The Exchanges

Q Why should people in the general public care about finance? So it's like, so you're, you're doing, you're investing in all these things, this sounds like something banks used to do a lot more of, and now there's other people that are doing them other than just banks, and what, what's, what's changing here, and why does it matter?

A I actually think that technology finding its way into financial services is actually taking down costs for consumers. So if you think about the average American who used to overdraft many times a month and pay, I think, 15 to 25 dollars every time they overdrafted as, in addition to interest and return check fees, now you have a couple dollar a month subscription for a company called Dave that provides insurance that allows you to overdraft as much as you need And they obviously have some limits, but for the inadvertent 20, 30 dollar overdraft, they'll ensure for your small payment that you don't get any of those fees and you don't have the interest payments with the banks.

AI assessment note: “technology finding its way into financial services is actually taking down costs for consumers.”

Q What'd you, what'd you learn from the founders there? What'd you learn from the team?

A Um, that, you know, I was never thinking big enough. One of my favorite stories was, I remember I was helping my friend Brian Rakowski, who was, You know, this guy had first job out of school, out of Stanford, and he was given, you know, Gmail. That was his project, was to help Gmail out. And I remember sitting on the, the product review with him, because I was helping him out at that point. And I remember the Gmail team saying, oh, maybe we should like, um, maybe we should give people 10 times what, uh, they're being given today on Hotmail. Because remember, back in those days, Hotmail was the dominant email service, and so was Yahoo Mail. And they had 25 megabytes of storage. And you would sit there and go like, wow, it's, you know, not a lot. But, you know, back then it was, State of the art that you would get 25 miles, uh, 25 megabytes of cloud storage. And I remember the Gmail team sort of, you know, pitching an early version of this to, to Larry going like, you know, we think we can do 10 times and do a 250 megabytes, which sounded like, you know, a lot back then, right? You know, you're giving, you're getting a quarter of a gig. And Larry kind of sat there and was like, why are you wasting your time on like a 10 X improvement? That doesn't sound very much. And everybody's like, well, that's a lot. You're getting 10 X. He's like, let me start something, you know, sort of …

AI assessment note: “that, you know, I was never thinking big enough.”

Q You've ended up investing in a very large number of successful companies. Tell us, tell us a couple of those stories. What were some of your early wins that you had?

A Boy, you know, I'll probably tell you my favorite, which is this fantastic company with this very humble founder in Australia, Melanie, who started Canva. And Canva is today is YouTube for design. They're the reinvention of design, you know, every school Student, uh, is using it for their homework, and it's one of these things where I've never seen such incredible traction, incredible growth, and such a commitment to reinventing how people build beautiful things and make design simple, and so, you know, I think it's one of the most valuable, uh, companies run by a woman entrepreneur, uh, that's private today, and then, you know, I think it's one of the fastest growing private companies, uh, on the planet. They're based out of Sydney, but it's one of these companies where, you know, we were privileged enough to find Melanie and Cliff, her co-founder, uh, and, uh, lead their Series A, and it's been this amazing journey to sort of watch them Uh, you know, build, you know, the next generation tools for building, you know, beautiful things. Uh, whether it's business cards, websites, presentations, it's the, it's a, it's a tool that basically democratizes design for everybody.

AI assessment note: “I'll probably tell you my favorite, which is this fantastic company... Canva”

Q control of the usual special interest. Uh, you know, you guys worked together at sports charters in DC. You closed down feeling schools. Like where, where did that push come from? It really, I think it surprised a lot of people. And I, I, I was just really amazed by what you were doing there. It took a lot of courage and made a lot of people angry at you.

A No, I really appreciate it. And you, you say that to so many people when we're in private groups also. So, To me, like when I was hired as mayor, like I was the CEO of the city and, and as a CEO, your job is to take on the toughest problem. And I thought from day one, that if we didn't tackle education, that our city would never maximize its potential. Um, so, so we went right out of the gate to, to take over the school system to actually have the mayor in control of the way Richard Daley and New York, I mean, it's Chicago and mayor Bloomberg in New York had done. And so once I got control of the system, I wanted to find the best, most talented, Hard charging and visionary CEO I could find, and that, that was Michelle.

