business use cases. But I do think a lot of what you're really kind of gesturing at is people want to use generative AI. There's an incredible demand for it. And that happened sort of seemingly overnight. It just sort of exploded into the world. And now suddenly everyone has a use case for it. And so when you're in the world of that many people want to get their hands on it, it's...
[0:56:41] The case that a pretty lean team can do a lot. That being said, we are also hiring. We definitely could do more if we had more people. It makes sense. It's just what I'm scratching at is the AI industry is very unique for AI people, for tech people, for the people that work there. But if you're the CIO of a Fortune 100 company, they don't care how unique and special you are. Guess what? They want an account manager.
[0:57:11] They want someone that they can call that's the same person. They want all of the normal things. That's right. And they're not going to be like, all right, Daniela, this is the best thing we've seen since sliced bread. And so we're just going to accept your terms. That's just not how the world works today. And maybe these AI companies will bend.
[0:57:34] Large organizations to their will. You're shaking your head. I don't think so. I don't think so either. That's not been our experience. That's not been your experience, which is probably a healthy level of skepticism from these CIOs. That's what I was going to say is I don't think it should be that, right? I think we still have the same customer-oriented responsibility as any business. We have to be available to answer questions and explain how the technology works and be on call if something goes wrong. You are right that that takes people and that takes time.
[0:58:04] To be able to sort of scale that up. I will also just kind of point to there are cloud providers also that we partner with who are selling the technology. That give you reach. Exactly. It's your channel. And as we're scaling up, that's a great stream too. At the end of the day, of course, our team knows the most about our technology. And when you're in this world of having so much demand and many use cases that are pretty scalable and repeatable, there's value in just having.
[0:58:32] Yeah, and maybe you go channel all the way, but the problem with the channel in emerging technologies and new categories is that your own team is barely keeping up with what's happening. The messaging, the product, the tech, everything, let alone putting a hop or two in between. It just becomes muddled.
[0:58:56] That's right. You lose control. I will also say, I think something that's interesting to kind of go back to this organizational scaling question that we were talking about earlier, part of why we felt like we had to grow faster than the sort of operational leader in me wanted was because we were trying to meet that demand. Totally. We looked at the universe and saw, wow.
[0:59:22] There's a lot of people that want to use Claude. That is a great problem to have, but that's also a real problem. And so we had to add engineering. I mean, a year ago, we didn't have any product on the market. A year ago, the only thing Anthropic was known for was doing research. We launched our first product February of 2023. So you can imagine the amount of growing that we had to do to be able to acquire the customers that we've acquired and do the type of...
[0:59:51] Business activities that we do now. And so I think that there was almost this just palpable need to go out there and get the bodies that were required just to do the sort of frontline work that had to be done in order to get Claude to market. And so now we're entering a phase of saying, okay, there's even more demand. We really got to scale up. That leads to some sort of interesting trade-offs for us around how quickly do we grow the team?
[1:00:21] Again, in a perfect world, I would love to not grow quite so fast. But to your point, our sales team, our marketing team, they have to be bigger. We have to grow. We have to add more people to be able to do the type of customer conversations that you just talked about. 100%. It's very much like, in some ways, such a classic story of a tech company that's going from a product.
[1:00:45] to a customer company. And I feel like we're in the midst of that transition in AI today, where it's almost a more extreme version of that expression, where it's a research lab that then becomes a product that then goes into like, oh, we're going to commercialize this, that then has a multi-pronged distribution of both B2C and B2B. The B2C is fueling the zeitgeist.
[1:01:13] Because now my mom's asking me about OpenAI or whatever, right? And using these tools, then that's feeding the things that she's doing at work. Then all of these individuals, kind of like a classic PLG company, are then going to the CIO and being like, we're using this. The CIO is then like, wait a second, how do I reconcile half my organization using this and putting our company private data into OpenAI?
[1:01:42] And then what the PLG companies for a while were saying was like, oh, it's just product. This is how it's going to work at organizations. You're going to have a lot of people yelling at you and sorry, you're just going to have to use it. And it's like, actually, the CIO can just put a kibosh on this whole thing. And what you end up then realizing is, well, maybe I need salespeople to quell the concerns of the executives with the dollars and help.
[1:02:12] Those executives navigate their own organization because they're really complex and complicated and go through procurement. And that is like maybe eight people on one account. I really like the way that you put that, that AI is sort of in this process of transitioning from being kind of a product led industry to being a business led industry. And I think the way that we often.
