Castelion Hypersonic Missiles and Crunchbase's AI Powered Predictions | E2088

This Week in Startups is brought to you by… Squarespace. Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www. Squarespace.com/TWIST LinkedIn Jobs. Post your first job for free at https://www.linkedin.com/twist AdQuick. Visit https://adquick.com/twist and mention TWIST to get $1000 off your first campaign. Today's show: Jason interviews Bryon Hargis, CEO of Castelion, about the United State's hypersonic technology gap with China, how we're catching up and what it will take to win the hypersonic missile race. Alex interviews Jager McConnell, CEO of Crunchbase, about their newly released, scary accurate, AI powered predictive models. Timestamps: (0:00) Episode teaser(1:14) Introduction of Byron Hargis and Castellion's mission(2:20) Significance of hypersonic munitions(5:55) Impact and challenges of hypersonic systems in defense(7:16) Castellion's development and cost of hypersonic weapons(9:39) Squarespace. Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www. Squarespace.com/TWIST(12:28) Shift in defense contracting and venture backing(18:25) Sales strategies and founder insights in defense startups(19:38) LinkedIn Jobs. Post your first job for free at https://www.linkedin.com/twist(21:10) China's manufacturing capabilities and defense paradigm shifts(24:37) Role of automation and robots in manufacturing(28:58) AdQuick. Visit https://adquick.com/twist and mention TWIST to get $1000 off your first campaign.(31:37) Crunchbase's transition to predictive insights(38:01) Crunchbase's technology stack and AI models(41:36) Financial implications of AI-first model for Crunchbase(48:43) AI integration challenges and financial performance(52:50) AI-driven predictions and their accuracy(54:24) Future prospects for Crunchbase: Acquisition vs. IPO(55:22) Upcoming guests: Vlad from Robinhood and Raul from Superhuman Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com Check out the TWIST500: https://www.twist500.com Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp Checkout Castelion: X: https://x.com/CastelionCorp Website: castelion.com Checkout Crunchbase: X: https://x.com/crunchbase Website: https://www.crunchbase.com/ Follow Bryon: X: https://x.com/hargsb LinkedIn: https://www.linkedin.com/in/cbh/ Follow Jager: X: https://x.com/jagermcconnell LinkedIn: https://www.linkedin.com/in/jager/ Follow Alex: X: https://x.com/alex LinkedIn: ⁠https://www.linkedin.com/in/alexwilhelm Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis Thank you to our partners: (9:39) Squarespace. Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www. Squarespace.com/TWIST(19:38) LinkedIn Jobs. Post your first job for free at https://www.linkedin.com/twist(28:58) AdQuick. Visit https://adquick.com/twist and mention TWIST to get $1000 off your first campaign. Great TWIST interviews: Will Guidara,Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta,Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland Check out Jason's suite of newsletters: https://substack.com/@calacanis Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com Subscribe to the Founder University Podcast: https://www.youtube.com/@founderuniversity1916

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  It's honestly sad. What quantity do you think is a large quantity of missiles to order in a year? Right now, and like in the US, if you're talking about a reasonable size, like long range missile, what is a large quantity?

• 0:13Jason Calacanis?

  Four to five figures, thousands to tens of thousands would be a large quantity of missiles.

• 0:19Speaker 00

  Way less.

• 0:20Jason Calacanis?

  Hundreds?

• 0:20Speaker 00

  Way less, way less.

• 0:22Jason Calacanis?

  Dozens?

• 0:23Speaker 00

  Our highest end systems, we're not at the capability of producing dozens per year, but that is like their ultimate eventual output when everything is working. Yes, it's.

• 0:33Jason Calacanis?

  Pathetic.

• 0:35Speaker 02

  This week in startups is brought to you by Squarespace. Turn your idea into a new website. Go to squarespace.com/twist for a free trial. When you're ready to launch, use offer code TWIST to save 10% off your first purchase of a website or domain. LinkedIn jobs. A business is only as strong as its people and every hire matters. Go to linkedin.com/twist to post your first job for free. Terms and conditions apply. And Adquick. If most of your advertising dollars are going to digital ads, it's time to diversify. Out of home advertising like billboards offer low cost, high value reach. Adquick makes it easy to plan, buy and measure all in one place. Visit adquick.com/twist and mention twist to get $1,000 off your first campaign.

• 1:14Jason Calacanis?

  Hey, everybody, welcome back to this week in startups. Very excited for our next guest. His name is Byron Hargis and he's got a startup in the defense tech space. It's called Castellion and welcome to the program, Brian.