AI assessment note: “as a CEO, your job is to take on the toughest problem.”

Q These are really unhealthy. But isn't it, isn't it good for people to be striving to work harder and build things though as an economist?

A This is your point. This gets to your point. So you said, okay, so does that mean that, you know, money doesn't matter and none of it, no. Money per se doesn't matter, but what you're actually being compensated for matters. So Josef Schumpeter, the father of modern entrepreneurship, the understanding, the professor at Harvard, but, but he was a visionary. He said that entrepreneurs actually don't care about money per se More than the fact that it's a scorecard for what they're building. This is the key thing. And so for entrepreneurs who are being really successful, don't think that your satisfaction is actually coming from the money. It's not. Your satisfaction is actually coming from the prosperity that you're creating.

AI assessment note: “Your satisfaction is actually coming from the prosperity that you're creating.”

Q And they go like Mach five, right? Is that, is that, is this what I'm thinking of? Is a Mach five engine or how fast are they?

A Oh, that's the best part. The faster you go, the better they are, unlike most types of jet engines, because the faster you go, the more pressure they can develop and the higher your compression ratio of the engine is. So they actually start working. There are subsonic ramjets going back to World War II, but you really want to be going Mach three, four, five, six for them to really come into their own and beat other engines. And until now, there's never been really aircraft that could survive that regime of flight. And so ramjets were just inefficient. Impossible to start the support equipment required to get them into the air and going fast enough was too heavy. But very recently, I think ram jets have come back into this world where we do have composites that can handle vehicles going that fast. We have secondary propulsion systems, uh, like basically electric propulsion that allow you to do things like, uh, you, uh, I guess, a variable compressors that would allow you to use them from zero speed and then switch over to full RAM at high speed. This was actually what I was working on when I, when I was at Facebook, I had a side project. Me and a few guys were in Atherton, and we got kicked out of Atherton because we were making so much noise. But we were building a project called ASRAM, the assisted ramjet, and it was a ramjet that could take off from zero and then transition t…

AI assessment note: “you really want to be going Mach three, four, five, six”

Q How did that emerge? Was that a natural thing where a few of you on each side kind of got it and decided and made that choice? It was the personalities of the people involved, or was it actually the, the, the market forces? Like what actually brought that together?

A It was a combination. I mean, like, I think it was market driven. We weren't just doing this as like idealists who thought it was for the greater good, but like we knew that if everyone saw VR as something that made me sick or that was low quality or that was janky, that it would not be good for any of our products. And so, I mean, personality wise, like me at Oculus, like I at Oculus was a really big, you know, I pushed this really hard that we should, You'll be kind of helping other players in the industry. At Sony, I think it was, uh, Shui Yoshida-san, uh, at PlayStation, he was really big on doing the same thing. Like, we would go to trade shows, and we'd both have our closed doors demos running for investors and game developers, and we would always make sure to bring them over and show them exactly what we're doing, explain exactly why, and Talk through, talk through ideas where we could, uh, share technology. Same thing I think with Valve, a lot of the guys that were at Valve, some of who ended up later coming to Oculus, which is its own crazy dramatic story, but, uh, you know, at the time it was really clear that we were all doing it for same business reasons. We weren't, we weren't communists. They're just like, Oh man, no need to run a business. We all just need to share and share the love. We all figured we would be.

AI assessment note: “It was a combination. I mean, like, I think it was market driven.”

Q Give me some, give us some stats on ghosts. Like what, how fast does this thing go? How long does it go for?