[1:02:41] The way that that shows up is we really want to be able to help businesses understand how to get the most out of this technology. And there's a lot of things you can do, like explain it and write docs and things like that. But to tie it back to this idea of AI is still best with humans in the loop, with humans in concert with them. There's a way that we are still the best at explaining how to get the most out of the technology.
[1:03:09] The AI technology is capable of doing so many things, and it still sort of needs a human interpreter to help people understand the vast myriad of use cases that could be right for them in their business. And so I just think it's an interesting kind of question that you raise around sort of where the industry is going. And I do think there will continue to be more research and product innovation that will come.
[1:03:32] Out naturally because it's such an incredible space of so much innovation happening. And I think you're right that there's kind of this third leg of what is kind of the business story and where is that going that is emerging more now. Yeah. And it's being masked. I think this is a crazy shortcoming of most of these AI companies that's being masked by.
[1:03:57] Seemingly infinite demand, which by the way, over the last few years is exactly what happened with, and I'm putting in air quotes, all these product led growth companies where they had so much inbound that they never felt any urgency to actually create an outbound customer oriented organization. And they have product and technical oriented CEOs that aren't oriented that way. And so now.
[1:04:23] Many of those companies are getting their ass handed to them because they don't have the DNA of serving customers. They resemble more of what you do and these AI companies do, which is like research companies, tech companies, product companies. Do you agree with that? I agree. And I think in an interesting way, the sort of fun and opportunity and challenge right now is to go back to this kind of story of like, what is my job?
[1:04:53] For the whole business side of Anthropic, how do we actually turn this into a thing that is not just, like you said, sort of the cool shiny product, right? It's not just a demo. It's something that's providing real business value for organizations today. And I think that that frame is actually one that our engineering and research teams have internalized.
[1:05:19] I think that there is a lot of discussion at a place like Anthropic around not just what can these models do. Sometimes it's really cool to think, wow, you know, they can do math problems. You're like, that's great. Is that going to help the businesses that are- Right. You're still describing here's what we do, not here's what we solve. Exactly. And I think we talk a lot about what are the customer solutions that we're selling? Believe me.
[1:05:43] No one is immune from looking at these things and saying, my God, they can do so much. What can't it do? And I think that inspiration and, again, sort of childlike sense of wonder at the potential for this technology. I mean, I'll go back in time a little and just sort of remember, I worked closely with a team that did sort of early language model stuff at OpenAI. And GPT-2, we were all surprised to learn new French.
[1:06:11] We're like, we didn't train it to know French. How did it know French? There was some French in its training data. And so it knew French. And I think that moments like that continue to happen in this technological development and in the sort of arc of story of the story of generative artificial intelligence. There are ways in which you don't want to be overly prescriptive and just say, well, this is what the models can do today. Go sell me a solution. Because the technology itself is still evolving and it's still getting better at things.
[1:06:37] But it's a little bit of a dance because you also don't just want to say it can do anything, sign on the dotted line. That's not really a business model. You have to be able to tell your customers, here's what the technology is going to do for you. Here are the problems it's going to solve. Here's the pain it's going to take away from you and the organization. And I really do believe that that is going to be the story that AI companies are going to have to tell this year. I completely agree. And by the way, I think you saw this at Stripe. I think you saw this transition where...
[1:07:06] It was probably pretty similar. And they were this incredible, they are this incredible technology that can do so much. And they have had so much demand for so long. And then they flipped the switch and they became obsessive about creating a systematic engine that supports this technology.
[1:07:36] That's right. I agree with all of that. And I will also say something that I think made that possible at a place like Stripe was that
[1:08:00] Patrick and John and the whole executive team, they were already obsessed with a customer experience. And I think if there's sort of one thing that Anthropic, the business side, right, not the researchers, that they are the thing that sort of makes me feel excited about the next year, although there's many things, is they have that customer obsession. They're on the phone with businesses saying, great, what didn't work for you? We built this solution for you. Why didn't it work?
[1:08:29] And I think there's a way that companies, when there are sort of a lot of markers of, oh, things are going well, you can lose that hunger. You can lose that customer obsession and that orientation towards, are we really meeting your needs? Or are we sort of telling ourselves that we're meeting your needs? And I think that is really the thing that I most want to make sure that Anthropic does right, is keep that customer obsession, especially as we're building something that is so new, right? Where we're going...