• 1:29Speaker 00

  Hey, Jason, thanks for having me.

• 1:30Jason Calacanis?

  All right. I like the American flag back there. I love defense tech and our researchers were really excited when they found your startup to have you on the program. So we're really excited to have you here. Maybe tell the audience a little bit about what you're building.

• 1:47Speaker 00

  Absolutely. So Castellion is really focused on bringing back deterrence through strength. And if you kind of subscribe to the Reagan era piece through strength, like how do you, how do you maintain peace? You have to have actual strength against an adversary and adversary has to know that. So we're focused on building very affordable, long range strike munitions, specifically hypersonic munitions to get after kind of the current problem set that we see with having very limited options against pure adversaries such as China.

• 2:19Jason Calacanis?

  Let's take a moment here, just to level set with the audience. We hear hypersonics all the time. Now, I think we all assumed that inter ballistic missiles were going at a very high rate of speed. We see things come out of battleships and submarines. They look like they're going really fast, but you know, for civilians like ourselves, I think sometimes we don't understand what what hypersonic means.

• 2:45Speaker 00

  Let me, let me take that piece wise, because it is, it is kind of a confusing topic. So the hypersonic itself, like the kind of like general term is usually like associated with going five times the speed of sound or greater. So Mach five. The actual technical details of that are even more complicated. It's really that you're getting into compressible flows, you're in, you're not in equilibrium. So Navier-Stokes becomes much harder to actually compute when you're trying to figure things out. And generally, what it means is that if you don't, if you don't actually do testing, it's extremely hard to model a hypersonic system. And you are correct that when you look at things like ballistic missiles, or

• 3:26Speaker 00

  like a reentering space vehicle, they all do enter hypersonically. So like typically, like if you're at orbital speeds and you're reentering the atmosphere, you're coming in at least at Mach 25. And so ballistic missiles, you know, space capsules, everything is they're all start hypersonic when they come back into the atmosphere. Now, when you hear like the Department of Defense, or like a company like ours talk about hypersonics, it, we're really actually referring to a subset of that. And so very specifically, like a ballistic missile follows a ballistic trajectory. In other words, it's a very predictable controlled trajectory that usually goes through space. It launches from the surface, or what have you, goes into space and then reenters and comes back and it follows like the same arc

• 4:11Speaker 00

  that you would throw like a baseball. Hypersonic weapon systems, what what the DOD typically refers to, is really like when looking at how it flies, it typically flies a very different looking flattened trajectory. And so you're not leaving the atmosphere. Flying at a very prolonged distance horizontally at very high rates of speed. The reason that hypersonic systems are kind of like all talked about right now is really when you're trying to get after say like a peer adversary such as China, they put in tremendous investment to basically negate American capabilities regionally close in close into

• 4:52Speaker 00

  the coast of China. A hypersonic system fundamentally, when viewed at it from the from the other side, it doesn't look like a ballistic missile. And the reason that's important is traditionally ballistic missiles usually also imply that you might have a nuclear warhead on the front. And so you don't typically start launching ballistic missiles at other nuclear armed countries for fear that they might confuse what you are doing.

• 5:16Jason Calacanis?

  That's fascinating.

• 5:18Speaker 00

  Yeah.

• 5:18Jason Calacanis?

  Okay, so just to recap here, these things go well, five times the speed of sound. That's Mach five. Airplanes go under Mach one even, right? We fly below the speed of sound in commercial airlines, but there's boom, a new, a member of the twist 500, the supersonic passenger plane and the Concord obviously would break Mach one. But we're talking about five times the speed of the Concord. The second note is these things tend to fly closer to the ground. They're not interballistic. They don't go into outer space or to the upper atmosphere. When you go into the upper atmosphere, you have less air. So you can go a little bit faster. Yeah. So these things are fighting against wind and they're going five times

• 6:03Jason Calacanis?

  as fast. And I think the reason this is important, or the introduction of the this capability is so important, is that you can't defend against hypersonics, or it's incredibly hard to defend against hypersonics. In other words, the Iron Dome, if Hamas or, you know, whoever was dropping bombs on Israel, the Iron Dome wouldn't catch a hypersonic. Am I correct? That that's the reason this is so important?

• 6:29Speaker 00

  It's not impossible. It is much more difficult.

• 6:32Jason Calacanis?

  All right. And we have a video here. What are we seeing? Maybe you could sportscast this. Many people are listening.