A Uh, so cruise at about 85 miles an hour. It can fly for up to two hours on a battery and up to six and a half hours on a hydrogen fuel cell, which clips on and places the battery. Uh, the really cool thing about it is like the airframe is incredible. We can carry a ton of payload. It can fly for every long time. It's also, uh, we call it ghost because it's basically inaudible and almost invisible. Uh, at a hundred meters, it is inaudible to the human ear. And because, because it's a helicopter, it's very, very skinny and very, very short. And we stack everything all in a line. When you look at it dead on, it's like, it's like this big, like this is the front of the thing. So try looking at a square and it's like, you know, three inches tall, three inches wide, a hundred meters away. Basically invisible. And when you get a few hundred meters, like let's say 300 meters, you, you are invisible.

AI assessment note: “cruise at about 85 miles an hour. It can fly for up to two hours”

Q It's interesting though, because you, it's rather than join a big company next, You decided to start your own company, but I guess you'd be gotten big very quickly. I guess you might be able to say, so you still, you think you have those advantages again right now?

A Oh yeah. I mean, going into my, going, going into this as a, going into this as my second company, it's very different. You know, I said before that Oculus, I started it not because I thought it'd be a successful business, but because I thought it'd be important for the long run and a lot of fun to work on. Um, I had no idea it was going to be such a case of right place, right time with Anderall. It was very different. I said, this company, I am going to grow it to be large and influential very quickly. There's a huge market distortion that I can take advantage of. The companies that are doing defense work are totally incompetent when it comes to machine learning, AI, computer vision, and the companies that are good at those things refuse to work with DOD. And from the beginning, I mean, in our first pitch deck, it said, we are going to become a company that saves taxpayers hundreds of billions of dollars a year and makes tens of billions of dollars a year. And I'd say like that, that was a big goal of the company from the start, because if we don't grow this to be a many billion dollar company, it means that we're just another lark. It means we're just another supplier into the system. We didn't actually change the way.

AI assessment note: “Oh yeah. I mean, going into this as my second company, it's very different.”

Q much more sophisticated. They can memorize paragraphs. They're these hard things we had to do. They were harder. Then they gave you more skills to do things. And it seems like nowadays we've gotten rid of the stuff that's hard in order to get there faster, but then you kind of lose some of the depth basically. Is that, is that, is that an analogy for this kind of interface?

A You're right. And I think businesses are doing what people are naturally wired to want, which is the fast reward, the thing where you don't have to put in the time where you get that quick, you know, the quick dopamine hit, not the longterm return on productivity. And so where, where this ties into neuroplasticity, I think, is that there is enormous potential for things like brain to computer interfaces or peripheral nervous system interfaces that require an enormous amount of training to be able to use. I think people are thinking about these things the wrong way. They're thinking, I want to have a brain to computer interface where I put it on my brain and without any training, I just naturally think things. I think a word and it does it, or I think something out and it types it. I think the most powerful uses of these types of, uh, Human and computer interfaces are going to be ones that do require many hours of training. You know, imagine if I could, let's say, wear a band that is detecting a lot of the, a lot of the, Electrical activity in my median nerve cord in my arm. If I can learn to receive and send data along that cord in an arbitrary way after doing a hundred hours of association exercises, I can be hundreds of times faster than going through the traditional, you know, gesture movements or, uh, you know, schemes that people have come up with. And I think that's going…

AI assessment note: “You're right. And I think businesses are doing what people are naturally wired to want”

Q What do you encourage? Are you encouraging people to build them at their homes? Are they encouraging people to contribute to them? What's the-

A Yeah. So, so anyone listening to this podcast should go to million-gardens.org and donate 10 dollars, 20 dollars towards Getting gardens into the homes of, uh, in underserved families across America and, and Canada. Uh, we're hoping to do Mexico as well. Um, and eventually we'll go international, but that's where we're focused for now. So go to million-gardens.org. And, um, uh, it's a beautiful, uh, program. We're going to do potatoes in the spring. So when you, when you harvest potatoes, it's like digging for gold. So it's, we think it's the most fun one. For us to do in these, these sort of little green gardens. And, uh, and then we're going to do tomatoes in June. We'll do that around the, the summer.