[1:08:58] Every some number of months or years say, guess what? There's new model features. Now you've been using it to solve this problem. You can also now use it to solve this problem. But I do think that that kind of business customer oriented component of what we do is equally important to the technological innovations. Yeah, I can't wait to see how it shakes out. Because if this ends up being something where the B to C,
[1:09:27] is so powerful that it's Google-esque, then this doesn't matter. Then this is kind of a moot conversation. But like, there's been one of those companies, you know? And if it turns out that this is going to resemble basically every other business that ever has in the world, then what history would tell me, and maybe I'm wrong, is that the terminal value of this company will be determined by the quality and strength of distribution.
[1:09:56] And I'm going through talking to companies like I just spent time with the MongoDB CEO and with Frank Slootman. These are people that are very distribution oriented. Very. Mongo has no business being this big of a company, especially with the competition that they have. It's insane. But their execution was so remarkable.
[1:10:24] That they are now a giant company. Anyway, you're like basically valued at their valuation now, so what do I know? But you see what I'm saying, right? I do. And I think, again, I think this is one of these questions around what are the ways that the AI industry is different and what are the ways that it's the same? Every company and really sort of every...
[1:10:52] Industry within the technology industry that sort of has its moment that AI is having right now. It's really easy to tell yourself stories and ways about how you're different. And I think there's a huge benefit to keeping grounded and saying, okay, parts of the technology might be different. And this might really transform the ways in which people work and live and what we spend our time doing.
[1:11:20] And there's also fundamental things about the business that are probably not going to look that different from any other business that's existed in this space. So I like that you are kind of bringing it back to this idea of what are the ways that it's different and what are the ways that it's the same. And I remember Claire would relate stories to me about how the Collison brothers would try and reinvent all of these things, like OKRs and performance management.
[1:11:47] You're laughing because you were probably in the mix on this. I was. I remember those days. Yes. And she would be like, guys, you can do this. You are literally smart enough to do this. You could reinvent every business process if you wanted. But is that what we want to do? Yeah. The way that I like to think of it is every tech company in Silicon Valley is innovative. Like we're all innovators if we're here. Where do you want to use your innovation chips? That's something I think about a lot.
[1:12:16] Companies that try to do everything differently, everything their own way, in general, they're not successful because they're not focused. We're like, okay, are you innovating in how you do internal performance management? Are you innovating by using different tools than everyone else? How you set up a C-corp? How you set up a board? Right. But that's what I'm saying is you kind of have to pick what are the ways that you are innovating? And what are the ways that you're sort of...
[1:12:43] And I think our sort of zone of innovation is around the technology and the social mission. We're saying, hey, we're building models that exist almost nowhere else in the world. That's a big bet. That's a wacky innovation chip that we're using. And we're also going to do that while remaining.
[1:13:09] We're steadfastly committed to ensuring that we're doing it in a way that's beneficial and safe. And so are we going to really try and be wacky innovators in terms of how we do performance reviews? Probably going to shock you, Jubin. We're not. We're a little innovative, right? Instead of doing quarterly planning, we do it three times a year because that works better for the research cadence. Do we do OKRs? Yes. Do we do performance reviews? Yeah. I remember when I joined Stripe, this is a funny anecdote, but...
[1:13:38] Stripe like didn't believe in managers when I was there. That changed quickly, but- Like until Claire got hired. Until Claire arrived, basically. Yeah. Around when Claire started. Yeah. When Claire started. Which was like 150 employees. Yeah. And Stripe is an amazing company. One of the best we've ever seen. Amazing company. And so I think it sort of speaks to this idea that you kind of have to pick where you want to do something differently. Because any company that's doing everything the same, they're also probably not going to be successful.
[1:14:08] And so I think a lot of the magic lies in where do you want to be a little bit out there? Where do you want to sort of follow your version of the standard thing that makes companies successful? That's not to say that you can't do the normal things with your own flavor. It's like feedback at Anthropic feels different than feedback at OpenAI, than feedback at Stripe. We have our own color to it. It feels a different way. But in terms of the structure and the tactics, not that different.
[1:14:38] This is what I get scared of in general of being in the bubble. Because, again, if you look back historically, what ends up happening is these tech companies have breakthrough innovations. The first people to consume these products are other tech companies that are built on shaky ground, relatively speaking, other tech startups. They're all sloshing around a bunch of venture money.
[1:15:08] Buying each other's products and tools, which isn't really giving you actual real life feedback, because then you realize most of the money, most of the budget, most of the dollars are not in the Bay Area. They're in the Fortune 500 or in the Global 2000, which is not the Bay. It's everywhere else in the world. You have 10 sales reps that are all in the Bay today, probably, I'm just guessing. And you become distant from the reality of what customers
[1:15:37] Want. And over time, these worlds converge. And I think maybe the only thing that I'm just trying to impress into this conversation is just like, I think how quickly that convergence happens will be a competitive advantage for an AI company, for some of these companies right now, just like it has been for all the wave of companies before then. And I talk to these CIOs all the time.