• 6:37Speaker 00

  Absolutely. So this is some of the team. We develop a lot of the hardware in-house. Very, very uniquely in aerospace. Most of aerospace, like you typically hear the primes as integrators. We do actually a lot of the manufacturing of all the systems that are inside the missile ourselves. That's actually our marine MKR. It's a, it's a heavy duty truck that can basically pick up a shipping container, put it on the back. We, we made a launcher for that to make a mobile launcher to be able to do accelerated testing. Obviously has applications for the army and the marines, but here we're, we're actually testing a prototype of the, the upper stage of a hypersonic weapon system.

• 7:15Jason Calacanis?

  Got it. And so are, which piece of this puzzle are you building? Are you a provider to other people building the hypersonics? Are you building the full set or TBD? What's, what's the plan here?

• 7:26Speaker 00

  We're doing both. So very concretely, we are planning to provide full all up rounds, especially at the lower end of the cost scale. And, and I view that as absolutely necessary that as a country, we'd be able to do that. Cause one of the kind of key tenants of like American defense is we make the most exquisite systems, but they're very expensive and we tend not to have a lot of them. When you're looking at an adversary like China, that's put a lot of effort into manufacturing and their defense, defensive capabilities. We need to actually have sufficient quantity to actually deter them because they're, they're not going to be scared of having like a few very high end missiles.

• 8:05Jason Calacanis?

  What will these cost? Do you think, what do you think they're going to cost ballpark like for?

• 8:09Speaker 00

  For our like smallest weapon system, they'll probably be on the order of one third to one quarter. What a much less capable, but comparable in size system currently costs the US government. If you compare them like on a capability basis, like what could this weapon do versus like our weapon, it's probably one tenth the cost.

• 8:28Jason Calacanis?

  Okay. So I have a, I'm sorry, a very naive question because I didn't serve in the military and you know, I've been behind a desk here doing podcasts. Why didn't the military industrial complex make more affordable missiles? More affordable tanks, more affordable planes over the last couple of decades? Where we saw in consumer and business technology, a massive decrease in price. All we've seen on the other side is a massive increase in price. Now I'm sure that these providers were adding features. But we all know, you know, you can buy a smartphone today for but $100, $200, $300

• 9:12Jason Calacanis?

  in Android phone, obviously, and you could have capabilities that are literally, quite literally, 100x what somebody had but 20 years ago. I mean, it might even be 1,000x, you know, based on the camera and the whole, the whole set. I'm kind of getting at why economically, politically, systems, what, what's the issue here of this opportunity suddenly emerging to take 90% out of the cost structure?

• 9:40Jason Calacanis?

  Squarespace makes stunning professional websites ridiculously easy. It doesn't matter if you're just selling a product or a service. Maybe you're just sharing your ideas. Maybe you're an artist. Maybe you're a consultant. Squarespace is going to give you all the tools to make this happen. And as you know, Squarespace is always adding cutting edge features. This is why I've been using it for over a decade and Squarespace is actually the longest running partner here on this weekend service because every time I need a new feature, Squarespace adds it. And the new AI tools are unbelievable. They have one called Design Intelligence. It's like having a world-class designer sitting next to you in your office and you just answer a couple of questions and their AI builds

• 10:22Jason Calacanis?

  you a fully customized site that's perfect for your brand, that's unique to you, and it does it in just minutes. Personalized layouts, on-brand visuals, premium content, it's all ready to go. So start your year off strong at Squarespace.com/twist for a free trial. And when you're ready to launch, go to Squarespace.com/twist to get 10% off your first website or domain purchase. Squarespace.com/twist. Thank you to Squarespace for making such a great product that we use year after year at an affordable price and we really appreciate your partnership.

• 10:53Speaker 00

  Fundamentally, like the, the incentive structure that the traditional primes have worked under for a very long time and has generally been accepted by the government has not driven them to those types of improvements that you've seen in commercial industry. Like if you had, if you had a traditional prime, build your cell phone. Probably be over a hundred thousand dollars, maybe a million dollars. Only a select few people would have one and that's good because that's all they would be able to build. It really is like you got cost plus contracting. That's one aspect. There's basically no commercial penalty in terms of like taking a long time or it being expensive. Cause all of the, all of the expenses are paid for.