AI assessment note: “anyone listening to this podcast should go to million-gardens.org and donate”

Q Was it, was this, was this normal that like, were a lot of your friends doing these jobs too, where you were from? Was that like a thing that a lot of kids were doing?

A Yeah. I mean, once you turn 13, you can get a official, like a work permit and, and, and, and, Most of my friends all had jobs. Um, and we worked and did athletics and sports and, um, and then went to school. Um, and so I would usually get home when I was in high school, I would, I would go to football practice. And then after football practice, I got a job at a, there was a butcher nearby. Um, and I would go and help skin deer, um, during hunting season. And then I would get home at about like 10 o'clock at night, do my schoolwork, and then turn around and go to school at six in the morning.

AI assessment note: “Most of my friends all had jobs.”

Q Product engineers back home. And it's, it's interesting because nowadays with AI, I guess it's easier to, to be more across the spectrum or, or do you still, or do you still have the really strong distinctions though?

A I think actually, so we still have the really strong distinction because it's like, are you accountable to your customer succeeding? Are you accountable to the product succeeding? Obviously you want both of these things to work. And so there's a question of, do you think you're going to get them to work by collapsing it to a single person who's accountable for both? Or do you think you're going to get to work by wiring up an elegant tension that you manage between the two? I think the thing that most people don't understand about this is, um, it is an unstable equilibrium. So if you were going to pursue a strategy like this, it actually requires a dictator to sit between these two organizations to manage that tension, to make sure it's a productive tension, it doesn't get counterproductive, and to, to make the strategic calls along the way.

AI assessment note: “we still have the really strong distinction because it's like, are you accountable”

Q decay in America, and one of the few bright spots really has been your leadership in restoring our judicial system to its constitutional role. For our audience who don't follow the courts closely, can you start with the history and context? How do we get here with the courts? Did some of the current problems start in the 19 sixties with the Warren court? Like, what's happened over our lifetimes?

A Sure. Well, really, Joe, this goes back to the 19 twenties or 19 thirties, when you had a bunch of, um, pointy-headed intellectuals in the law schools that said, uh, look, law really has no fixed meaning, no objective meaning. Uh, it, it's kind of whatever you ate for lunch. Uh, and that sort of slowly morphed into the 19 sixties era, where the left began to use the law as a really serious driver for social change. And you know, the Warren court played right into that, uh, and began to use law as a political tool. So that's really where it all began, Joe. And the Federalist Society in the early 19 eighties was meant to be a response to that, was meant to be a way to bring the law back to what it was, which is an objective, knowable, um, system, uh, that allowed people to govern their own affairs, Uh, and, and to create, uh, accountability and transparency inside of government.

AI assessment note: “this goes back to the 19 twenties or 19 thirties”

Q We need some kind of asylum for them, not a jail, right?

A Well, here's the problem. So because of the asylum word and the stigma associated with that, remember that most of the people we're talking about in this situation are low income, therefore it's all Medicaid. And as so often in government, you ought to follow the money. So local governments plan provision based on, on their reimbursements from federal Medicaid for this mental health treatment. When Medicaid was set up in 1965, There was a rule put in place because there was public outcry against insane asylums and all of that, right, that says we don't want these big institutions. So there's a rule called the IMD rule, Institutions for Mental Disease, which says that you cannot, states cannot get Medicaid reimbursement for any facility offering mental health care with more than 16 beds, very precise number, which is tiny. So you end up with these very inefficient Outpatient kind of services instead of modern, effective mental health treatment.

AI assessment note: “So there's a rule called the IMD rule, Institutions for Mental Disease”

Q impact. Obviously the ADL has been something that's important to our people. We're both Jewish. It's something that's been important to our people for a very long time. We'll talk more about it cause it has a very complex history, but maybe tell us a little bit about. What the ADL did in the past and why it was important. Like what, what are some of the successes it had?

