← Dialectic
Dialectic
48: Henri Stern - Principled Enough to Be Pragmatic
01:54:50
Founder of Privy talks Privacy, Crypto, Stablecoins
Loading summary
Transcript238 lines
- [00:00]
A
The sheer arrogance of Silicon Valley in thinking that you can just do things to the French mind is so absolutely incredible. And this is why Silicon Valley is what it is. And there's so much to be excited about that we think we can reshape the world. And we can, in fact, because we believe we can. We wrote a strategy paper early on which basically said there are three markets. There's the extremely die hard crypto natives for whom storing an mnemonic is a non negotiable fact of self sovereignty and we will never be able to convince them. There is crypto developers who want to build for a mainstream audience, and then there are mainstream developers who want to use crypto. And basically we said we will never win one and three doesn't exist. And so we have to go for two. We have to build for basically crypto devs who want to cater to a wider population that will teach us the things that we need so that when three comes about, we'll be ready. And that is like, we just got massively lucky with timing that three is coming. Now three is here. Most of our biggest customers aren't here because they care about crypto. They're here because they care about the impact that this technology can have on their users. Pivot was probably the hardest moment in the company's life where we were doing an offsite and just the atmosphere sucked and it was like six or seven of us, the entire team, and at some, some point I think I said so like, clearly things aren't working right and like you felt the body language change and people relax. We have eyes, of course it's not working, we all see it. But now it's like part of the conversation and we can do something about it. The conversation that I think had a real impact on me a little bit was I went to see Roelof from Sequoia and I asked him what would it take for Sequoia to be excited to do ra. He was like, you're talking to me about how far you've come and I don't give a shit. The only thing I care about is am I excited, looking forward. And I think that was like such a wake up moment. The obvious searing pain of trying to create something from nothing is so wild in your mind that you end up saying, oh, but we've done so well.
- [01:40]
B
Welcome to Dialectic Episode 48 with Henry Stern. Henry is the founder of Privy, a crypto infrastructure company that allows every developer to put a crypto wallet in the hands of its users. In the middle of 2025. Privy was acquired by Stripe but remains a independent organization within Stripe, and Henry works across Privy and a number of Stripes Other Crypto Efforts I got to know Henry when I was a venture investor at the crypto firm Paradigm, which shortly after I left ended up investing in Privy. Henry is one of the most thoughtful and principled people I know, and I deeply admire the way he is able to combine ideology and pragmatism to build in a space like crypto that is laden with incredible amount of both speed and velocity and change, oftentimes dogma and ideology that ends up going too far and people trying to push things far beyond they should or exploit things. And I think the way he has been able to remain focused to serve developers and customers and build a product and a team and company that he can be really proud of is deeply admirable. We talked extensively about morality and technology, privacy, crypto wallets and where all of this was and where it's going, the thought process and the decision to join Stripe and what it's been like since he joined and where crypto is going, even if it might be a little boring at times. I even managed to make Henry play philosopher with us despite his best efforts. I hope you enjoy the conversation with Henry and before we get into things, I would like to thank Notion Dialectics Presenting Partner Notion is a collaborative workspace for your life's work that has natively integrated AI and agents into the place where you and your team do the actual work. Over the course of the last year or so, Notion has totally transformed itself into an AI native company that is first and foremost seen with custom agents, which allow you to have little guys in your workspace that can automate everything you don't need to do or don't want to do and allow you and your team to focus on the important work that you don't want to delegate. The biggest news in Notion's world lately is Notion's developer platform and workers, where you can design a wide range of ways that agents can better help you and your team focus on the actual work while they take care of the busy work. The range of what's possible with Notion Developer platform is pretty incredible and still very early days, but I just saw something from Brian Lovin about how you can basically get your entire Twitter graph downloaded into a Notion database categorized updating constantly. I'm really excited to play around with this more, both in terms of how I present dialectic episodes to you all, but also the ways that I can hopefully continue to push the bar on how I research for these conversations. It's a delicate balance between trying to find a way to manually sink my teeth into all of my guests ideas and also find unexpected places and unexpected things that I might not have found otherwise in the research process. Thanks again to Notion and you can also learn more@notion.com dialectic and with that, here is my conversation with Henry Stern. Henry Stern, thank you for having me here at the Privy Office.
- [04:57]
A
With pleasure. Thank you for, for being here.
- [04:58]
B
We've got all the secrets on the whiteboard.
- [05:00]
A
Let's see.
- [05:00]
B
We.
- [05:01]
A
These are, these are, this is, you can see this is hard hitting work we're doing here.
- [05:03]
B
Yes, well, if we have to explain anything we, we have the whiteboard available to us. I want to start with first of all, thank you for doing this. This is an honor. I've known you for a long time and I've been rooting for you for a long time and you keep winning. So
- [05:20]
A
yeah, it takes a village, as you know, but we have a long way to go.
- [05:25]
B
Indeed. At the preview website at the very bottom, the bottom right you have a phrase, it says technical decisions are moral decisions. Why is that important to you?
- [05:43]
A
Sorry, I'm thinking. I mean the reality basically is like as software eats the world, the question of what is the part of moral valence that software builders are accountable for just like grows. I think when we set out with ASTA to build Privy, we were building something that had to do with ownership and with privacy and with how the web works. I want to start with the premise that we are fallible and we will make mistakes and the best thing we can promise is we care so deeply and when we make the mistakes we will react accordingly. And that's what it stood for. And I was like we should actually just have this on the site and just have it be something that we remember because whenever we start larping as we know truth and if you don't agree with us then you're a heathen or conversely some nihilistic converse. But they kind of meet of like it's all about pragmatic distribution and this has no. Stop overthinking it. I think that the truth is more nuanced and so I think starting with a premise of this is going to be complicated but we're in it and we're thinking about it and we're accountable to our work. Was, was kind of the bottom line that we wanted to set in starting
- [06:58]
B
the company you mentioned, sort of like close to the metal in some sense. The story of computers is like abstraction. Yeah, like getting farther and farther away from the metal. Obviously there are elements of crypto that involve technical rigor and proximity, ideology, all these things. But like it is super. And I want to talk about different elements of this, whether it be privacy or self custody or otherwise. But like what we've sort of seen is that like people prefer convenience over whatever ideological or technical or trust the metal thing at Infinim. And so how do you think about that? Maybe under the lens of that statement?
- [07:39]
A
Crypto is a very interesting foil. And I have many views, but it's a very interesting foil because it is the best and the worst of a lot of like creative endeavor. Then you have a lot of extremely ideological people who are fighting for what they see as a future path for the Internet. And that's like the best version. And at the same time that often veers into such an extreme level of non pragmaticism and belief that actually everyone will adopt these experiences that if you actually just take a step back for a second, are so obviously broken and there's no world in which it happens. And so I think what we had was a few experiences where we would go to talk to a user, we would say, this is how we've built it. We would get absolutely basically destroyed for like, but you guys are building this on aws. How can you trust a centralized cloud? And we're like, everything's encrypted, end to end. The keys are managed by the user. And they're like, yeah, but no, the data won't be available if AWS decides to censor you. That's a really good point. There are solutions. We can come up with it when that becomes the threat model. But also for now, the threat model is more about privacy than censorship. And you would talk yourself hoarse. Nothing would happen. This person would say you were demeaning the space and then you'd go back to this person three months later and they were taking PII and dumping it in a postgres database on their side. And you're like, what happened? And they're like, well, gone for gone. In for a penny, in for a pound. Insofar as I'm going to do this, I might as well just do it easily and I will take on the risk. And I was like, there's no intellectual consistency here, basically. I think a lot of the lesson here was one, let's move away from trying to please everybody. Let's try and figure out what's right for us as users, where you're right A lot of tech is about intermediation, but we want to know that if we scratch, we can figure out how things work. And so that stood out as becoming like, this is the right line for us to have as a company. And if you go to our security website, we have a security faq. Makes me deeply uncomfortable. It's very transparent. It's how Andrew McPherson, who runs our security team, ended up joining the org. He was like, this is a very unusual security faq, but it's very honest in a way that I haven't seen and makes me want to work with you guys. But I think that was the line. I think intermediation is a fact of life. I think privacy is not something people will go out and buy in and of its own right. But I think that doesn't mean that we can't try and build better tooling just the same. And this is why I'm in many ways, and it's obviously a very complex issue, but a big Apple fanboy is like, sometimes things are just nice because they can be. And catering to human qualities is not a good product building thesis. Catering to the seven deadly sins is a much better product. Deadly thesis. And in our case we tried to appeal to privacy and we actually found that appealing to convenience and greed were easier, but baking privacy and sovereignty in was the way that we could do it.
- [10:43]
B
What do you think? Maybe back to that first statement, sorry,
- [10:45]
A
I'm painting greed with a wide brush by saying basically the ability to monetize your products better capitalism is greed under that definition, which I don't find it to be.
- [10:54]
B
Yeah. And you're abstracted probably even a further level in that ultimately you serve developers. Yeah, for sure. I talked to a philosopher late last year ad nauseam, kind of about a bunch of elements of kind of how things have started to fall apart. And one of the big parts of it is that at least in his view, this Silicon Valley or the technology industry writ large has this sort of view that technology is morally neutral and so you should just build technology and it's about what people decide to do with it. Yeah, I guess I'm curious about your reaction to that idea in general and then how you sort of how that sits next to this. Technical decisions are moral decisions which if not, I don't know if it's in complete disagreement with that idea, but it's certainly not in total alignment.
- [11:45]
A
I think it's just a narrow field of view in a sense. Like if you. Again, like software has eaten the world in many ways. Like technology is everywhere. And if you take a wider lens at it, maybe the technology itself is neutral, but the way in which you market it, the people you cater to, you're ultimately trying to spike the ball to tell people how to use it. And that's certainly not neutral. And so I think I broadly, I guess, agree with the idea, which is the same technology can be used for great things or bad things. Nothing is inherently one way or the other. But basically, I think by the time a human, it's like Schrodinger's cat. The technology is the moment you've looked at it, it has moral valence to it. And again, I'm less interested in defining what is moral and what is immoral, and I'm far more interested in people not being hypocritical and owning up to what they're trying to do. That's my main piece, is you should own basically the outcomes that you're trying to drive the world towards. And as someone building tech, then you have a responsibility towards that. I interned at Palantir for a hot minute and absolutely drank the Kool Aid. But during that internship, there was a Privacy and Civil Liberties Board that they would put the interns in contact with to just show them this is how we make morally complex decisions. And at least at the time, the viewpoint was, as an engineer on the team, you need not work on anything you don't want to work on. However the company may choose to work on things that you personally disagree with, please voice it. We will take it into consideration, make a company decision, including this Privacy and Civil Liberties Board, as to whether or not we want to continue doing this. And ultimately the bottom line is there will be someone to do these things. And given a choice, we would rather sit at the table and be a part of building it, rather than let someone else who cares less or do
- [13:40]
B
the Google thing where you're just like, we're not going to. We're not going to bother. We're not going to work with the government at all.
- [13:45]
A
Exactly. So I think that's my point, is you lean in and you own up. But the tech itself, as a cold instrument, I agree, is probably just an instrument, but as soon as a human touches it, that's gone.