[1:16:04] It's a huge part of my job. And they think it's kind of a joke. They think that all of these companies are really immature and they are doing an inordinate amount of pushing back, not leaning in. They're pushing back and they're getting pressure from their boards and their CEOs to canvas every part of the organization. And the question that everyone's asking them is, where can we use AI?
[1:16:34] Which is fundamentally a crazy question. I also think there's sort of something you're getting at here, which is sort of a bold thing for me to say as someone whose business is selling into other businesses and particularly into enterprise. I don't think every business should be using AI today. I think there's some use cases that make a ton of sense, right? If you're searching across a lot of information and you need a high accuracy result from that.
[1:17:01] You might want to consider using generative AI. It's pretty good at that. If you are writing code, potentially it could be a great partner for you. There's many other use cases I could go into, but I think there's sort of this idea, and this is kind of what I mean. I keep coming back to this idea of how much are you buying the hype? How grounded are you in the reality of what's actually happening? And sometimes in business conversations, we tell a potential customer, we don't think we're right for you.
[1:17:29] I mean, why do we say that? Because we don't actually think we can provide value to them today. And so we'll say, hey, come back to us. This is what's on our roadmap. Come talk to us in six months or a year. But I think this idea that generative AI can solve every problem, that feels like a story that's older than the hills. Every new technology can solve every problem. I don't think that's true. I think there are a lot of use cases today at a lot of businesses where generative AI can save time, can save money, can save staff, but it's not universal.
[1:17:59] And if anything, I think this idea that these sort of bigger businesses are kind of being pressured to adopt it, that doesn't actually feel right to me. And that doesn't feel sort of ingenuous to the values that I think we should be espousing, right? Which is we should be providing value for our customers. Yep, I totally agree. And by the way, we've seen this parallel with the public cloud providers. When the public cloud providers came out, they were everything to everyone.
[1:18:26] And over time, what ended up happening was they were Google built TensorFlow and they started having very specific use cases that solve very specific problems. And they would use that as a foothold to then expand into an organization. And if they didn't do it, then someone built on top of Google was going to do it or built on top of Amazon. And it was going to be a very specific purpose built tool, like today, a legal summarization tool.
[1:18:55] That is going to go mop that market up because they speak legal. They go directly to the chief legal officer in conjunction with the CIO. They tune the model directly to that use case. And that's all they're doing is talking to that customer. And I think we're starting to see that all over the place, just like with code. We're just starting to see these use cases picked off. It seems to me, again, this is just from an outsider looking in, that the foundational models.
[1:19:23] Are still everything to everyone. And that's okay when growth is historic. Like literally we're talking about like going from zero to a billion dollars in a year and a year and a half. So like doesn't matter right now. Everyone's like, yeah, venture capitalists, yep. The big cloud providers that are putting it in, yep. The teams, yep. It creates a little bit of like, Jim, what are you talking about? I digress, but that's just my observation. Do you disagree?
[1:19:50] I think you're getting at something really real, which is that being everything to everyone is not actually a business solution. And I think that you can simultaneously feel incredibly excited and in awe of what the technology can do today and say, which I say often, I think this technology is still in its infancy. I think that we're still in very early innings in terms of figuring out what problems it's best positioned to solve.
[1:20:19] This whole kind of phase of generative AI development is not that old. I just sort of keep coming back to this feeling that the distance ahead of us is vastly greater than the distance behind us. And there's a lot of excitement right now. And excitement is a great place to start. But that's not a finish line. And so I think from where I sit, so much of my...
[1:20:40] Energy is spent on thinking about the future, right? Where are we going? What can this technology do in several years from a research perspective, certainly, but also how do you take that and translate it into something of value to people today? Yep. And it seems like maybe this is where the crypto industry hasn't completely gotten to yet, which is why it hasn't been adopted in the way that people thought it might have. And also why in the early days of the public cloud, it was only financial services firms using it and they were using it for disaster simulation tests.
[1:21:10] Where they needed to burst a bunch of compute that was unpredictable. Burst it up, burst it down, perfect use case. They sold that. And then once they sold that, they built on top of that. CIOs at the time really just wanted to be told, you should use AWS or GCP for this use case right now. And I think in a lot of ways, CIOs today want to be told, these 70 use cases that your board wants to see should be like three.