A Sure. Look, ADL was founded in 1913 around the Leo Frank trial. This was a Jewish guy who was wrongfully accused of a crime, ultimately sentenced to death. There was exculpatory evidence and he was, uh, his, his death sentence was reduced to life imprisonment and the mob was so enraged they tore him from his jail and they hung him from a tree. And the wake of his trial and death led to really the creation and the acceleration of ADL. So when it was founded in 1913, the mission statement that they wrote was that its purpose would be to quote, stop the defamation of the Jewish people and secure justice and fair treatment to all. So ADL has always thought we have to defend the Jews, and we also, we do that by defending others. When everyone is safe, the Jews will be safe.

AI assessment note: “ADL was founded in 1913 around the Leo Frank trial.”

Q And, and have you, have you tested this out yet?

A Well, that's the cool thing. This has been built and done before. So everything I'm talking about, the physics of how this works has actually been proven. So, uh, back in the 19 sixties, they built a reactor, uh, it called EBR two, not the best name, but EBR two in Idaho. It ran for 30 years and it demonstrated these amazing safety characteristics. In fact, on April third, 1986, they put it through these phenomenal. Like basically full scale safety system tests where they challenged this thing in, in crazy difficult ways. They ran it at full power. They locked the control rods out, the things that go in and shut it down. They locked them out of the reactor so they couldn't go in and then they shut off all the cooling to the system.

AI assessment note: “This has been built and done before. So everything I'm talking about... has actually been proven.”

Q So how'd you and Michael connect the side of buildings together? Like, he has a pretty unexpected background of his own, like firefighting and...

A Yeah, yeah, yeah. So, so Michael and I worked together at, at a previous company where I was the VP of product. He was the tech lead. He was the best engineer at the company. And, uh, interestingly, like, after we, you know, he, he left that company, Uh, and he had, he had done the fire academy when he was younger. He has really interesting background where he had gotten really passionate about that. And so after that company, he went off to be a professional firefighter. Um, yeah. And so he, he did that, uh, uh, uh, I think it was, it was during, I think COVID during the worst of the fires. So he was doing wildfires.

AI assessment note: “Michael and I worked together at, at a previous company where I was the VP”

Q like they'd want, like, the people to have the relationship where they're hearing from them, because they want a relationship with a person, because that's a really important, valuable relationship for us. But then in other cases, there's probably lots of people we're not getting that we could be getting. What parts should people do? What parts should a computer do? How do you, how do you think about this?

A Yeah, yeah, it's, um, what, what people say, people often ask me to, like, Judge my credibility. They're like, when do I not use AI moderated, right? It's, it's like, ah, they're testing me, like, when a buyer will ask that. The answer is, like, if you are, want to have a relationship with that person, then have a relationship with that person, right? Don't, like, put AI between you and somebody you want to have a relationship with. So, like, a hundred percent. But the question is, like, what is the, kind of, cadence at which you want to have a relationship, and what is the cadence at which you want to understand more? So, I would, like, people are all the time mixing. I'm gonna have a couple of interviews with people who are, They'll call it, like, uh, red carpet research, right, when you're really talking to somebody with deep relationship, and then the rest of the year, you're sending them AI interviews, right?

AI assessment note: “Don't, like, put AI between you and somebody you want to have a relationship with.”

Q And you said, I think there's, there's three core pillars, I guess you say, There's principles, partnership, and policy. Like what, what does that mean?