- [13:59]
B
Right? And part of the, I think intrinsic to this idea, at least his criticism was if you believe technology is morally neutral, you sort of take this view that we should just. More technology is always good. Maybe put it a different way, you are morally motivated is my sense.
- [14:19]
A
I come out of it thinking action is good and Creation is worthwhile in its own. And having to lie to yourself as to why you do it or what it means for you to do it is the thing that cheapens the whole affair. And so I think broadly that I am the thing that impresses me so much with us. I was having this conversation with my wife yesterday, which is the sheer arrogance of Silicon Valley in thinking that you can just do things to the French mind is so absolutely incredible. And this is why Silicon Valley is what it is. And there's so much to be excited about that we think we can reshape the world. And we can, in fact, because we believe we can. But I think like, you know, it's the, I'm a lot of very mission aligned companies that espouse giant worldviews and then the actions just don't match up to the world. That rubbed me a lot more poorly than say, I want to build the world's best toaster oven. Because I really like toast. I think as long as people speak it the way it is and acknowledge the complexities, I'm good with it. I generally find that more technology is probably better and if nothing else, more technology is inevitable. We are on an accelerationist world path. And so again, I'd rather sit at the table than move to the woods somewhere and try to not participate.
- [15:41]
B
You aren't building a toaster. You are also a French mind. How did I guess the root of this? And I think one of the reasons you're such an interesting person is you are deeply pragmatic and you've been successful. This isn't just this purely like looking ahead, idealistic thing. I, I realize much more, much more ahead. And you're super principled and you, you do care about certain ideological things. And so I guess I'm curious about like maybe twofold, like one, what, what caused you to maybe from previous parts of your life or academia or whatever to want to just do things? And two, like, why this part of the world? Why is this important to you and why does this matter in a way that maybe no offense to toaster makers but like, is a little bit more of substance.
- [16:32]
A
It's a lot of questions. I'll go to the last one. I mean, I think there's so much path dependency. Basically. I, I went to grad school, so, so why this and why this field? The honest take is I have a lot that I'm very excited about in terms of building and I think that when I was in college, it's very embarrassing, but is a very good snapshot of a version of a Millennial, but I'll go with it. I watched the Social Network in college, and the thing that struck me was not actually the whole story about Facebook and so on, which is great, by the way. Good movie. But the thing that struck me is there's a moment in which Jesse Eisenberg playing Mark Zuckerberg, I think, has to relabel files and writes a Perl script for it. And I was like, fuck, what am I doing manually relabeling my files every day? I need to do computer science.
- [17:28]
B
I'm the only person with that take away from the Social network, but beautiful.
- [17:31]
A
And so I came into it thinking I need to learn to talk to computers because I cannot be a functional human if I don't do that. And then I got to really love the people who were in the computer science department at Columbia at the time when I started doing it. And then I think through that, when I went to grad school, I thought I was going to do coursework in AI, because I was like, this is where the future is headed. And this was in 2016, 17. And then I took one class by a professor called Dan Bonet, who is a cryptography professor, and I was like, oh man, I just need to take all of his classes. He's amazing. And that's how I ended up basically doing the crypto curriculum more or less, and ended up in this field. And so I really find that there are so many areas of deep meaning in human endeavor, and I just happened to, via pure path dependency, gravitate towards this one. And I think the future of ownership and money and what it means to be a stateful actor on the web as pertains to these things is something that I find very interesting. But I don't think I was born thinking like, I want to work on crypto.
- [18:37]
B
When did it flip for you maybe to go back to the French part? Like, when did it flip from academic to entrepreneurial? And obviously that's a dramatic wave. How did that gradient happen?
- [18:48]
A
Yeah, I think. I mean, I think it took a bit, but I. So I finished my undergrad, went to work with a friend of mine, and we both had like, you know, real jobs lined up, quote, unquote.
- [18:59]
B
This is the first company with.
- [19:00]
A
The first company I started exactly with Zach, who started Row now, which is a really successful direct to consumer healthcare company. And I think we really both just wanted to work together. And we went through yc and at the time, I think what was most shocking to me was just how that people Took us seriously, like we were just making shit up as we went along. And like, people just saw that we were going to grind and like, were willing to give us a shot. And I felt that was very, very infectious as a.
- [19:32]
B
Well, I guess this was pre grad school, actually.
- [19:34]
A
This was pre grad school. This was pre grad school. And I went into it and I felt like I had unfinished business with cs. I got into CS too late and I was like, I really should try and get a PhD or something at some point, but let me do this. And basically when Shout, the company we started, failed, I went back to grad school. And a core part of my endeavor there was do I get a PhD or not? And I think at that point it stood to me that I wanted to build products and what was a good place to do research and build products. And that's how I got into crypto. Post fact was like, actually this is the job I should take out of grad school was in crypto, where I can do research, but it leads to products being shipped.
- [20:10]
B
Remind me, you, when did you move to the us College or earlier?
- [20:14]
A
No, earlier in middle school.
- [20:16]
B
And so did you. Was it because of that that you didn't like, what, what, what created this sort of more American bug in you around some mix of ambition and to go back to the agency or whatever?
- [20:28]
A
Like, I mean, I just spent. Arrogance. Yeah, the arrogance. I became arrogant when. No, I mean, I think the honest take is I lived in New York a very long time through when I was like, you know, 10 or 11 to college. And then I think probably a lot of it, I was in a French school and a lot of it, I think came through college. And the people I hung out with. I started my undergrad at Stanford. I was around folks who were just very unusual in the best of ways in like, you know, one of the projects I was very lucky to get involved in through One of my RAs, Alan was we built a tree. Like we built. We went in a backyard of one of the row houses and this was like the wide eyed freshman joining along because it sounded fun. But basically this is folks who are really into theater tech who had decided to try and build a physical sculpture to something. So I think one year they built Egypt in a dorm room where they put down a tarp, stolen sand from the volleyball courts and put it down, and then tried to rebuild that. And this year was a tree. And they were trying to build the tree of life from Norse mythology and they wanted to host dinners in the tree. But what they did is it was a metal structure that had been welded together with then some chicken wire, monster mud, and we had shaped this theatrical tree that then ended up having a table for 12 people and hosting theirs and so on. And so I was just like, this is really fun and cool. And I think finding myself in that environment felt very intoxicating. And so I think honestly, the take is a lot of it is just I was the pure product of that, plus going through YC in 2014. It's like a very specific part of the world at a very specific point in time that ended up making a big impression on me.
- [22:14]
B
When you went back to grad school, did you think you would do another startup?
- [22:16]
A
Yeah, no doubt.
- [22:17]
B
Okay, so there was always sort of a matter of time.
- [22:21]
A
And I think, in fact, like the first startup, we ended up shutting down. And I specifically remember at the time thinking, I do not want to be a founder for founding's sake. There's nothing inherently valuable to the act. And so unless I have an idea that I believe is worth having in the world, why should I do this? And obviously my mindset changed a lot because I came out of grad school being like, I want to build something. And as long as I start with that premise, there are too many interesting things that are broken in the world for me not to find an idea that's going to be interesting.
- [22:55]
B
Speaking of that, I suppose
- [22:59]
A
I think
- [22:59]
B
it'd be cool to trace a few of the arcs of privy.
- [23:01]
A
Sure.
- [23:02]
B
I think it was on Cheeky Pint you said to John or to Zach, like every year we look at the company and it's unrecognizable from the prior year, even the last few years. You're also dealing with a market. Crypto is a market that is both, probably. I think a lot of people had an experience of it not moving as fast as they would want and it moving maybe in directions that they might not want. I think you probably have experienced both
- [23:25]
A
of those
- [23:29]
B
and I think that has made it a really interesting substrate for your deafness in balancing ideology and pragmatism. So maybe to start, you mentioned it earlier, like you have all these old medium posts about like, how do we actually have a world where we can smuggle in privacy or whatever it might be. Let me see if I have it. Well, one of. One of the titles of this post is great. There's too little to be gained from treating your users data with respect, which is probably, I think, maybe broadly still true or still felt and really an example of like, nobody actually cares about this. Why does digital privacy actually matter?
- [24:13]
A
I think for me, basically just the reality is, like, the majority of our lives are now spent in the digital medium, and you have this data exhaust that you're putting out every day. I actually just went through a digital audit by a firm, or they went off of my name and they tried to pull everything. And it's really impressive. And I'm very aware of what's out there and what is intentionally out there versus not. And there were a few things that surprised me, but broadly, it's pretty wild. They went through my entire family tree, all of the things both in crypto terms and not. It was a very interesting experience. And basically I'm just deeply aware of how much digital exhaust we all drag with us. And I think there is this disconnect in terms of risk and danger, of the fact that our sense of fight or flight has been developed in the physical realm. If you go like this, I feel that viscerally. But the amount of risk I'm taking on in a digital surface is very real. And yet my brain is not at all wired to perceive it in the same way.
- [25:16]
B
And so unless it like, slaps you in the face.
- [25:18]
A
Yeah.
- [25:19]
B
AKA somebody finds a Snapchat of you doing cocaine or whatever, and people lose their minds. But. Or the icloud photo leak.
- [25:26]
A
But. But even then, just the feedback cycle is so different than it is in the real world that I think just like, we're not exercising the same judgment in digital spaces as we are in physical spaces. And so at least I grew up just very aware of, like, those risks and dangers and thinking, how do we build back into that?
- [25:45]
B
Why did you grow up so aware of them? From a technical standpoint, I think from
- [25:50]
A
a, like, technical and personal standpoint, I think what I really valued, maybe as a, you know, immigrant coming to the US and so on, like, you know, the most, like, privileged version of like, you know, random white French dude coming to the US but like I said, I think it was being able to control how I present myself to people is extremely important. And the idea for a long time of people being able to Google someone made me very uncomfortable, which is it was always the case that you could show up and you decide what you brought with you into the room. And now actually there is this hidden baggage that you're dragging with you everywhere. And that felt very real to me. And so I think to that end, I think digital privacy ended up being something I really care about. And I was very interested in and still AM And I think the thing that's so interesting in the moment today in tech is just that the arms race has changed completely. The launch of Sora makes it that you can spend a lifetime trying to be careful about your digital exhaust, but now all you need is two angles of your face and three sentences spoken aloud and that's it. You have been ingested by the machine, who can now output you at infinity saying anything that it wants.
- [27:06]
B
We won't even need to do this podcast.
- [27:08]
A
That's the point. One of the things that came up in this audit was there's too much of you on YouTube and the deepfake potential is infinite now. And that is just a cost of living at this point in time. And it's true, but so I think that's very interesting, which is we're in a shifting threat landscape and maybe this is the place where I reject the French ethos the most. This is going to be a very Europor themed podcast, unfortunately. I love France, I feel very French. But the French response to global warming is turn off lights when you leave a room. And it's so clearly not a useful mindset as opposed to thinking about how do we change how nuclear waste is recycled. And so I think the acceleration has been to see the world as it is, be truth seeking, understand the way things actually function.
- [27:54]
B
And.