[1:21:39] And here's the three. That's right. And this is how we're well suited to solve that. That's a lot of the conversations that we have today. Yeah. It's like, here you go. Here's the solution. The 10 people. But that I do think there are ways it's adding value now, but that's not limitless, right? That's what I mean about sometimes it's like not the right use case for generative AI, right? And I'm like, you have to go find a specialist to do that. So I definitely hear where you're coming from. I think it's an incredibly exciting place to be. I think that having a technology with so much demand,
[1:22:09] And then you're in the job to build the scaffolding around that to build a real business that serves customers because that's what businesses do. And that's your job. Literally, that's what you do at one of the kingpins. And that's very exciting. Thank you. In the high growth handbook, Claire has a section of working with me. You probably know that section well because you've probably had the firsthand version of it.
[1:22:38] What would that read for you? It's actually really funny that you asked this because I think around 10 or 15 people at Anthropic, we all wrote working with me guides and we still have them. And what is, can you explain what that is? Yeah. A working with me guide is like an operating manual for you, the person. And so it's actually even meta interesting because people will fill them out differently. Like there's no template. And so.
[1:23:07] Very kind of engineering brained people will generally have like a clear structure and lots of frameworks and more intuitive people. It'll just sort of be like a list of words, right? It looks really different. But this idea is basically we should demystify for each other in advance. Like what am I like to work with? What are ways to get information out of me in a way that's really going to help you? What are things that are going to trigger me or make me not be my best self around you?
[1:23:35] And so we all wrote those for us at Anthropic around like 10 or 15 people. What's the low light and highlight for you? What's the thing that triggers you? And what's the thing that like revs you up? So I think the highlights for me and kind of a lot of what I wrote about in that document is I think in my heart, the thing I most am is a people leader. And what I mean by that is I have managed basically every function in a business across
[1:24:04] Stripe, OpenAI, and Anthropic. I've run sales teams and recruiting and people, and I've managed research teams and engineering and product. I've done everything. And the way that I have been able to do that, to varying levels of success, of course, is I'm actually not an expert in any field. I'm an expert in people. I'm an expert in figuring out
[1:24:33] What is the problem and who are the people that are going to solve that problem? The thing that tends to light me up is a new area or an existing one that has a real challenge and figuring out who are going to be the people to work on it. How do we get them organized and structured in such a way that they know what they're doing and they can move clearly towards a goal? And I love doing that type of work. That's basically what I do at Anthropic all day. I think about who are the senior leaders?
[1:25:02] We're hiring. What are they going to work on? Are they the best person in the world for that job at this stage of company? And I get to do that. I mean, it's a privilege to get to do that across so many different teams at a place like Anthropic. In terms of sort of low lights or what I don't like to do is I'm not the expert in anything. There's always a person in the world who is a better focus leader than I am.
[1:25:32] If I'm temporarily in charge of a particular function and there's no leader for it, my number one goal is getting myself out of that job by finding the right person to be leading that function. And when you think about interviewing, I do a lot of interviewing in my role today. I bet you do. And I probably do less interviewing than you do, but it's still obvious to me that it is very f***ed up. The way that interviews are done is so broken. Do you agree with that?
[1:26:01] Yes. I mean, I think we do interviews pretty well at Anthropic, but yes, interviews in general are not great. Why do you think you do interviews well at Anthropic? Well, I have to start with a disclaimer here, which is that I think interviews are important and have signal and they are one piece of a bigger picture when you are assessing talent. So the number one.
[1:26:28] I would sort of say to people that are in high interview stages of their companies are you have to be really clear about exactly what you're looking for. And I know that that doesn't sound earth shattering, but I actually think it's a step that people skip way too often.
[1:26:50] Roles that you're hiring for for the first time or they're very significant roles, right? Either because of their seniority or a piece of the business really relies on them. You have to invest a lot of time upfront in thinking about what do I actually want? But then you also have to adjust as you talk to people and sort of learn the shape of the role. You almost never want to start your interview process with the person who you think you should hire because you're probably going to change your mind about what you want at least two or three times.