• 11:33Speaker 00

  You have essentially a kind of a peacetime posture where because these systems take a very long time to develop and they are expensive, like the government isn't buying any. And of course, as you know, if you don't, if you're never going to sell, but you know, a handful, of course they are going to end up being more expensive. So like you have, it's not just like, I don't want to just blame the primes because it's not a hundred percent on them. It's really just the kind of traditional aerospace, I'd say, market has been very backwards. In fact, like the space market, which I'm very familiar with was like this pre SpaceX on both launch, launch vehicles, satellites, you name it. It's this exact same problems. There's just been a new entrant that fixed it.

• 12:16Speaker 00

  And I think that in this market, a new entrant can also fix these problems. There's literally enough margin with different, different techniques and processes to bring that much cost out.

• 12:27Jason Calacanis?

  So if I'm summarizing you correctly, a paradigm shift has occurred. The old paradigm was this cost plus paradigm. And there was very little competition. There was competition, but it was competition from a handful of players, maybe, you know, count them on one hand. And their incentive was to give the government what they wanted, how they wanted it, not to look at it and say, from first principles, what's the best thing I could make for the lowest price? Whereas in Silicon Valley, we have a rabid competition with a large customer base. There's only one customer here, the United States. Am I understanding this paradigm shift correctly?

• 13:08Speaker 00

  Yes, there's absolutely a paradigm shift and I'm going to give you one more like very concrete example, like it's just one part of a huge problem, but like even with what you said, typically, yes, the government writes the requirements and they are buying to like the requirements, the system needs to do this. And you know, the major primes are building to that. If you're under a cost plus contract, it's not like, it's not like most folks working that are thinking like, just how can I pump up the price and you know, stiff the taxpayer. But fundamentally, if a customer asks for something and it's actually very difficult, like it hurts manufacturability, it causes, it's going to increase the cost greatly. If you're planning to sell these things under a firm fixed

• 13:49Speaker 00

  price model or commercially, you will fight to the death that this is a bad idea and you're going to go literally generate conflict between you and your customer, which is always like uncomfortable. But in a cost plus environment, why do that? I'll just agree to it. Oh, look, oh, the schedule is now longer, it's going to cost 3X as much. But this is what you wanted and I, I told you the consequence and you said, do it. But you don't fight. And I think like, that's just like a small example of like, what has been part of the problem.

• 14:21Jason Calacanis?

  It's kind of like, you know, your dad or your grandpa goes into the, I'm kind of the dad in this now, but you know, your grandpa, your dad, like they like to go to a certain restaurant, they like to eat like this New York strip steak, they get their whatever, New York strip steak, they like their martini, like the side of mashed potatoes. And the chef's like, hey, I want to do something a little more interesting here. And the customer's like, yeah, I like my steak and potatoes and my martini. Thank you very much. You kind of, you're kind of stuck there if you are the provider, because they told you what they want, you kind of hinted at something, maybe you want to try a seared tuna, they don't want it, they want the steak, that's what they always knew. And so you're kind of left with having to just build these things and then show them and then say, hey, we

• 15:04Jason Calacanis?

  built something, here's the capability, here's the cost, might this interest you? That's how the industry is kind of moving now, you think?

• 15:11Speaker 00

  It is, I mean, obviously, the space is now venture backable. Like, we're venture-backed. That's relatively new. Like, honestly, Andro's success unlocked just an enormous potential in the fact that now these types of companies are venture backable.

• 15:27Jason Calacanis?

  And they're standing on their shoulders. I assume you would agree of SpaceX, which showed, you know, hey, you can, you can have the government as a client and it will work out pretty well. And you can actually make things that kick ass that they didn't ask for, that they will eventually buy from you. And all my friends who are SpaceX venture capitalists, investors, founder fund, et cetera, they're feeling pretty good about the investment now. So it's a kind of a big unlock, yeah?

• 15:49Speaker 00

  Yeah, totally. And so like the SpaceX thesis, right, there's still like a commercial aspect to it. Andro's definitely more like defense only, but there was a very big software focus. We're down here, like, we're making hypersonic missiles. There's not a obvious, like, you know, what's our SaaS subscription software play? There isn't one. And that literally was uninvestable, probably like three years ago. And so like this market is now unlocking. And the reason I say that is like, if you are, if you're in this like traditional kind of market with selling to the Defense Department, everything you do, like if you don't have outside capital that you can go raise, you have to find funding for it.

• 16:27Jason Calacanis?

  How long will it take you to go from raising your first venture dollars, you know, outside investment, let's say not your own seed capital, but from first day for outside capital to first day revenue in from the government? What's the number of months or years, do you think?