- [27:54]
A
And given that fact, how do you then counter position to lead to the best outcomes as opposed to trying to rewrite the world? So I'm coming back to the prior question, but I think crypto's biggest sin is trying to reinvent the Internet and financial systems from scratch. It's a very useful intellectual exercise, but that is not how change happens. Change happens based on where we are today, to then moving the machine in the direction that we want. And very seldom, in my opinion is there such a disruption that completely negates past innovation. And we just get the Neanderthal to compete with the Homo sapiens. It ends up being some inner binding of the two. And we're seeing it play out in AI today. But I think the question of how do you protect your digital privacy on the web when AI just makes it that the level of input that you need to destroy privacy is infinitely small is actually quite interesting. And so I'm actually not here lamenting AI and saying is the end of everything, just saying the stakes are super high and the tools at our disposal are very interesting. And to the earlier point still of moral valence, of tooling, creation of napalm led to the creation of the best burn creams we've ever had. And so the solution isn't the problem if we know how to look for it is I guess my hopeful take.
- [29:09]
B
You use the words destroy your privacy. In one of those old medium posts you have this interesting distinction between privacy and security. But I like to go a little deeper on you say you can. You can define privacy as the state of being free from unwanted public attention or observation. This is different from security, whereby a product or service works as its creator intended and does not unwittingly expose or leak information. The relationship is one way. Bad security means bad privacy, but good security does not necessarily mean good privacy. Secure products can degrade their users privacy. In that sense, security is about creator intent, while privacy is about user intent. Outside of like making sure you are secure or safe, what is, what are the ways that privacy actually matters?
- [29:55]
A
I think to me it's like this the most hokey sentence ever, so bear with me. I think it just comes down to like, you know, how much right to self determinism do we have? And that's the distinction between like privacy and security. Security is systems work as intended. Privacy is to me, people can have the expectation that the things they think are happening are in fact happening. And so it's much more in the eyes of the beholder. And again privacy, there's just no single definition for it. Some people love being extremely public on Instagram and they have a very different. And yet they can have a sense of privacy that matters to them, but their definition of privacy is very different.
- [30:30]
B
Yes.
- [30:31]
A
Yeah. I don't know. I think in a world in which basically you were just the composite of all the things you've ever said, written and done, this maybe has some weird tie ins into cancel culture as well. But in a world in which you are just those things and that is what exists of you, then there's just very little ability to make willful decisions to change things in life. You are the sum product of everything you've ever said or thought and saying. Actually I've changed my mind on this. Doesn't really matter anymore. And so I find that there's something deeply problematic with not being able to have maybe humans. This is. We're so far down a rabbit hole, so please take us out. But humans, I guess intrinsically have some version of a decay function where the things I thought a year ago are not that interesting to me anymore, the things I thought six months ago a little bit more and so on. And so there's probably actually some weird curve where the Things I thought when I was seven years old remain extremely important to me. But I think the way we're able to filter out opinions, hold opinions, and then over time filter them out to get to where we are today is really important. And I think in a world in which everything is recorded and stored and can be brought back, it just completely changes our ability to, like, you know, determine what we want to do moving forward.
- [31:46]
B
Lightness is a virtue. Lightness can be. You don't want to. Accountability matters. And being able to move through the world without the weight of everything is freeing in a way that I think is like. This might seem so strange, but I was listening to Kevin Kelly talk about having how having, like, a belief in the future is really important for being able to move forward in the world. And this almost feels like the inversion of that. It's like not having to, like, know that your past is chasing you.
- [32:16]
A
Yeah.
- [32:17]
B
Again, I can be easily, like, taken out of context and applied it. Like, I'm not. It's not an argument against accountability, but.
- [32:25]
A
And maybe this is like, you know, this maybe just points to my deep cowardice, which is one way to. One way to do it is to try and be extremely vocal about the ways you change your mind and have original viewpoints felt strongly that you embrace. And you have a handful of figures on the web who have very strong ideas, very polarizing ideas, and they change very quickly. And again, as long as I can see some version of an intellectual through line for how they got from A to B, I love that and I find it to be so inspiring. But I think the alternative is that's not very private. That's the point. And I think the alternative is don't engage in that way. Don't let people see these things. These are conversations that we get to have in this case on a podcast, but typically one on one. And I've also, in this, been very careful to not espouse any opinion here. I'm like, spewing about what it means to have opinions, but I've not said anything of any valence on any topic. But that's the other point is basically, given that I don't know how this plays out, being more private is the way in which I get to keep my thoughts to myself, have those thoughts and move about in the world unencumbered by having to relate my present set of actions to my past set of thoughts. In a sense, yeah.
- [33:38]
B
And maybe to tie it back to identity, there's some level in which maybe. Maybe part of what this informs is an ability to like, move lightly through the world with multiple identities too, that, that don't need to be like the classic thing, like who you are with your mom versus your wife versus with your best friends from college. Like, those being different doesn't mean you are in conflict as a person. And yet our digital lives, like, make it harder for those things to be harmonious.
- [34:03]
A
There's this rumor going around crypto circles and stripe that I am in fact the heir to the Patek Philippe family. And that's because there's a Henry Stern.
- [34:14]
B
I saw this. I saw this in my research and
- [34:17]
A
it's obviously very untrue, but I've had people come up to me and say, man, your watch. It's so cool. What a fuck you to your family. I'm like, I have no idea what you're talking about. But then I found out that that's what it meant. And. And I find it really fun when basically that version of the substrate of information that we have in the world ends up being completely wrong. And so actually that's a fun idea that I like to toy with, which is can you have misinformation canaries about yourself on the web where you just put out every here and there a completely false fact, but just to kind of trace how data makes it through in identity? So various musings.
- [34:55]
B
It's fascinating how we are just like. So we. We evolved to like have this gossip and these sort of like local context of information you might have heard about a person. And now over the last, really last 25 years, but sort of over the last 80 years now there's a possibility that actually someone can know a tremendous amount about you or have a tremendous sense that they know a lot about you and bring that in. And some of it's stated and so. Yeah.
- [35:23]
A
But I think the other interesting question, and for what it's worth, no one cares about me, which is so lucky at the end of the day, the version of we're trying to change that here. Clear. Thank you. There's a Victor Hugo coach to live happy, live hidden. And the point is, anonymity in itself is such a blessing. And I think it's a very complex thing because I only have a Twitter because of work personally. But also I'm a deep lurker. I enjoy reading other people's stuff. Exactly. But I don't enjoy putting out my opinions because it feels so valent in a sense. But I think the point is this, which is living in a world where we basically went from phase one which was we exist in the moment and the activation energy to capturing something for the future is wild. Or actually maybe phase one happened with the printing press. So there's the pre Gutenberg days where it was oral history for the most part and you had to get a scribe to do anything. Then post Gutenberg, capturing information for posterity becomes that much easier to probably the advent of the Internet and microblogging where now putting out information to be captured for posterity becomes that much easier. So we went from the activation energy and putting out basically long lived information has gone way down. But we were always saved by the fact that no one cares. And the amount of work it would take to do a really deep psychological draw up of someone on the Internet is probably not worth it economically. And that has changed. It is now really easy to send out three agents to cut these things. And I'm curious how your research for the podcast has changed and so on. But my guess is just that basically you can get a deeply harmful psychological profile of anyone for really cheap in a way that used to be reserved for blackballing someone in a deep way. And the question for me is what happens next? And the reality is society will be too schizophrenic if it becomes too easy. I mean this is what we're seeing in many ways with what has happened from 2018 to now in terms of what was the moralization to now just tech itself is an interesting foil for how society is evolving. I find it's a good Rorschach test for you know, cancel culture to now like very abrasive masculinity to like some version of what will come back. And we're just swinging back and forth here. And I guess my question is, in a world where everybody has that power, I actually suspect it becomes a lot less powerful.
- [37:57]
B
Yes, the overturn window has to. I always used to joke back to my crude Snapchat analogy, like at some
- [38:03]
A
point we'll get everybody's got dick pics. Is your dick that interesting? Yes. And. And I think. And I think like that's like how do we keep carrying on meaning in a world in which everyone's sticks are out there? Is the question. That's our sound bite. Yeah.
- [38:20]
B
Two things that come to mind and we'll finally move on. Although I could talk to you about this all day is I think his name is Nathan Jurgensen, this guy who used to write for Snapchat. It's possible I have the name wrong, but basically this philosopher in residence thinking about the impermanence of digital identity or the permanence of it. I feel like you would find it interesting. I don't know what he's doing now, but it's an interesting sort of dumb dumpster dive on. Somebody was thinking about this actually quite early. The. The other one. The other thing that brings to mind, actually, hilariously, is I don't know if you've heard Tyler Cowan talking about how he's writing for the LLMs, and his whole thing is like, he wants the LLMs to know as much as possible about what he thought. So I think part of it is like someone, he's gone, but also, I don't know, he's reading some interview and he's like talking about Iceland. He's like, now the LLMs know what I think about Iceland. And like, if people want to know
- [39:08]
A
what I think about it, they can ask you. They can ask. And for what? It's sort of, this is maybe my draw to action, which is ultimately what we're going through right now is the quintessential intellectual masturbation exercise on so many levels. And I think again, Silicon Valley kind of sometimes talks itself up in histrionics around all of the ways in which. These are all very deep. But I think then taking a step back and seeing what action was taken, how has it actually impacted the world? And is this good or bad? It ends up being quite simple. And so I think being able to move in and out of this context to just, is this company producing something valuable? Does it make me happy as a consumer? And it isn't that much more complicated than that. I find it very important to get back to. Maybe this is why working in tech in New York, this is a call to arms to tech in New York, which happens to be much more easily to ground in reality than probably if you're operating in Silicon Valley.
- [40:04]
B
In what ways does privy still work on privacy?
- [40:07]
A
It's really tough. To a large extent, we pivoted. The core product failed. And I think a lot of the infrastructure and architecture that we built remains like data encryption, key splitting, running secure environments in which we can provably show that no one knows the information. It used to be about knowing information, now it's about securing keys. But the core concept is very much the same. Just the substrate of data has moved from intelligible information to now key information. So I think that the reality is basically it works on it, because I do believe that encryption is key to privacy and having everyone have keys is key to enabling encryption to exist throughout the web. And so I'm probably holding out hope that the self sovereignty project of Web3 will come back as we distribute stablecoins across the world and as we make it that digital currencies work. And I think the project itself is a risk in a sense that there have never been darker days for Web3, and yet I've never been more optimistic about crypto. I think finally the space is having real deep impact on people who otherwise don't care about the technology, which is maybe the definition of impact, is the only space that has ever so deeply pandered to itself for so long. So I'm as optimistic as I can be about this space and about the ability to do meaningful, impactful, positive work in this space. But also I think this idea of we're going to build an alternative Internet that is user owned has never been in more dire straits. And I think my hope is that Privy is still working on privacy because we are sort of laying the groundwork for how to enable people to have keys to signify intent and encrypt data
- [41:53]
B
to jump forward a little bit. I think around the time I met you, you were, and I'm sure I'm skipping over some stuff, but you were starting to think about embedded wallets, which sort of became this massive, amazing wedge in the crypto wallet at the time. And certainly for most of the early history was this sort of idealistic. Go back to the point you were just making this idealistic idea that you were going to bring your self custodial wallet wherever you went. And it wasn't just your money, it was all of the things you were, your identity, whatever. Do you think that there's still merit to the idea of the wallet in that form factor at all?