[1:27:18] Once you go through the process. But the more you've kind of written down in advance, I really need a builder to grow the sales organization. I'm just making this up. I don't need an amazing sales leader. I need someone that's going to go higher in scale. That's going to look really different in terms of what you're actually assessing for in the interview process than I need an exceptional individual sales leader who's going to lead from the front. Those are just completely different people and completely different profiles. And often the interview team,
[1:27:47] Has totally different pictures in their head of what they're actually looking for. And a job description is a great place to start, but you probably also want something internally that's like, this is really what we need for this role today. This is what we need for this stage of company. This is why we need this person. And then I think it takes a lot of time. It is a lot of work to source the right people and interview them. But outside of just writing down what you need and interviewing them, another step I think people miss,
[1:28:18] A lot is references are just wildly signalful. And I don't just mean, hey, Jubin, can you give me three references? I'm often like, hey, I'd like someone from this job. I saw that you left this company after only a year. I loved your story about why. Can I talk to someone that you worked with there? And people's reactions to that are often really interesting, right? Sometimes more information comes out that's useful in assessing them. But especially for people that I'm going to work really closely with and for senior leaders in a company,
[1:28:47] I think the best relationships stem from mutual understanding
[1:29:13] about what you're both signing up for and clarity about what the job is. And as much as possible, music to my ears that we've done recruiting well is someone comes in and they say, this is exactly the job that I expected, or this is pretty close to what I was expecting. And so in some ways, I think interviewing is a little bit of a two-way street. The more information you're willing to give, the better people are going to self-select into or out of that position. And then also you have to do your homework and it's an incredible amount of work.
[1:29:42] And you have to really think critically about what it is you want in that function. I think that's very well said. I've heard, not dissimilar to Claire, that you are very serious about reviews. True? Very, very serious about performance. Yes. Tell me, what does that mean? So feedback is a word that does not sound very sexy. And I recognize that. So I'm going into this conversation.
[1:30:11] With full self-awareness of that. Managers do so many things. And one of the things I try to do at Anthropic is really demystify for managers, especially new managers. What does management even mean, right? Like what is your job? But I think that feedback is one of the core tenants of not just management, but being a good employee is knowing how to give feedback well, respectfully, and thoughtfully.
[1:30:40] We take feedback very seriously at Anthropic. And what I mean by that is, I'll give you an anecdote. When we were maybe 15 people, it was like May of 2021. I said, hey, we've all been working together for some time now in very close quarters. And we are all going to give each other feedback. And I made this spreadsheet. It's probably still somewhere in my Google Drive if I were to dig it up. You got to send it to me. I will. I'll find it and I'll send it to you.
[1:31:10] On the left-hand side is just everybody's name. And then across the top is everybody's name. And I said, everyone in the company is going to go and give feedback to everyone else in the company. And you have to write it down. It doesn't have to be fancy, but you have to say at least one thing that you really appreciate about what they're doing and how you think they're having impact. And one way that you think they could increase their impact or something that they could be doing better.
[1:31:37] And you have to practice it with someone else before you deliver it. And then I made everyone go do that. And they had to mark in the spreadsheet when they had done it. And I went and I asked each person, how did the feedback conversation go? Meaning the spreadsheet was like an ex-I talk to Daniela.
[1:31:54] Yes, Amanda talked to Daniela and Amanda talked to Catherine. So the feedback's not in the spreadsheet. No, the feedback's not in the spreadsheet. It's just an accountability system. It's an accountability system. But I was like, hey, you put an X here. How was the conversation? Like, how did it go? And because our team is amazing, they all were like, that sounds like a great idea. Let's do that. And what I saw was pretty soon afterwards, you saw behaviors change.
[1:32:22] Not in big ways, but in sort of subtle ways. And there was even sometimes a little bit of language change. Someone in a presentation was saying, hey, I got feedback that I don't communicate out enough. And so in addition to this presentation, I'm starting a feed channel that's just going to tell people what my priorities for the week are. And I heard someone else say, oh, I got feedback that people really like my updates. So I'm going to send them twice as often. Or I got feedback that when I'm arguing.
[1:32:49] An idea I come across as overly intense, so I'm going to try and consciously argue the other side. And it sounds so small, but if you imagine we're 15 people and we set this norm, this cultural expectation that you are honest with the people that sit across from you. And you're not a jerk about it. Sometimes I've had to say, hey, I hear that feedback. Can we like...
[1:33:14] Make it a little nicer. But in general, I would prefer to be in an organization where people are trying to be honest with each other and trying to give feedback and coming from a place of wanting the best for the other person. And maybe they get it a little bit wrong than a place where people are not communicating feedback to others. And I think in the technology industry in particular.
[1:33:38] We're much, much more often in a world of people not sharing feedback as often as they should be. Because it's mostly introverted technologists. Right. And it's hard to talk about. It's hard to say, hey, you're doing this thing and it's making my job harder. And so I think just kind of instilling this idea that conversations like that are totally normal and totally expected. They're part of your job. They're in our matrix of expectations for people. In everybody's performance review, we're like, how did you do at...