A Yeah, so we started with the declaration in December, and that declaration was really an articulation of our first principles. Principles that we share, you know, what are the, the core, the North Star that we want to work towards, and, um, and the things that are really going to bind us together. And so obviously there's a lot of clauses in there about having a pro-innovation approach to regulation, Um, there's a lot on supply chain security and the, the protection of critical infrastructure. Um, then we want to basically, you know, take that, those principles and say, okay, how do we translate this into action? And so we decided to, uh, set up two work streams, one on policy and one on, on projects. Um, the one on projects are the products that I mentioned earlier, the idea of having, you know, three or four different, Key flagship project projects that, um, we want to productize and roll out to, you know, the rest of the Pax Ilica ecosystem. And then policy workflows are, you know, conversations that we have to have around the protection of an intellectual property, the protection of sensitive technology, critical infrastructure, and the promotion of an AI friendly regulation agenda.

AI assessment note: “we started with the declaration in December, and that declaration was really an articulation”

Q So what's the Trump administration's approach? You're helping these places industrialize as well. What does, what does that mean?

A So I think it's projects that move the needle. So we think about it. Look, we're the largest energy exporter in the world, the largest producer of oil, and the largest producer of natural gas. So our RIOTH kind of lays the path of what the key sectors to invest in are. Energy, critical minerals, and technology. We have the right to invest anywhere in the world. No restriction there. To block choke points from China and Belt and Road, but also to move the needle for others. And, you know, but take energy as a continuum. You go somewhere, you want to build a port where US LNG can be exported to. We need a re-gas plant there to send it to the country. Maybe we need a critical mineral, lithium, cobalt, or something in that country, but mining is incredibly power intensive. That's where that LNG, or now, that natural gas comes in. And along the way, you can also supply consumer power to villages, to small businesses, large businesses, and you grow an economy.

AI assessment note: “key sectors to invest in are. Energy, critical minerals, and technology.”

Q the, by the late eighties, he was making like, I think it was a billion dollars over a four year period or, or maybe a billion, some huge amount of money relative to everyone else. Did that put a target on his back? Like why, why did they want to take down Michael Milken? What, what was, what was the, what was the energy of people who were against him?

A Yeah, I'm not so sure because he personally was making that much money because I don't know if people really realize that at the time, but Drexel was making a lot of money. If Mike was making that much money, Mike's deal that he'd made from Drexel at a time when it was a very small market was that something like 30, 30 some odd percent of the profits of his department would go to the bonus pool he could allocate. So if he was making that kind of money, Drexel was making over three times that much money. Yeah. So here they were basically taking market share. They were a disruptor. If you use today's term, they were a disruptor. Um, you know, a lot of people don't like Elon Musk as he's a disruptor. People didn't like Bill Gates at Microsoft because they were a disruptor. They, you know, people were after, uh, you know, number, you can note any number of technology companies. And so number one, he was taking market share. Number two, because he developed a market and people believe in what he was doing, he was able to finance people Who wanted to acquire other companies that they thought were undervalued in the market. So when he would start financing people that would do what they would call a hostile takeover, um, the companies that were established companies, companies like you, you know, California that had tremendous political clout were wondering, Hey, wait a minute, how co…

AI assessment note: “So yes, there was a huge target on their back.”

Q In February of 2020, President Trump actually granted, of course, Michael Milken a full pardon. What did that mean to you and to the Milken family?

A It meant a really a lot, Joe, in that Once Mike pled, and he knew this, that he pled, he now was going to go through the rest of his life with a cloud over him that he is a convicted felon, ok? And he, the only way to get any kind of acknowledgement that maybe he didn't deserve it, or he's not that bad, or he's done good things in his life, would be a presidential party. There's really no other way of doing it. I have a whole chapter in the book devoted to, we discussed it with the Clinton administration, we discussed it with the Bush administration, um, and, you know, and the pardon process and, and how you go through it. It, it, it met the world. Not only did President Trump grant him a pardon, but when he called Mike to tell him that he was getting a pardon, he said, you know, and this is a real pardon, this, you've never done anything for me. Okay. Um, you know, there's no quo here. It's a pardon. And then the White House issued a press release that went in great detail of the fact of what Mike has done in his life for, in the world of finance, for companies to get capital that couldn't get capital, what Mike had done in philanthropy and cancer research to save lives, and the fact that even in the press release it talked about that what Mike pled to were all novel, as I think the term they used, you know, novel interpretations of the law. So, you know, it meant, it meant a …

AI assessment note: “It meant a really a lot, Joe”

Q as deep as, as the PhDs, but I go decently deep. But now with AI, everyone could just pretend, right? And by the way, it's useful for me too. Cause I'm like, oh, how does that, how does that really nuanced deep tech thing work again? And you just like ask it. And then all of a sudden you're like asking really good questions. It's a little dangerous right now.