- [42:31]
A
Yes, and I think in a sense this was the interesting take. We had the biggest issue with identity writ large. Like how do you define identity on the web? Typically people will say this is about basically portable identity. I record something once I get to have it everywhere it is controlled by me, not by a platform. That's at least a very common way I would think about it. And basically any work on identity ends up being a question of how do we incentivize platforms to play along in a game that cuts out their edge and that this intermediates them from their users, which is to say their lifeblood. And so I think that's the core reason why identity projects end up going astray, is if you are pure enough about following the identity path, you end up building a protocol that you can never get anyone to adopt, because all of the people who have the greatest impact in adopting it are the most disincentivized from adopting it. And yet I think what we'll end up with is maybe federated versions of identity, localized. I think there's basically a gap between. You only exist within the platform ecosystem or you exist at a government mandated level. And I think basically being able to federate different parts of identity between platforms is still a very viable dream. And I believe that having basically digital currency that can flow between platforms is actually a very easy testbed for figuring out what a wider wallet can do. If I have a wallet that allows me to take currency in a digital form from platform to platform and then saying currency is just one type of data, but I can take other data, I can take KYC information, I can take. My driver's license is just a spring from there. So I think there's still merit to the wallet. But in itself, I would recommend that no one work on that wallet in itself because it doesn't solve a lived problem for any user in a very direct way.
- [44:20]
B
And yet in some inverted way, as I understand it, feel free to correct me, the embedded wallet was really the, I don't know if that was the exact pivot, but that was the sort of like unlock to really make things start to work for you guys, at least in terms of the kind of core goal of serving developers and having. Do you disagree?
- [44:42]
A
No, I completely agree. But I think what we did in doing this was also we, you know, there's a reason the embedded wallet doesn't have a first class feature right now around like, you know, Social Security number encryption or things like that. Like we moved to saying not only will we give, we will basically move from being this like horizontal platform that can capture any data to being a very verticalized one that can capture one type of data, which is what assets do you hold and how do you make them easy to spend? And then what we did was we'll say, and this moves from being something the user has to willfully, intentionally bring with them to this is something that a developer can build into their platform. So they are now building a global store value or fintech, whether or not they sought it. But now value can be captured at the app level, controlled by the developer in the UX that they want. And so I think the unlock was saying developers can build with crypto without needing their users to be crypto people.
- [45:36]
B
I see. And is that the, is that what made the company work wildly?
- [45:43]
A
Yeah.
- [45:45]
B
When did it, when did it Go from. You're working on allowing customers to basically manage their users data in a way that was responsible. It wasn't really working. You were acquired by Stripe last year and you were quite successful in some dimension of things. And you have.
- [46:05]
A
If we wanted to put maybe the
- [46:06]
B
crudest analysis on it, there are millions of people who have preview wallets by way of your developers. Like, what, what was the inflection point?
- [46:15]
A
I think there were a few chapters, but basically through the first version, it was purely vision built and it was, this is what we think ought to exist in the world. Can we get people to agree? We launched a product in like December of 2021 or 20. Yeah, 2021. By April, Asta was like, this is not working.
- [46:36]
B
We weren't.
- [46:37]
A
She was, she was like, this is too hard, we're pulling teeth. And I was like, yeah, but like, you know, this is about grit and so on. By July I was like, all right, yeah, you're right, this is like too hard. It should not be this hard to get people to care. And like, we are trying to reshape nature as opposed to the market always wins. And we are trying to change human behavior and that is not the right path.
- [46:57]
B
What were you, were you most trying to change? You really were just trying to sell developers something they didn't care about, something
- [47:04]
A
they didn't care about and something their users weren't asking for.
- [47:07]
B
I see.
- [47:07]
A
And we were trying to convince them you should care about this. And by the time you're doing that, you've already lost the game, you've lost the plot. We're not an advocacy group, we're a company building products. And so I think by July I came to agree with her and we had a few paths. We were like, okay, what have we learned?
- [47:29]
B
You had already raised from Sequoia, we'd
- [47:31]
A
already raced from Sequoia. And I think this is where there's real truth to the Idea Maze really matters in so many ways. We pivoted, but it's so related to what we were doing before. And I think part of the question we had over that summer was do we just reset, start something entirely different, go to AI or whatever? And I think the takeaway is like, no, we still really believe that there's something here in the idea Maze. We just went in through the wrong gate. So let's get out of the maze. What did we learn about the path we charted? And let's go back in. We were pretty methodical about it and we came to a few ideas how
- [48:04]
B
sacred was selling to developers very.
- [48:07]
A
Because I'd worked on consumer products before and I know how hard it is. And so that was not questioned.
- [48:13]
B
But it was always developer, specifically developer infrastructure.
- [48:16]
A
And I think the sense here was like, it's what we know. We're both developers, we care about the craft. And frankly, again, if you want to be selling the means of creating the future, that is a very good way of having impact. And so developer tooling is the way in which we get to start. That wasn't questioned, interestingly enough. But I. Yeah, and so we got to this idea of we need to change onboarding. It started with onboarding and we knew that embedded wallets were a part of it. But actually the first launch was not embedded wallets. It was enabling people to off into crypto apps and link their existing crypto wallets to an off system more easily. And we basically just. We'd seen the data, and the data was people want to have key management be easier, so we'll have to build that next. But we're trying to cut very thin MVPs. And so basically by January of 23, so a year later, after the initial launch, we relaunched with this new product and we had one foot in front of the other goals. Our first quarter, we were like, we need to have five customers in prod. Our second quarter, we're like, can we make it 12? Our third quarter, it was like, can we make it 25? And we got very lucky because there's an application called Friend Tech that used us. And we were very clear that that used us. That blew up and then everybody started copying them. And so we went from our goal being 25 in Q3 to having about 150 developer teams building with us at the end of Q3. And that point we had enough data and enough customer contact that we. Oblivion or irrelevance was no longer the threat model for why this wouldn't work. It was us just messing the product up, not being able to scale.
- [49:55]
B
Was there any fear of will there be? Because interestingly, we knew each other back then. I remember. Well, before Friend Tech, you had, like, envisioned this correctly, which is like, I need to get enough developers that one of them is my. One of them rips Basically after that one, I guess, do you think that was a real, like, a reliable strategy that can actually work? And then two, was there any fear that, like, will there be another front tech?
- [50:20]
A
I think the fear was, can crypto. Yeah. Exist beyond the, like, you know, called following? It has. Yeah. And I think we were basically very Comfortable or at least appropriately uncomfortable. Making a core bet which was we have to have one like we can't stack bets. I think that was a very important like mindset that we had which is like the startup can be a series of like, you know, bank shots and pools where you have to hit like three sides before you get the ball in. There needs to be one core bet that we're making, but then if that macro core bet works out or doesn't work out, we can succeed or fail. And I think our core bet was crypto, be it in the web 3 version and digital assets version or something else is such a fundamentally interesting idea. Having a state that is not owned by any party, but that every party in the world can connect to across the Internet is so clearly valuable. And the first application of it is magic Internet money, which is the old name for stablecoins, I guess now I
- [51:19]
B
was going to say back then it almost felt like like Web three didn't work.
- [51:24]
A
Yeah.
- [51:25]
B
And so in some sense like it was, at the very least it didn't seem that obvious then that there was still this like very long run bet that stablecoins and like the quote unquote boring stuff was still.
- [51:39]
A
I think that for us the point was friend tech gave us relevance and enabled us to test things. And I think the guiding star was can we work with tasteful developers? If we find customers who have taste, what they're working on is far less important than the fact that they understand what they're working on.
- [52:00]
B
What does taste mean?
- [52:02]
A
It means, I mean the negative signals is someone who is mimicking a trend as opposed to doing something because they believe in it for first principle reasons, whether the first principles are correct or not. And it means somebody who understands their audience and their customers deeply and is making decisions based on that understanding rather than just based on proxies for what an audience might be. And so I think we've been able to just spend a lot of time with our customers and identify who has real vision for what they're trying to do and then let's just be better for them. And so I think we basically spent this first. Chapter one was pre pivot. Chapter two is this period of getting to a place where we had enough of a customer base to start iterating. And chapter three is how does this stretch the product and what are the places in which we can just build better tooling for these people? Leading to chapter four, which is seeing the emergence of stablecoins and a new class of user coming into crypto and making sure we were well built to support that use case. And so I think we wrote a strategy paper early on which basically said there are three markets, there's the extremely die hard crypto natives for whom storing a mnemonic is a non negotiable fact of self sovereignty and we will never be able to convince them that actually these are really viable self custodial systems that they should use. There is crypto developers who want to build for a mainstream audience and then there are mainstream developers who want to use crypto. And basically we said we will never win one and three doesn't exist. And so we have to go for two. We have to build for basically crypto devs who want to cater to a wider population that will teach us the things that we need so that when three comes about we'll be ready. And that is like we just got massively lucky with timing that three is coming now three is here.
- [53:54]
B
I mean it wasn't like it. I don't know if you got massively lucky. It was over the course of four years.
- [53:59]
A
Yeah, we were patient and lucky. You hung around, we hung around. But we're now in a place where basically most of our biggest customers aren't here because they care about crypto, they're here because they care about the impact that this technology can have on their users. They're here to build better financial systems. They're here to enable people to hold US dollars globally if they don't have access to local currency or stable currency. They are here to build better products for orgs that already exist and are massively scaled. And we're really well set up to work with these people because we have two markets, we have the folks who are really pushing the technology and these are the deep experts, the crypto natives who are forcing us to be faster, to scale globally. The reality is DEFI is just doing so much more transaction volume today than stablecoins. So we know that we can scale to hold any stablecoin like thing because we've scaled to support Hyper liquid and Uniswap and Jupiter and on the other side the you know, fintechs and marketplaces will kind of actually distribute this technology more widely. So it's very much a two pronged approach to product development and being very clear around what is our customers optimization function here. So we can make sure that the product is right for them, but then making sure that like we're able to work with startups and enterprises in a way where they feed each other in terms of product quality.
- [55:19]
B
You referenced on an earlier podcast, having written this document, I think probably in 2025, but maybe it was 24 about the sort of two paths between trading and silver coins and not choosing.
- [55:31]
A
Yeah.
- [55:31]
B
Maybe that's a similar cut on what you were just talking about. But like, you. There was somewhere else you talked about. Like, part of the role is to be a hoodie amongst the suits and a suit amongst the hood.
- [55:43]
A
Yeah.
- [55:44]
B
How has that been? Maybe it was especially relevant pre stripe, but it's certainly still relevant today. And I, like, I. I know that's you. You've talked about this, like, make making sure that you're. You don't just go after the. The stablecoin kind of rise.
- [55:59]
A
Yeah. I mean, I think that was probably the second hardest moment. Like, the pivot was probably the hardest moment in the company's life where, you know, we. We were in a room on 600 California Avenue or Avenue, I think, in San Francisco. It was a WeWork. We were doing an off site and just the atmosphere sucked. And it was like six or seven of us, the entire team. And at some point, I think I said so clearly things aren't working right. And you felt the body language change and people relax. Being like, finally, at least we're not just. Yeah. And at least we're all. We have eyes. Of course it's not working. We all see it, but now it's like part of the conversation and we can do something about it. That was like, the start of broadly a very painful. I mean, it was already quite painful to be working on something no one cared about. But it was the start of just figuring out what are we doing. And it was done very publicly. We have one teammate, Ankush, who was like, this is the first job out of college. I think it was like, he was probably like, a few weeks into the job, and I'm, like, just so thankful that he hung on because that was not a fun time to be around the company.