• 16:44Speaker 00

  Oh, no, it was months. It was months in our case.

• 16:46Jason Calacanis?

  Wow. So you picked something that they really needed. So in six months, nine months, you had, you had money coming in from the government in the bank account?

• 16:53Speaker 00

  Correct.

• 16:54Jason Calacanis?

  How is that even possible? I thought the government moved slow and this took years. What, what happened? Explain it.

• 16:59Speaker 00

  I started my career as a, an engineer working in defense and aerospace and worked my way up into doing government sales, which is extremely esoteric.

• 17:11Speaker 00

  And really, the reason I did that was I actually wanted to be part of driving what are we working on? Because typically, as an engineer in aerospace, like you're told what you will work on. And so I found like being part of the sales process actually helped, you know, like pick what we are working on, what we're trying to solve.

• 17:28Jason Calacanis?

  This is like, there's two very important founder lessons here. And we like to always point them out when we're talking to founders. If you want to be the CEO eventually, the sales team, especially in the early days of a startup, they're the closest to the customer. And they really start to understand what the problem set is, what the needs are, what the willingness to pay is, what the quantity is. And you know, that, if you look at just that subset of tasks, it sounds like the CEO's job, it sounds like the founder's job. The sales job is the founder's job. Eventually, you know, you have to get a customer and that customer has to be delighted.

• 18:09Jason Calacanis?

  So this is actually a really interesting lesson here, I think, is don't look down on the sales job. If you are a young person and you can get into sales, you know what? I guarantee you the CEO is going to come talk to you at some point. They're going to be like, hey, how's customer X, Y, and Z doing?

• 18:26Speaker 00

  Yeah. I mean, to your point, it's not, it is sales and like, yes, you're trying to make sure that you're getting to a product that you can actually make money from. But sales fundamentally, before you make your sale, is really like product development. Like what, are we building the right thing? Is this what the feedback is? Is this what we should be building? Are we building something no one will buy? And everybody is already telling me there's no way in hell they're going to buy that. Like you are, you're the voice to the rest of the company. And if you're doing sales correctly, you are doing product work as well.

• 18:58Jason Calacanis?

  Yeah. And this changes over time. But this is such an astute point. If you look at your first couple of sales executives as product discovery and customer discovery individuals, then you can take in your mind half of their salary in comp and put it towards product development and half towards the sales process. And that will make it easier for you. Now, of course, once you have a product completed, well, then you're doing consultative sales. You're, you're saying, hey, here's what we have. What are your needs? And you're kind of matching it up and closing the sale. And then if you really have a great product like SpaceX does now, the third phase is you're picking up the phone and taking orders.

• 19:39Jason Calacanis?

  All right, we all know if you're a founder, or even if you're a small business, you're thinking about your company 24/7, 365 days a year. That's the life of a founder. This is not clock in, clock out, nine to five gig for you as the business owner. So when you're hiring, you want a partner that's as equally as committed as you are. And that's of course LinkedIn Jobs. LinkedIn Jobs is like your co-founder. They're going to make it so simple for you to post your jobs for free on LinkedIn, where there are 1 billion members. You're going to be able to share what you're posting and actually keep all the promising candidates organized in one place. And also LinkedIn is going to help you quickly write a job and get it in front of the right people,

• 20:22Jason Calacanis?

  whether you want to post for free or use some promotion to get it in front of even more qualified applicants. So do me a favor, don't take my word for it. I mean, you should, I know what I'm talking about. This is where I find my great people, but just understand that 72% of small businesses using LinkedIn said that it helped them find the best candidates. So find out why more than 2.5 million small businesses already use LinkedIn for hiring. So here's your call to action. Post your job for free. Why wouldn't you do it? It's free, F R E E. That's a good price. LinkedIn.com/TWIST. Once again, that's LinkedIn.com/TWIST to post your job for free. Terms and conditions do apply.

• 21:09Jason Calacanis?

  But this is so important because to our previous discussion about the paradigm shifting,

• 21:16Jason Calacanis?

  I think what I'm taking from this is a really important insight, which is you can build exactly, the sales team can get in there and build exactly what the customer wants. But then there's also room for creativity and the founder saying, I want to build something that they don't know they need. So I want to get to China. Everybody is scared about China's low-cost missiles, low-cost production and the velocity of their production. And the thesis I hear from my friends, you know, in the deep state and who are in and around it is, we have an adversary that can manufacture stuff at a fraction of the cost and at a multiple of the speed. What do you think is going to happen?