A It is. Yeah. I think like AI as a technical co-pilot for diligence is actually one of the most interesting use cases for AI. As an investor right now. I don't think it's a replacement for human experts necessarily, but it certainly is a huge help. If you're, if you come across a really interesting company in a field that you know a little bit about, but you're not an expert in, you can get really, really deep within the matter of hours, and then pull in kind of human experts to help you to go the last mile. Whereas, you know, five years ago, it would have been this kind of mad scramble, kind of calling people, trying to assemble folks that might have taken days or weeks. That can be done much faster now.

AI assessment note: “It is. Yeah. I think like AI as a technical co-pilot for diligence”

Q Like, cause it's all changing so quickly, even for me in the middle of it, probably as much as anyone, like, what is that worth? And it's like, you would know a lot better than me cause you have all the data. Like, how does that workflow work?

A Yeah, for sure. So I think it's a couple of things. I think one is, um, Getting all of the data in one place. And I think the way we think about it is that firms need a mechanism to collect and store all their information on their portfolio. As we said before, kind of in an apples to apples format, that's clean, that's auditable. Um, that's, that's kind of traceable back to where the data came from. Um, they need a way to access that data. We think about it as, you know, we have this deep commitment to interoperability at our company. So We think that when an investor has their data, it's their data. They should be able to use it, you know, with their own AI agents that they're bringing to the table or with software that they're building on top of standard metrics or in a spreadsheet. Or on our platform. Um, and then lastly is, uh, kind of robust tools that help them to leverage that data as well as the corpus of data that exists in an aggregated anonymity.

AI assessment note: “firms need a mechanism to collect and store all their information on their portfolio”

Q part of this in a lot of ways. And obviously everyone in society is part of it, but a lot of people are listening and they're not building things with AI right now. Like, how's it different for you? Like, tell us about this. Like you have a team, obviously you have a lot of technical people on your team. Like how is AI changing how your company works internally?

A Yeah. Uh, dramatically. Um, and I, and it really feels like the, the, The, the rate of change is increasing every day. The, the sort of the product releases, the conversations we're having, uh, it's super exciting and it's a lot to stay on top of. I see the first place where we saw it change the most on our team had to do with data operations. As I mentioned before, we sort of ended up building a human team to go and work with our customers and help them to process our data. And the first place we launched AI agents in our product was to assist the data operations team. Now AI is doing the vast majority of the work. So that's been a big, big, big shift.

AI assessment note: “the first place where we saw it change the most on our team had to do with data operations”

Q sasspocalypse was a big term last month, uh, where, I mean, this is obviously, this is a high growth sass company, but it is like, but also you're moving into AI and you're an AI company as well. Like, should people be afraid of the sasspocalypse? Like, which are the certain companies that you're actually, you're bearish on because of it, but then like, how do you think about it?

A Yeah. I think that the sasspocalypse likes most things. There is, there's a, there's a lot about it that's correct. And there's certain things that are probably overblown. Um, I think that the thing that's the most correct about the SaaSpocalypse is getting us to question all of our fundamental assumptions around competitive barriers to entry for software companies. Um, historically the number one, you know, competitive barrier to entry, um, for software companies has been switching costs. You buy software, you build a bunch of workflows on it, you get a bunch of users using it, you get a bunch of data into the system, and then it's very painful and expensive to switch to another platform. Because of AI, that's changing. Uh, it's cheaper for competitors to build software. It's cheaper for people to vibe code their own software. And, um, it's also easier to migrate data because of AI agents. So I think switching costs have gone down. And so I think it's right for the way that software companies, um, the way that software companies have been valued to change in light of that. It turns out though that there are other competitive barriers other than switching costs. So for example, for our business, one of the major competitive barriers that we have is around network effects.