- [57:02]
B
Still here.
- [57:03]
A
You're still here. Wow. And he's amazing. And his job has changed 18 times in the course of, like, the Org. But he runs a lot of what we do in product generally. But I think, like, that for summer, we were doing, like, we'd set it up so we would do random research to try and figure out what we would do. The second hardest moment was basically much more of a private moment, but it was me and Asta and just seeing the market is changing. And this is when this was October 24th. I wrote this doc that I sent to one investor Being like, I think we're headed towards a wall.
- [57:41]
B
And like, like Hyper Liquid is already a customer.
- [57:44]
A
Hyper Liquid is already a customer.
- [57:45]
B
Like the trading stuff is working.
- [57:48]
A
The trading stuff is working. But we, you know, I think our stance was always where we are. We want to be big fish in a small pond and then grow the pond. And we will not focus on monetization and we will not focus like we have built a SaaS business. And so in many ways, like this was the actual like crux of the doc, which was we're seeing the boat depart from the dock. We are on the dock. We have to either jump on the boat or stay on the dock.
- [58:13]
B
The boat is stablecoins.
- [58:14]
A
The boat is a new vertical coming into the space. Stablecoins, tokenized assets and generally the fintechs entering the space that will clearly be massive and requires a set of changes to our product and the org that we need to make to stay relevant. And I think the point was if we stay on the dock, we cannot be a SaaS product. We have to figure out how to be a very different type of product because there are great businesses to be built in trading. But we have not built such a business. We never intended to.
- [58:43]
B
SaaS and trading, exactly.
- [58:44]
A
It doesn't make sense. And conversely, if we want to jump on the boat, that's great. But we're giving up what was hard earned but also what was, I think a very meaningful set of customers who have great ideas, who are really insightful, who we want to keep working with. And so I think the point was this needs to be a very intentional, deliberate decision that we make for how we want to be in the world. And I think the point was we have to do both. We can't control the timing of stablecoin adoption and not using the platform we have would be such a mistake because it's a valuable platform and I actually think it feeds product quality in a deep way. And to the point I think we are now on the latest part of that transition. We have made the transition. Most of our customers are moving stablecoin volumes. They're fintechs again. They're folks like Ramp or Klarna or Deal. And these are very modern companies who are here for the user impact. But at the time it was we actually have a platform. We just need to rethink our customer base a little bit in order to make that movement. The conversation that I think had a real impact on me a little bit was I went to see Roelof from Sequoia around the Time we were raising a Series A right after we had raised it and I asked him, what would it take? What would it have taken for Sequoia to be excited to do ra? And we talked for a little while and he was like, you're talking to me about how far you've come and I don't give a shit. The only thing I care about is am I excited? Looking forward. And I think that was such a wake up moment for the obvious searing pain of trying to create something from nothing is so wild in your mind that you end up saying, oh, but we've done so well and no one
- [60:27]
B
gives we're not dead, we're not dead.
- [60:29]
A
Exactly. And so I think that that's been a very useful lesson to carry forward, which is the only thing, like, let's restart the clock today. What do we have to do? And this is why I find just the entirety of say, SpaceX as a company and the way that we're operating in AI give a fantastic business. And you're now going to, I'm speaking a little bit provocatively, but you're not going to risk the entire business to go after the next bigger, biggest thing. And so like that ability to only look forward is like, you know, it goes back to the good at that.
- [61:02]
B
Yeah, Elon's good at that.
- [61:03]
A
So that's very important.
- [61:06]
B
So easy to say from the outside looking in, but like, I guess you could say, like, obviously you should have done both. What. What were you comp. What did you have to compromise on and what did you not compromise on to actually make that happen? And why was it so hard to do both?
- [61:19]
A
Yeah. And to our credit, maybe the thing I like is we did not ape into the space. This is at a time like, you know, I really do credit like Zach and Bridge for inventing a lot of how people think about stablecoins today and making Fetch happen. I think we were basically making. What happened? Fetch happened. Sorry, Mean girls reference, please.
- [61:40]
B
I thought you were talking about some tactical fetch.
- [61:42]
A
Yeah, fetch, It's a technical term. I think we were early enough that it felt very risky to make this bet at the time, which now every company that is still operating in crypto has pivoted to stablecoins. So it seems very obvious post fact, but I think we were lucky to see it a bit earlier. I think there were probably three core trade offs. One, marketing and product positioning. We rebranded, we did a number of things to actually self actualize for the market. We wanted to build toward to actual decisions in the stack privy While key reconstitution was all happening on device at the time and we were like man and key management wallets were very tied into auth because most of our customers did not have authentication systems. We're like, well there's no world in which a bank is going to rework their AUTH system to integrate wallets. We've completely fucked up the stack here by tying these things together. So we need to rethink core abstractions, we need to rethink the relationship between wallets and authentication, we need to rethink key management because it can't keep happening on device. It needs to be like. And this is what led us to TES and secure environments in the cloud where we could provably have keys that no one could touch but the rightful owner. But also it did not need to run locally in order for that to be true. So it was a deep set of. It took us the better part of eight or nine months.
- [63:01]
B
Just like a resource, it was focused
- [63:03]
A
in resources and then it was clarity internally. The question of who are we building for then comes up a ton. And I think basically, you know, building a resilient culture in startups is getting everyone very comfortable with ambiguity and with saying we don't know, but we know that we don't know so we can figure it out. And I guess my core theory of startups, great start, test sentence is you basically have like three audiences. You have your users, your team and your investors in that order of importance. And whenever you start to talk to one of these audiences in a way that is too unlike the way you talk to the others, you're setting yourself up for failure or reckoning. And so I think trying to keep the right tension on the rubber band so that we know the ambiguity, we're transparent internally, we know we haven't figured it out, but we have to figure it out. That's the most costly thing is just feeling on a day to day basis, man, this could be wildly wrong and we're wasting precious brain power trying to figure out this made up thing when in fact we're doing pretty well on the crypto native side of the house. And so why should we threaten to destroy the entire enterprise in order to make this leap when in fact post fact not doing that would have been disastrous. And again, this is so easy mode, right? I had this other conversation, I'm going to keep name dropping Sequoia Partners. I had this other conversation with I think Alfred Lin who was talking about a company that was at $100 million run rate and had to decide whether they were going to destroy their business to start a new business. Like we were doing this as like a teensy, like, you know, 20person.org seeing something that like was, was in hindsight very obvious. But like, there are companies that are having to reinvent themselves at such massive scales and that's like real courage in many ways.
- [64:49]
B
You are, I think you benefited certainly post pivot from being like very customer obsessed through this period. Despite a customer base that was a little all over the place and moving target. Where do you think vision has remained really important?
- [65:08]
A
Yeah, it's a really good take. And I would argue that this is one of the things I'm not very good at as a founder, which is to say customer obsession in many ways is the vision itself.
- [65:17]
B
By the way, lots of amazing content have been built basically on that alone.
- [65:21]
A
We're working in an interesting place that is complex and hard technically, that is morally valent and important. I believe in the future of how the web gets built and as long as we pick the right customers and we kill ourselves to serve them really well, things will kind of work out. And I think that's been a lot of it. I think from there the parts of vision are, imagine a world in which every Uber driver can get paid out globally in a currency of their choice, can save in the app directly, and can refinance their next vehicle to build a bigger business off of that. That world exists today and that world is no longer just one that only Uber can build. Any three person team actually has the means of doing that. It's extremely powerful. Like we're rebuilding the financial stack from the ground up in an Internet native way. And like, I really do believe that, but I think I'm like a very skeptical person. And so when I hear like, you know, founders talk about vision, I'm like, oh, this person's trying to sell me bullshit. And it makes it less exciting to me than actually talking about the craft.
- [66:27]
B
And well, and. And that last example, or frankly a lot of the stablecoin examples are like kind of boring to Americans who have like a pretty good financial system. So like the vision part of it's like lowercase V, like.
- [66:39]
A
Yeah, that's right.
- [66:43]
B
When was the acquisition? Mid 2025.
- [66:46]
A
Yeah, it closed in. We now sit in June of last year. Right.
- [66:52]
B
You run Privy as a somewhat separate org inside of stripe. And then you also work on, I think, some stripe things. Like what is Privy today? What is the right way to model what it is and what it will continue to be. It's a great question versus being just folded into Stripe's sort of massive crypto effort.
- [67:16]
A
I mean, I think a lot of what was exciting about joining Stripe was just the quality of leadership at Stripe. Here you have me brown nosing up to my bosses, but the reality is that it was generally just impressive to see the scale of ambition, the commitment to craft, and the humility with which all three being true together. They were approaching the space. And that made me feel like if there were to be a home, this is the right home. And I think the point was, when we entered, I think we had sort of three principles, which is to say we need to continue to operate as a startup. If we stop, we die, because this market does not exist and we are in market formation phase. Same is true for AI in my opinion, by the way, which is the market is annealing and we need to heat the product and the market to the same temperature to shape where the market goes. But product market fit is not a good framework for working in a market like this today.
- [68:06]
B
What do you mean?
- [68:07]
A
The market moves too quickly. If we start settling down and just trying to scale the shit out of something, it'll move out from under us. And so we need to be much more dynamically questioning everything all the time and making decisions that don't scale accordingly. So I think that was like the first. The second is we can't just be a stablecoin shop. I think stablecoins are the lion's share of where value is being built. And I think they are such an exciting development. But I think there's a lot of interplay between different parts of the space that needs to be fed over time as well. And so we should continue to work on both. Inappropriate measure, but we should continue to work on both. And then I think the third is, like, we have to be able to work with, like, Stripe Bridge competitors. I'd include Tempo in that now. But, like, basically a lot of this stack is being built as an open platform. And if we try to overly verticalize the product, experience needs to be seamless in the parts where we are working together. The developer should not have to see where one product begins and the other ends. It should just feel like a cohesive whole. But for those developers who want to shape their own stack, the pieces should work modularly as well. So those were three very clear things that I felt strongly about coming into the conversations.