[1:34:06] Culture and feedback and communication. That's the thing we talk about with every employee. And we do performance reviews every six months. Everyone in the company who's been there for more than two months gets a review. And that includes me and Dario. We get reviews from the board. Everyone gets feedback. They write a self-review. They get peer feedback. They get feedback from their manager. And if they're a manager, they get upward feedback. And it's kind of a bold thing to do at a startup.
[1:34:35] That was not what it looked like at either Stripe or OpenAI around that size of company. But I think it served us really well. It's helped us to identify when there's a performance issue early and handle it. It's also helped us notice when like something, it stopped a lot of things from going off the rails because we just talk about it. And there's a lot of conversations I have around.
[1:34:57] Hey, I don't think you meant it this way, but this didn't make me feel good. That's either going on under the surface or it's going on and you're talking about it. So my feeling is feedback, management, not sexy terms, critically important if you're trying to be a wildly successful company. Do you tie compensation to the performance reviews? So right now we're at a stage where we're just trying to get people to be honest with feedback. And so everyone at the same level makes the same. Over time, that might change.
[1:35:26] Because I think it's smart, by the way, because it does create perverse incentives. It does. When you start tying, because then you can just game it. Right. People want to game it. Exactly. And people don't want to negatively impact someone else's comp. Right. And that's exactly what leads to people not being honest with each other. They're like, oh, well, I don't like this about this person, but it's not that case for us right now. What else do you do? Anything else that you are particularly proud of that you think is unique or different? So I think from a management,
[1:35:56] I don't know if this is just something I think that's unusual or something I maybe a little bit of both. I believe that middle managers are one of the most important groups in a company growing at the rate that Anthropic is growing at. And I think they're actually often ignored or not given enough attention. And they're one of the most important.
[1:36:25] And to unpack that a little bit, I personally probably spend the majority of my time with senior managers, with my direct reports who are the leadership team of the company. But line managers generally know the most about what's actually going on. They're on the ground. They have people working for them who are doing the real day-to-day work. And they actually hold an incredible amount of useful information.
[1:36:55] ICs do too, right? ICs are the best, but from a scaling perspective, like you can't talk to every 320 person company. You can't talk to everyone at the company. But middle managers, they carry an insane amount of water for the company. They're hearing direction from way up here and they're trying their best to like do what the company is telling them. They're usually shotgun promoted. Right. They're often like, I just wanted to be an engineer, but it seems like you all needed a manager. I'm here for it. And there's almost this way that,
[1:37:25] They kind of just get thrown into the job. And if their manager's good, they're paying attention to them, they're supporting them. But there's not like a structured way for them to learn from each other and from the business more widely. What is my job and how do I know if I'm doing it well? And so something I did for a while, and now we've made a bigger version of it, is whenever there was a new class of people that became managers, I would just meet with them like once a week.
[1:37:54] And I would be like, what's on your mind, guys? And so it was people who were across the organization. Informal. Yeah, informal, but in a little group with them, four or five. So they got to bond and be friends too. Because a lot of their questions, whether they were a new manager in research or in product or in sales or in recruiting, they were like, how do I know if my team is doing well? Or I have a really tough feedback conversation and I don't know how to do it.
[1:38:22] It was, I hope, useful for those people. But I have to tell you, Jubin, those conversations, I swear I learned more from them than those people learned from me. Oh, for sure. Because they're telling me what challenges they're facing. And sometimes it's about a person, but it's often about something that's a system that's not working in the company or a piece of information that there's no other way I could have gotten it. I could have talked to my leadership team 50 hours a week and not gotten this gem that somebody at
[1:38:50] The frontline manager level can tell me in 10 seconds. Now that we're a lot bigger, that doesn't scale. But I do a manager round table where we talk about management themes and it's primarily for line managers. Folks that are managing ICs, sometimes managing managers, but rarely. And again, I predict I learn more from them than they learn from me. I think it's a really smart way for you to continue.
[1:39:14] To cultivate ground source truth about the organization. Exactly. I think a lot of the way that things can go wrong is that leadership just gets really divorced from the day to day. And there's different ways that people keep in touch, right? Some people use their product only at their company, right? They don't use any other product because they want to make sure it's good. Some people...
[1:39:35] Whatever, they dive into the code base themselves. And I think my sort of version of this as the people leader of the organization is I have to know what's happening on the ground for the people that are managing the people that do the work. Totally. Do you have a favorite interview question or one that you lean on? One that you ask most everybody? This is a little bit of a sneaky one, but I'm going to give it to you. So when I first meet someone, so first interview contact, I always ask the same question, which is tell me about yourself.