AI assessment note: “there's a lot about it that's correct. And there's certain things that are probably overblown.”

Q Tell me a bit more about that. You guys have obviously deployed with a lot of the biggest companies in the world. What are some of the more impressive results you guys have seen in the wild?

A Yeah. So the earliest results that we, where we, where we saw and we were like, oh, there's, there's something here where they started with, um, basically like modernization programs. So people that had large legacy existing things, they needed to transform. And if you sort of did the math to scope out how long it would take, maybe it would be like a two year project. And, you know, relatively quickly by like late 20, 24, we were measuring You know, somewhere between a six to 12 x productivity gain for those types of projects. Meaning that, you know, one hour of human time spent managing Devin was worth like six to 12 hours of that human time doing the work themselves. Um, and so that was a big, that was like a big early result, and we started doing lots of, lots of engagements where our customers would use, would use Devin to just refactor, migrate, modernize these large systems. Now the interesting trend that we're seeing is, Um, a shift from really, uh, reactive to proactive engineering work. So, you know, if you think of the early days of the internet, most of the packets that were sent on the internet, um, it was like a human clicking a button or, or visiting a link or initiating some requests. And then at some point it, it totally flipped and now most of the packets are initiated by machines talking to other machines. And I think we're now seeing the sort of the flippenin…

AI assessment note: “we were measuring You know, somewhere between a six to 12 x productivity gain”

Q and making it talk to each other, ontology, the processes, and we had all sorts of, Different frameworks over time. There are conceptual frameworks we'd use when you go in. Do you guys have, like, your own conceptual frameworks for business value, and do you have something called ontologies? Like, like, not to, not to get the secret sauce, but other things like this you could, you could tell us.

A We, we really look at it from the perspective of the software development life cycle. So, we go inside an organization, um, they are, they have a way of doing things, right? Of, of developing software, starting from planning and deciding what they even want to write, to understanding all of their existing code and process, to then maybe scoping it out, and And maybe writing some code, testing it, fixing it when it's wrong, iterating on it, deploying it in production, monitoring it. You know, there's a pretty standardized software development lifecycle at this point. And what's happening is agents are just eating more and more of the cycle. And it kind of started with the writing of the code. And now we're like well past that, right? And so in fact, one of the more recent products we, we shipped is called Devon Review. It's because we observed that there's this totally new bottleneck in the software development, uh, in the software development lifecycle that wasn't the case previously, which is there's this abundance of code being written by AI now. How can humans even keep up with it all to understand what's going on? Again, we work with, you know, a lot of like regulated, large, complex organizations that are, they're running mission critical systems, and you can't just sort of vibe code, you know, your way and like YOLO merge, uh.

AI assessment note: “We, we really look at it from the perspective of the software development life cycle.”

Q When you, when you did that, where would you, when would you fire? Like you'd, you'd turn around and get them behind them or what?

A Oh, for sure. I mean, that's that the whole idea essentially is to, is basically to force an overshoot, to get the, ah, you know, to get the guy you're trying to beat, to spit out in front of you. And, um, you know, you can, you can do a turning fight where you're just continuously, you know, in, in a, in a high G and trying to rate and get the nose around to get Behind the other person, or, you know, you can get where it's a, you know, a little bit more artistic, and go up in the vertical, and you try and get, ah, you're just trying to get behind the guy. Um, and that's, you know, when you start getting really slow, people get afraid to bring their nose up. Um, I love it.

AI assessment note: “basically to force an overshoot, to get the, ah, you know, to get the guy”