- [69:11]
B
Is Tempo already a part of the
- [69:13]
A
conversation, or at least not for me? And so a lot of that was an explicit part of the conversations that we were having. And then I think it came down to some version of do I believe the powers that be, that they mean this and that we get to do really excellent work together and that in places where we disagree, we will come out better for the disagreement with something that is actually something I can be just prouder of building. And I think so that's a long winded way of. This is the core substrate within which I operate. And then I think the point is prev's two things. Privy is a developer tool that people can use to embed wallets and access and ownership of digital assets into their platform. And Privy in many ways is like Stripe's wallet stack. And so within Stripe products, Treasury Link, we are working to power basically the, the, the, the inner piece. And we're like spending a lot of our time with Stripe teams to help shape this. And then we're spending a lot of time with our other customers that are extremely important. And again, this is like the two feed each other, which is the work that we're doing for Stripe benefits everyone because it is like extremely demanding, it scales, it like has all of these requirements. And likewise a lot of our customers are running so far ahead of the market that we get to build a future proof tool. So I think that is the core definition of Privy. And then increasingly we're trying to figure out where do we make it that one plus one equals three. Like I'm helping run the on ramp team at Stripe and this is one where Privy basically just really needs on ramps to work better in order to enable people to load up their wallets with assets, after which it becomes much more interesting to have a wallet. And so the question of like, how do we work both with on ramp vendors across the board because no product is perfect for everyone. But how do we also make sure that the integration surface between these products is really good and that the integration surface across the entire crypto ecosystem, which is to say Tempo Bridge, Privy, Stripe is really good. That's what I would say. Like my, I guess three hats today are focusing on stripe products wholesale that touch stablecoins and crypto. And there's many of those and I just manage a few of them. Which is to say like, you know, merchants getting paid with stablecoins and on ramps, it's Privy. And then it's how do we build an ecosystem that can run at speed on point solutions and yet build a greater whole as we do?
- [71:43]
B
So is a wallet just like a way for a developer's app to let Users have crypto.
- [71:50]
A
Yes, in my definition of it, yes. I think there's. My sense is you have consumer wallets. It's a way for a consumer to hold and trade crypto. You have developer wallets, embedded wallets, which is a way for developers to build these assets into their app. And then basically digital asset accounts, as you were saying, is like this higher level abstraction that brings together the many integrations, that brings together card issuing, orchestration wallets, so that a developer doesn't have to think about putting the stack together themselves, but can just build with one thing and it just works. Bank account in a box, so to speak.
- [72:22]
B
What does it mean for Privy to. You talked about customers. Like, what does it mean for Privy to be a startup within Stripe? Because versus you working on other things, like on ramps at Stripe, which is another team. Is Privy just a team at Stripe? Like what does that.
- [72:36]
A
It's an interesting question and one where, you know, my answer is probably very different. The more I do work within Stripe and the more I understand Stripe.
- [72:43]
B
Yeah.
- [72:45]
A
And the short answer is yes and no. I think one of the great things that startups have going for them is continuity of vision and continuity of culture. And in a sense, Privy is another team within Stripe, but it's another team that's had basically very consistent staffing and direction for four years. And so we have a lot of internal memes that exist through pain and good times that can carry forward and that's extremely powerful, which accounts for like some difference in how we operate.
- [73:17]
B
So I think, like, I'm an American. And I'm a New Yorker.
- [73:20]
A
Exactly, that's exactly right. And I think the reality is like actually the Privy and the Stripe culture are very well matched for each other and that's not a mistake. Like, we literally looked at Stripe in terms of how do we want to build this company. And so we tried to fashion ourselves after the bizarro version of or after the marketing version of Stripe. And we're very lucky in that the marketing version of Stripe and the reality are very close together. So that worked out pretty well. But yeah, I think culturally we share a lot of things and I think there's cultural quirks, ways in which we work, some operational tendencies that are a little bit disjoint. Just because we want to keep running at scale for our customers and trying to fall in line has no intrinsic value. What does is being able to do good work together. Quickly.
- [74:06]
B
What do you say? Maybe to go back to some of our ideology and pragmatism, I'm not the best person to ask this and you may not be the best person to answer this. So I don't want to. We don't need to spend a ton of time on it. But, like, there is concern. Yeah. It's nice that Privy's a team within Stripes. Nice that Bridges and team in Stripe Tempo is not just Stripe. Like there is concern amongst people in crypto. You're building the monolith and that like great. After all of this, like, you're going to build a really amazing performant database. Ish thing that basically, like the mostly good guys at Stripe, like, it's a pretty benevolent God. What do you think? Like, conflating many things. If you were pitching a Henry from 2020.
- [74:52]
A
Yeah.
- [74:52]
B
Who is unconvinced on this, what would you say to him?
- [74:57]
A
I have three takes. The most blunt one to the Henry from 2020. And this is not a good sales pitch for the folks who are very worried about this. But it's like, what do you prefer? A space that is irrelevant, that has no adoption, or a space that is making pragmatic decisions to drive things forward?
- [75:10]
B
I think they would say that on some level that's a cop out.
- [75:14]
A
Yeah. I really don't think it is. First, I think that reports of crypto's death have been greatly exaggerated, to quote Winston Churchill. I think we should be as optimistic as we've ever been. And now is a great time to build a business into, you know, again, crypto. Is it stable? Coins. Like the word itself carries too much. But, like, I think it's a great time to be thinking about what does ownership on the web mean and what does it mean to have like global financial. Rails. So that would be the first version. I don't think it's a cop out because I genuinely think some people, like, would rather
- [75:48]
B
nothing at all. Yeah.
- [75:49]
A
Than this. So that's. I think the first part. I think the second part is there is a real commitment and this is such a lived reality to an OpenStack. We work with LightSpark and Plasma and Solana and Base and these are all very close partners, some of whose infrastructure we power. And just like the reality is we have made commitments and hopefully lived up to them. And I live to get called out, basically because that's the point to actually building in an open ecosystem and an open stack. And so now basically I'm saying Stripe is an ambitious company that wherever it runs into issues, we'll try to solve them by building software, which is what we know how to do, especially when we feel like we have good instincts for what the software should be. And this is how I think Tempo was born. It was born out of issues with chains that needed managing. And I think choosing not to do something that we think we can do a good job of because it looks bad is such a bad way of trying to have impact on the world. So I would say I think it's a very valid concern. I think the commitment to an OpenStack is absolute and so the pieces work well together and you don't need to do the whole thing if you don't want to. But also I wish there were a lot more people in the space trying to build end to end solutions that could transform, transform developers and consumers lives. And I genuinely think it's a really good thing when some people do. So that's I think kind of my take. And again whenever people dunk on us on Twitter it's usually ah, of course this is the Stripe acquisition. Privy's gone downhill since or some version and it doesn't happen that often thankfully and we work our asses off. So I'm glad it doesn't happen that often. But let me say our incompetence is very much our own. In the many places where Privy falls short, it's because we are incapable in those instances. It is not because there's something else. And I actually think working in service to the user is a deeply ingrained cultural value that both orgs share. And that's kind of true north.
- [77:57]
B
It's something I was thinking about recently is there's a difference between in the AI context people are talking a lot about values and I'm more interested in incentives than I am in values. I don't think Facebook necessarily has bad values. Maybe they do, but I think they've had some bad incentives even. Even something as simple as like a commitment to being open source. Like what are the, what are the. Sort of like it's nice that you can say that but like what are the structural things that, that actually allow this to happen and have it still be aligned with Stripe's business goals.
- [78:26]
A
My honest take is, is, is Stripe being a private company and having very strong willed founders is a lot of that. Which is to say, you know, Tammy and Stripe Press existing Stripe climate being a thing, Atlas being a thing. Many examples of things that make a priori no business sense but happen to be intellectually interesting and meaningful. And I think Stripe's ability to try and I think Stripe's core belief, this is the core incentive. I think Stripe's core Belief that startups will always have a place in moving the world forward and that serving individuals and small teams who are trying to build things is intrinsically valuable is that core incentive. It is a belief in the future. In a sense, it's a very optimistic place in that regard. And so the point, I think it's
- [79:17]
B
closer to a value, but it's a value that they really are showing that they believe.
- [79:21]
A
Exactly. And so I think it's a very long again, these are all conversations that I had before joining and I kind of had my bullshit meter on to try and make sure we knew what we were getting into. And so at least for me, it feels like a very real hill that they'll die on, not out of a principle, but out of sheer economic interest, which is the future will be very exciting and the only way to stay relevant is to embrace it. And that means to embrace new things being built, including competition. And so I think frankly a lot of the decisions that have driven Stripe products being built in the crypto space have been born out of frustration as users that we can't build on this substrate and so we need something better and oh, in fact we have this idea. And so I think in many ways the short term economic interest would be much easier to not do these things right. So I think again it's up to us to stick the landing and make sure that the developer experience building on these Rails is absolutely excellent.
- [80:23]
B
What do you think Stripe is becoming?
- [80:27]
A
I've heard a sentence and I don't know that this is canon, so don't quote me on it.
- [80:31]
B
I'm not asking for the company view,
- [80:32]
A
but I think it's first great developer tooling that can be used for anyone to build meaningful businesses on the web is probably the lowest level true thing and you see it because Atlas is a way to build a company and strike payments is a way to monetize it. And now we have these financial Rails for money movement and all that. So that's the first thing is the platform that developers come to to power the financial infrastructure for businesses they want to build online and in the real world. I think the second one is like much more of a tooling first perspective, which is like kind of AWS for money in a sense. If you move into a world where money is data, then what are the core tools and infrastructure pieces that have to be assembled to build with that construct money. And I think a lot of the things happening in the agentic world speak to that, which is we are moving in a place where just Subsent transactions were never a very big concern, I'm guessing because no one was making subsent purchases. But in a world in which agents can stream payments for content and you're buying by the megabyte, terrible idea. But the physics of money change and I think that role as, you know, how do we build the tools to power it and how do we become guardians for like the sanctity of the basically like consumer and business interest in being able to rely on these rails to grow is I think a core part of what, what, what Stripe is, is, is becoming. And then I think, you know, the third. And here I'm just reading through the same lines, you know, everyone else is by looking at like, what is Stripe putting out in the world is like, hopefully an agent for like acceleration of tasteful things in the world. The commitment to beauty and craft. I've spent more time than I care to think about talking to Patrick about the floral arrangements at Stripe offices and the smell of Stripe offices. Not actually a joke. And so there is a real point of craft is in itself something that is worth pursuing. And if we can help people hone their craft by focusing on the right things and we ourselves can put out things that are beautiful, that is intrinsically valuable and something that moves the world forward. You can see I've drunk the Kool Aid, but I believe it.
- [82:55]
B
A couple questions on crypto, broadly. You have a line somewhere you say the difference in crypto between the theoretically possible and the actually true, which obviously runs through a lot of our conversation, I guess at a super high level. I mean, like I, I did a tour of duty in crypto. That's how I know you. There was this grand vision, there were different words for it. I never really loved the Web3 name. Maybe the thing that felt most important about it was ownership. And that could be digital ownership of different types of assets, some of which may be worth more or less than they were in the past. But even something as simple as like users should own the networks they use, the tools they use. Do you think that that dream is dead?
- [83:44]
A
I don't think so. I mean, I think things are cyclical and I think this is going to be a very cop out answer. But like, I think things are cyclical and I think frankly, again, the tension between hype and reality in any space is such a dangerous thing. And you can only go so far in stretching the band towards hype before reality and impact has to catch up with it or the band breaks and bubbles burst. And I think right now Us focusing on delivering value to the world through concrete economic tools that are empowering to people in the western and non western worlds alike is so fucking important to actually building the means of bringing that back. So I'm feeling very optimistic because I've basically never seen crypto deliver so much value to the world in a way that wasn't necessarily what we tried. But I think this will come back, if only. And this doesn't necessarily mean the crypto bros of old will come back. I think many of them have washed out forever because it's such a hard space to work in in many ways. But I think if you give it 10 years and impact can wash some of the stink off of crypto, then it just stands to reason that people thinking about what does ownership mean on the web. The web is becoming more and more important. AI in general, digital cognition is not going anywhere. These questions are becoming more important than ever. And we are building out again just a very important rail for reasoning with what does it mean to own things on this platform. So I actually feel very good because this is an essential question for the future and now whether it ends up looking like what we thought Web3 was going to look like or not. I've learned to be less fussed about trying to shape the market to my will rather than hopefully make forward progress in a positive way.