[1:40:05] You would be shocked at how differently people answer that question and how much information it reveals about them. Not necessarily good or bad, but when you give someone an opening to say, who are you? People choose to answer that in really different ways. Some people start telling me about where they grew up and their family and who they are and what their values are. Some people just laundry list career accomplishments. They're like, here's what I did.
[1:40:34] Some people talk about their teams or what matters to them, their values, where they wanna be in 10 years. Sometimes asking people who they are is the best way to learn who someone is. And when you have to work with someone day in and day out, you probably wanna know honestly who they are. You'll also notice a range of how long people talk and how often they check in with you. Some people will say, oh, that's a big question. Do you want me to talk for five minutes or 10? Some people will just talk for half an hour without checking in with you about,
[1:41:04] Is this actually what you wanted? And it's not the be all end all. It's just the beginning of the conversation. But I find it is really interesting sort of litmus test for learning about someone. The second is I always ask people I will be working very closely with. Tell me about best day, Jubin. Tell me about you on a great day. What are you like? Then I ask.
[1:41:27] Tell me about you on a bad day, like the worst day at work. You're super frustrated. Everything's not going your way. You're down. How do you show up? And the reason I ask those questions are number one, self-awareness. I can tell you how I would answer those questions. And like every reference that you talk to will probably say some version of it. They'll be like, yep, this is Daniella on a good day. This is what she's like on a bad day. The second is if you're stumped by that question, probably you don't know enough about yourself.
[1:41:54] And there's some work that needs to be done before I'm thinking about you in a senior position at a company. That's so good. Thanks. When I was doing back channels on you, somebody who knows Sheryl Sandberg personally was like, I have never seen anyone reincarnate Sheryl. That's describing you. That is incredibly nice. To whoever said that, let me know so I can send them a bottle of wine. Specifically, the quote was that she's a unique intersection of EQ and IQ.
[1:42:23] That I have not seen in anybody since Cheryl. And they're like childhood friends. So I can't wait to see what you all do. What a time. So thank you for doing this. I conclude all these things the same. The first, are you hiring? Given our conversation, are you going to triple again? I mean, we are hiring. Are you hiring for go-to-market? Yes, we are hiring for go-to-market. We're hiring a lot of people. And product and engineering and...
[1:42:52] Basically every function in the company, our website is basically mostly jobs that are available. So yes, we are hiring. If there is one characteristic that you hope that every Anthropic hire has, what is it? I can only pick one. Well, fine. Maybe more than one. Curiosity, humility, grit. And when you hear the word grit, what do you think of?
[1:43:17] I love that you asked this question to everyone. It's so good. And I love hearing what different people say. So I actually have two answers. I hope that's okay. Yes. My first answer is if you really want to see grit, the place you should look is professional athletes. I love that you had Alex Smith and Sean Livingston on the show. Some of that is because I'm from San Francisco. And so of course, Warriors and 49ers fandom runs deep.
[1:43:46] But I think there's something amazing about the way that professional athletes train and perform. I've been very fortunate to get to work with incredible tech tycoons. And I think what professional athletes do every day is just wildly more impressive because they actually show up and they put themselves physically on the line at peak performance at all times.
[1:44:12] So my real answer is if you want to know what grit is, look at a professional athlete. But my second answer is there's something I think we can learn from that approach in any industry that you work in. And what I mean by that is athletes are so consistent. They show up every day, whether it's raining or sunny, if they're feeling good, if they're feeling bad.
[1:44:39] If their team is in first place or last place, they come to practice no matter what. And they're consistent. They are focused. They're thinking about how to be the best versions of themselves in that moment, not just for them, but for their team. And I think if we can manage to do that, whenever I see people that are excelling, they're doing some version of that in their industry. They're looking around and saying, okay,
[1:45:08] How do I show up whether it's a good day and we're being hyped or it's a bad day and we're not? And I deliver the same experience to customers. I act the same way towards my colleagues. I'm respectful. I'm kind. I'm hardworking. And I focus on my impact. I'm like, what did I do today? What changed because I was here? And I think that kind of consistency and groundedness is what grit most means to me. What a B2B answer. Thank you.
[1:45:37] Thank you. Thank you for answering direct questions directly. You're welcome. I appreciate that. Thank you. Thank you, Jubin. That's it. Thanks for tuning in. Feel free to come back every Monday morning to listen to a new guest or go back into the archives when we've done more than 100 episodes. This podcast is a Kleiner Perkins production and the episode was edited by Eric Johnson from Lightning Pod. Thank you all.