- [85:33]
B
Yeah, I guess my feeling is it sort of feels that crypto has barbelled, which is that there's the boring part which is stablecoins which basically can't have speculation. I shouldn't say can't, but it's very hard. And then wherever there can be any speculation, there's max speculation, which basically means trading.
- [85:49]
A
But this is the interesting. There's a good Vitalik tweet which was some version of what we thought was going to happen. Crypto will become sane and start resembling normal markets. What actually happened? Normal markets look like crypto. And so I think crypto has been a forerunner and distillation of a lot of what's happening in the world. But I think you see it in so many other places and maybe in that sense crypto there is such a thing as crypto native culture. And I think it is a hyper concentration. It is the moonshine of the Internet in many ways where it just front ran what happens when you over financialize things to a large extent and when Internet memetics take on an intrinsic importance. And I guess my point is I think the barbell is real. I think in both cases each side of the barbell is seeping out of crypto. I think prediction markets in many ways, where a lot of the speculative liquidity from crypto, it's a bigger market because there are a lot of people who are using those in very different ways have no relation to crypto. But a lot of the folks who were dabbling in meme coins or prior to that in other spaces have recycled capital to move towards those markets. And so you're seeing two worlds meet there, and on the other side, on the stablecoin side, you're seeing two worlds meet of folks like me working with financial institutions and traditional global marketplaces to deliver real value to the users in that way. And so I think both sides are very interesting. And basically, I think the question of very soon we're moving into the age where saying I'm a VC investing in Internet companies so obviously makes no sense. We had that version, then we had cloud, makes no sense anymore either. And we're moving to an age where saying, like, I focus on AI makes less and less sense. And I basically just think crypto is slower. But there will come a point where saying, like, my products touch, digital value will be part of the fabric. And so I just, I don't know, basically that like, what won't happen, I believe, is crypto native culture will be everyone's culture and we will move to a cypherpunk world in which everyone's carrying a ledger in their pocket. I love ledger, but like, I just think it is going to become part of the fabric of the Internet in a much deeper way and it will not resemble what we thought it might. So, yes, Web3 is dead.
- [88:09]
B
I agree with all of that and the bar, like neither of those things is these more idealistic ideas of like, users being able to own a network.
- [88:19]
A
Oh, sorry, yeah, bringing that back. I am.
- [88:22]
B
Or. Or even something as like, whatever, digital assets that aren't rampantly speculative or whatever.
- [88:28]
A
When I got into crypto, what excited me was visions of limewire and Kazaa. And this was basically, we've taken the very cool distributed systems research from the early 2000s and we have now bolted incentives into them in a very deep way. And that will be the resurgence of federated and decentralized networks as the core substrate for building information. And like, you see it even right now with the AI labs, the fact that we're in a, you know, depending on how you look at it, version of an oligopoly for intelligence. There are so many fundamentally centralizing forces in the world. I'm just unconvinced that user owned networks. I think it is a beautiful idea and I worry that market realities and human nature make it that actually kiboots and isn't this part of the game of markets?
- [89:22]
B
Like aren't markets supposed to be how we.
- [89:24]
A
Well, so far capitalism is winning. So like the kibbutz and the commune, great ideas works at a small scale. The fact though.
- [89:31]
B
Oh, I don't mean to that I. I more mean like let a thousand flowers bloom. Why don't we have like the fact that there are two AI labs or three AI labs and like it seems that the power law is going to keep hitting harder and harder and harder, which is the. If you were going to create a bear case against capitalism and markets, it would be that they actually enable the extreme end of the power. I realize that's a no, no, I
- [89:52]
A
understand but I guess I worked. So I worked on filecoin for a number of years which is trying to decentralize access to storage. And like basically there are two things at play. There's like, you know, market forces and then there's like technological advancement. The reality is like Moore's Law is continuing to run and so the relative value of decentralizing storage is much less important because storage is in itself getting cheaper and so the market need will be reduced. And we were already up against centralizing forces in the market. And so I believe that we're seeing the same thing play out with AI. The incentives determine outcomes. There is clear demand for intelligence and GPU production and general ability to do matrix multiplication is growing like a weed where we are now energy constrained as the core bottleneck rather than chip constrained. I believe that there's too much path dependency for such a system to come out because people have thought it was a better alternative. To me, it's the same futility as asking AI labs to pause research while we figure it out.
- [90:51]
B
Yeah, okay, fair. There are two threads I want to pull out that I'm going to conflate into one question which maybe deliberately be difficult, which is there's this power law thing which it seems like the world's actually technology is making the world more power law and more concentrated.
- [91:08]
A
Well, yes and no, because at the same time I remember an era circa 2014-2018 where it was like we are kind of at the end of history in terms of what big tech can mean. Like you've got the FAANG at the time and like no other big platform can compete. Like AWS has got the, like, you know this and Facebook has got the that and, and that's just like clearly so not true anymore. Like disruption has come from left field. But so basically I think what is so cool and exciting about working in this space is we are just expanding what is possible for humanity to do. You've got SpaceX and multi planetary systems existing in a very real way. And you had Loom versus Starlink. So I'm just saying, I actually think old systems ossify and the power law is relentless, but innovation and sort of creative destruction is an unstoppable force.
- [92:04]
B
The power will keep getting bigger and bigger and yet we have new disruption.
- [92:08]
A
Exactly right.
- [92:09]
B
I guess that there's a question about optimism. My other question about optimism, which is
- [92:12]
A
a very optimistic take for what it is that I have.
- [92:14]
B
Yes. What I was going to conflate is are you really that optimistic about the future? And that was one piece of it. The other piece, though, you brought up the Vitalik tweet I had written down. You had retweeted Andrew Reed on the same note and you had likened it to Vitalik. He says everything is crypto now. The launch videos, the fake ar, the fake users, the affiliate kickbacks, the coordinated comps, the group chats, the random five from competitors, wild financialization of the world. Yeah, you are one of your. You are like, I always joke with people. Crypto is hilarious because you have like the worst people ever and then just like the most thoughtful, brilliant nerds ever. And they're like, they're combined into one thing that's less true than it was three years ago. What do you think? And again, you're not work. Although you guys pumped out fun uses privy.
- [93:02]
A
Like, we're very proud of our work with again, these very tasteful developers.
- [93:05]
B
What do you think is going to happen? Are we just headed to a world where we all have UBI and people gamble their lives away and like the government, like, that's overly nihilistic.
- [93:16]
A
But first I. Who the fuck am I to like espouse any of you with regards to this thing? So I'll just say I'm wildly outclassed. I can share my personal opinions as
- [93:24]
B
I'm more curious of if you're more optimistic or pessimistic and how you think about what.
- [93:29]
A
My take is twofold, which is I think the permanent underclass meme is very powerful. And for those who don't know, the meme is basically, if you move to a world in which intelligence is on tap and you have a slot machine for intelligence, then basically being the richest and being the smartest are the same thing. And there is no upward mobility anymore, because how much capital you have by the time singularity comes around is the direct determinant of how successful you will be in the future. And that is the escape from the permanent underclass is we have two years.
- [94:05]
B
By the way, a lot of the world, including probably a bunch of Europe, has believed this a lot longer than
- [94:08]
A
AI for sure, for sure, for sure. I mean this is like this Marxist theory. I don't know. But my point is I worry a lot about AI safety, but I also think humans are extremely resilient and we're talking about something that is unknowable. That is the definition of the singularity. But I don't know. I'll take that out of the equation for a second because that's another topic. I think my main response is I actually really believe in market forces and creative destruction and the fact that there are always ways of building newer, better things. And I think broadly the thing I'm most worried about is social upheaval as a response to these things. I think we probably do a, a very big tech backlash and I think France was never so innovative than in the post war years. And it's very hard to think that it did not take World War II and the Shoah and all these things to actually get us to a place where we could do this. So I think I'm basically macro, very optimistic about the future micro. I think the path to getting and to continuing is going to be a little rough. And I'm very worried about how polarized things are and. How much stock people are putting in. Things that I think are like just everything has become a litmus test for just. It's the definition of polarization. But I worry that the path to getting there is going to be a little bloody.
- [95:41]
B
Is there anything we can do about it?
- [95:42]
A
The point is to say, I think being intellectually consistent and nuanced across both sides because nothing is ever as bad or as good as it seems, is the only way I can find to deal with it. And so if you're utterly convinced about something and you can't be spoken out of it, I would treat that with extreme caution because it probably means you're wrong.
- [96:02]
B
Thanks for bearing with me on the existential question.
- [96:06]
A
Sorry, I'll just come back. I'm trying to think about this, thinking about this for my kids. I have very young kids and I think the main thing I'm trying to do is make sure they are gentic. Like make sure that they feel like they have the power to change things around them. And again, I've grown up with so much privilege that it's easy for me to think so. But the point is that remembering that the folks who are making the rules are no smarter than you are is, I think, such a powerful reminder that you should try and be an actor rather than a victim of events.
- [96:36]
B
How do you teach that?
- [96:38]
A
I think through small things. I think it's like my son is super into the Eiffel Tower and we made a bet that I lost and so we're going to go to the summit of the Eiffel Tower. It's like just showing. I honestly forgot what the exact bet was. But the point is showing him that I take him seriously as a kid and that basically I'm putting stuff on the line in the relationship. In this case, we get to go to the Eiffel Tower together. It's not a terrible loss, but I think the way to prove it is to have him question things and not to say have him question things and have him be an agent. So if he wants to ask someone something on the street, he has to go and ask someone something on the street.
- [97:17]
B
There is a low hanging fruit joke about how you're teaching your kid to be a wild gambler.
- [97:22]
A
True, true, true, true. In fact, I bet him yesterday a watch that he would not come wake me up in the middle of the night. And after two failed bets, it is now worth. That's a bribe. So he. Yeah, he had to earn it. That's funny. Ish. But so I think in that sense, the thing that maybe pisses me off most with tech is Galman amnesia. Galman amnesia. As you read a newspaper column, you think, oh, I know about this field and this is all bullshit. And then you read another column that you don't know about and you take it as gospel. And I think tech tends to forget this. And I think the way in which extraordinary tech leaders then think they know everything about everything is a little odd. And so I think just kind of remaining nuanced and humble, but also knowing that you can shape things is like the way at least I'm trying to behave.
- [98:14]
B
You wrote a founder variation of the Proust questionnaire back on Medium. What makes Asta wonderful and great in her own right and what makes her such a good compliment to you?
- [98:33]
A
It's a great question. I think in her own right. The meme I really like is Paul Graham's formidable founders. I think she's formidable. And I think it comes in two different ways, which is she is extraordinarily self confident and extraordinarily good at what she does. She is the epitome of competence. She's just very good and she's very self confident and so she doesn't waste time, but she's also very low ego and so accordingly she doesn't waste time thinking about the meta of it all or anything beyond what is the most effective path and who is the best person to do it. And that makes her just such an amazing person to work with and leader at the company because she has very strong opinions or she has very weak opinions and she knows exactly which is which. And she's very clear in how she states both. And so I think like Asta is probably the toughest, say manager that I've ever seen operate, but that's because she has maybe I don't know how good this is, but she has the slight tiz of being very no holds barred and straight with people but in a way that really pushes them. So I think she's in that sense really, really formidable. And I think she's a very good fit to me because hopefully me to her, but because I think we're very different in the areas where we spike and we have been able to build implicit trust so that we know where to lean on the other. And so I think basically there's probably some sanctity to our relationship at this point that goes beyond any given question or outcome. It's interesting because again, I started working with Z as a really close friend from college and we spent so much time assuming that our personal relationship was all you needed to be business partners. And I think with Asta, I didn't know her at all before starting privy. And so we came in and for me it was am I finding the right business partner? And it is costless if the answer is no, because then I don't have to see this person again that I, I have spent these last 20, whatever years not seeing. And so we went in very intentionally and there was for the first year a question that we kept asking each other which is when is the honeymoon over? When will the shoe drop? And we'll find out all the ways in which we suck. And there's enough mutual respect and admiration basically that the honeymoon hasn't ended. I can say all the ways in which she's not good at what she does and she can do the same for me. But we have found a way to build that trust and buttress each other in really good ways and there just isn't that Much ego. When we work together, it is about the work. And I think any relationship where when you work with someone it is about you or it is about them is a telltale sign that the fit is in there.
- [101:28]
B
Yeah. When you say different spikes is that domain expertise is that way of working with people, is that.
- [101:36]
A
I think it's domain expertise. I think it's, it's, it's things we value and respecting it. Like, for example, I'm a pretty atrocious manager. She's a very good manager in my opinion. And we know that about each other. I think I'm a very intuitive decision maker and she's a very rational decision maker. And we know this about each other. And so I think it's in the ways that we operate. And then obviously spikes. I understand parts of the crypto market exceedingly well, but I can't hold a candle to how she operates technically. And the way she's been able to scale the system in a huge way, the way Privy has scaled on a relative basis, kept four nines uptime throughout our existence, basically through periods like Fronttech where we had week on week, like 40x of traffic for multiple weeks on end is so impressive. So it's stuff like that.
- [102:26]
B
You have this three headed dragon of very impressive people at Stripe, Patrick and John and then Will Gabriel who's maybe a little less known. What have you learned from those guys?
- [102:39]
A
So I'm really impressed and I really enjoy working with them, the three of them, honestly, a lot of them is operating as a trump and the way in which they operate together is pretty wild and impressive to see. The force multiplication that you have is really great. And I think I've probably learned a few things. One is everything at Privy feels really personal to me in a way where any person leaving anything happening that's bad feels like such an indictment on my work in a way that they've been at it for decades. And so there is a sense of steadiness from them. And so they've been able to basically fuel ambition and get more and more ambitious over time, but also be steady in their vision and their execution and be resourceful in that way. And so that's one thing that I think I'm learning in real time by just observing them. And it's something that is happening. I'm very grateful because I get to basically scale Privy even though we're part of Stripe in that way. Two is, I think, balancing short term kinetics with long term structure and building it right from the start. I actually think Privy is quite good at that generally, but that's another area where there's a lot to be shared. And then maybe, I think maybe the third and fourth third is giving room for whims. Whims, yeah. Again, I look at Stripe Press, I look at climate, I look at these things that need not exist and yet I think both Stripe and the world are better off for them existing. And I think you kind of have to. My sense was always you kind of have to earn that like easy to do once you've got a very successful business that is like pumping out cash,
- [104:35]
B
but it's Google hiding its own.
- [104:37]
A
Exactly, exactly. But I think finding stage appropriate ways of having that exist is really, really important. And then maybe the third thing, and this is the thing I have most to learn on is leverage. They're running a massive org. And yet when I talk to Will, I feel like he is so deep in the weeds. And basically the breadth and depth means he has learned to scale himself endlessly, whereby he is able to be very high context and precise, but he also knows when to delegate and how. And so I think there's something to leverage and cultivation of talent that they're doing that I really want to learn from.
- [105:20]
B
What have you learned from Chloe?
- [105:23]
A
Oh, many things. I think Chloe's interesting because she's endlessly pragmatic and she's both very intellectual and drawn to all the things that we've talked about. But she has very low patience for conversations that spin and don't lead to outcomes. And so I think basically some versions of proper impatience with. All right, should I get off the pot? Where are we driving here and what is the thing that we want to get to one. And then I think she is also just much more civically minded and thoughtful about bigger societal ramifications. She reads economics books for funds and in that sense she is much more collison like than I am. But I think thinking about what are repercussions of what we're building and broadly the interplay between private and public is something that she's taught me a lot of beyond the infinity. Many things that we've learned in 20 years of pure life,
- [106:28]
B
just a few final things. You worked on the declassification engine Columbia.
- [106:33]
A
Yes, this is.
- [106:33]
B
Can you tell us a little bit about what was interesting about that?
- [106:36]
A
It was very interesting and it's such, you know, I worked on two things that are now that were AI related and that are now so quaint, which is so interesting. But one was a system for training machine learning models at this time it was. It was Markov chains. So the states, I guess where we were at in the NLP terms. But it was training it on Yelp data to try and work with the US Department of Food. I forget what it's called. Sorry. New York Department of Health and Safety, something like that. Basically food health. And the point was to look at Yelp reviews to try and figure out what restaurant kitchens should get inspected because they're making people sick.
- [107:13]
B
Wow.
- [107:14]
A
And the interesting problems were sick can mean very different things, depending on intonation and sentence. And so we have to train the models in interesting ways. Again, 20 minutes of vibe coded agent tick work today, which was like a research project. So the space moves. The declassification project was a similar thing, which was can you actually take redacted government papers and try to declassify them based on basically interpolation from other papers that have been declassified? Can you basically correlate data across different FOIA requests to try and understand what's going on behind the scenes? And take a best guess. This is how I remember it in any event, as to what's happening under the hood. And so it was both a data structuring problem. There's just the data, the classified data is a mess. And so how do you actually look through the full archives to reconcile things that maybe have already been declassified here but not there? And it's an inference problem of Cuba's Fidel. There's probably one word that's going to come after that one and it's pretty easy to tell what it is. But so how do you actually infer the likely piece behind it? It begs an interesting question, which is things are overly classified in many ways and so how do you get to a better standard? And again, I actually think this is back to how do systems evolve? You could spend some time coming up with a better classification system or you basically introduce a new process of declassification. And this is where obviously bureaucracies kind of go wrong, both in private and public sector, which is you create a solution to fix a problem when you should have just fixed the problem. In a sense, the solution. Yeah.
- [108:57]
B
In what way are you most French?
- [109:02]
A
I don't know. I think probably some amount of innate skepticism around vision and things like that. There's a very jesuistic mindset to not getting overly sold on these things. I think probably whilst I'm extremely capitalistic at heart and believe in the importance of, call it private enterprise in moving the world forward, I know what great government services looks like and the extent to which I've missed them in many a place in the US and then probably like there's a lot of stuff around food like bread sucks here. There's the sanctity of having a great meal. Probably is some version of it as well. But yeah, stuff like that.
- [109:53]
B
What's the best French film, or at least your favorite?
- [110:00]
A
Such a hard question. I've got. Impossible question. I'll give three that are very different parts of the spectrum. Even better, three that are different parts of the spectrum. The very wonky cult classic equivalent of scary movie, but with a French humor is called the City of Fear, La Cite l', Affaire, which is an old. I don't think it would be funny to anybody. Not French. And I don't know if it would be funny to anybody. Gen Z.
- [110:26]
B
Got it.
- [110:26]
A
But that is an open question. And that is like a very quintessentially French movie. There's old movie called the Rule of the Game. The Rules of the Game, Yeah.
- [110:38]
B
Sounds right.
- [110:39]
A
That is a great French classic. And then probably the third would be more recent. There's a movie called Heartbreaker with Roman Dulis and Vanessa Paradis, which is a very good French rom com about a guy who gets hired anyways to woo a woman. It's like the exact right type of French rom com that kind of works. And I'm assuming it has either already been adapted or will be adapted in the US Great.
- [111:10]
B
Can you tell us about your watch?
- [111:11]
A
Yeah. So I'm the descendant of a very famous. No, this is a watch I find really cool. Sean McGuire is the last Sequoia name drop of the. Of the session, but is the investor that we worked very closely with, Josephine Chen at Sequoia, a really good friend and had this watch that was like awesome and quirky and I lose shit all the time. And so this is the watch that I've actually not lost in a long time. But that is, you know, just something I like.
- [111:47]
B
My last question. What? You are very principled. Despite all the pragmatism. What do you hope to never compromise on?
- [112:01]
A
I think any place where I'm not saying what I think hurts me a little bit. And so that isn't to say I say relatively few things. I'm quite private. And I think being silent is, I think a very good default mode of operation. How do we save the world? People shut up a little bit more. That might be a really good place to start. And this is why part of this exercise obviously is ironic because I'm just spewing my thoughts on things, but I'm okay with not saying things. I think being asked to say something I don't believe is very, very hard for me, and so I think you probably see me get physically uncomfortable in those places. I think it is like espousing a worldview for convenience that isn't my own. Feels very hard and I hope not to change in that regard.
- [112:50]
B
Well, thank you for thinking out loud with me today. Nonetheless, that's all I got. Thanks again.
- [112:55]
A
Thank you.
- [112:56]
B
Thank you again for listening to Dialectic with Henry Stern. If you enjoyed it, please share it with a friend or give it five stars, a thumbs up like wherever you're watching or listening and subscribe or follow so you can get new episodes. I'd also like to thank Notion one more time for presenting Dialectic and making the show possible. As I mentioned at the top of the episode, the way Notion is working with AI both transformed the company from the inside out to be an AI native company, and for all of us to be able to work in our Notion workspaces with all our documents and databases and everything else, and use AI and agents that are natively integrated is deeply powerful. As I mentioned at the top of the episode, if you haven't checked out Notion's developer platform and what's possible with workers, it's still tremendously early. But I when I was recently in Washington D.C. with Notion Zone and Dialectics former guest Jeffrey Litt and he was telling me about the things that he is doing with workers. He's like got a rig where he basically just tells his agent like his rough schedule for the week is and what his goals are on nutrition, it builds a meal plan, it contacts Instacart and orders all his groceries. It's totally crazy. Things you maybe have seen people do with tools like OpenClaw is now possible thanks to Notion Workers, and again, it's integrated with the place where all of your work, all your documents and all your data actually already live. I think people have just scratched the surface and I'm excited to play around with it more, both in terms of my day to day life and how I prepare for and release Dialectic episodes. The combination of Notion AI, custom agents and workers have made it truly on the level with the most frontier AI tools that exist. The team is also growing fast, so if you're interested in working with them, you can reach out to me@workaxandall.com and I can hopefully route you to the right person if it's a fit. As always, you can learn more@notion.com Dialectic and thank you for listening. I will see you next time.