• 22:00Jason Calacanis?

  You look at the field in Ukraine and what that has wrought, which is basically they're not even fighting with industrial military complex items. They're zipping drones around with grenades strapped to them. This is a whole different warfare. So maybe your thoughts on where is China right now? How far is American manufacturing behind them? And how do we close the gap?

• 22:26Speaker 00

  Well, in the munitions space, when a lot of the defense manufacturing space, honestly, I my they're not ahead. China's not ahead in technology in most areas. Now in hypersonics, they are actually ahead. I would view them as ahead of the US, which is a weird place for the US to be. But in terms of manufacturing of those systems, they are generally doing quite well. It's very hard to win a fight if you are the first side to run out of munitions and weapons. And so like, kind of what you're seeing in Ukraine is like, I'm sure they would like to fight further distance from each other and push the other side back. But when you run out of those type of munitions, then you engage closer. And what you use to do that changes.

• 23:10Jason Calacanis?

  What's the path to winning if they're using quite literally slave labor in some of their factories with the Uighurs or absurdly low compensated wages that works six days a week, 12 hours a day? How does the United States compete with that?

• 23:30Speaker 00

  We don't have to compete on a dollar per dollar parity level. We're a very wealthy nation. We can't afford to pay more, but we do need to produce in a quantity and at a sustainable rate that allows you to, you know, basically, you know, atrophy their capabilities. Warfare is always like a back and forth. We do something, they do something to counter it. You have to do something to counter that. You need to just do that process faster than the other guys. And then you need to produce it at a quantity that overmatches anything they have. And that's where we're behind and what we are trying to fix.

• 24:05Jason Calacanis?

  So we need more factories and

• 24:07Speaker 00

  Absolutely.

• 24:09Jason Calacanis?

  It seems to me

• 24:09Speaker 00

  More factories. They don't have to get down to the same price as a slave labor, but you do need to produce the systems that can counter Chinese capabilities and at a quantity that matters. Now, if your system is much more capable than theirs, then you don't have to maybe make as many to counter those capabilities. So there's also like that factor that has to be looked at as well. So it doesn't even have to be exactly equal numbers, but you have to counter like their potential to hold you at risk.

• 24:36Jason Calacanis?

  The good news is we have slave labor coming to the United States. This is a little known fact, but we have unlimited slave labor coming in the form of robots. Figure. And I know that was like, you're like, where's he going with this one?

• 24:51Speaker 00

  I was like, oh dear God.

• 24:52Jason Calacanis?

  Dear God. Who's he enslaving? Robots. So automation of factories, obviously that's been a trend in our lifetime. But having actual robots who can fill in what the, you know, singular arm robots that are fixed, you know, into a conveyor belt and a production line, that's going to fill in the gap where maybe we're short on humans.

• 25:17Speaker 00

  It's honestly sad. What quantity do you think is a large quantity of missiles to order in a year? Right now, and like in the US, if you're talking about a reasonable size, like long range missile, what is a large quantity?

• 25:30Jason Calacanis?

  Four to five figures, thousands to tens of thousands would be a large quantity of missiles.

• 25:36Speaker 00

  Way less.

• 25:37Jason Calacanis?

  Hundreds?

• 25:37Speaker 00

  Way less, way less.

• 25:39Jason Calacanis?

  Dozens?

• 25:41Speaker 00

  Our highest end systems, we're not at the capability of producing dozens per year, but that is like their ultimate eventual output when everything is working. Yes.

• 25:50Jason Calacanis?

  It's pathetic. These missiles, the missiles you're building, all due respect. The complexity of building a missile and a Cybertruck. These don't seem. Yeah, I mean, I didn't want to say.

• 26:00Speaker 00

  It's a way less complicated problem than building a car. It's a way less complicated problem than building a satellite. So when you're talking about bringing in like a robotic workforce. We're so far away from like the scale that where that would actually pay back. Like we're talking about, let's just automate some of the tasks. Because you're still like a thousand is a very large order.

• 26:20Jason Calacanis?

  Well, since we're going there and we're going to be super candid here in this interview. I like having candid guests. Thank you so much for educating us.

• 26:29Jason Calacanis?

  Would. What is the skill level? I'm going to try to be delicate here. The skill level to put together a hypersonic missile. I'm sure you, let's say, if you need, you know, X number of people per missile, you know, to build a missile or to build 10 missiles a day or whatever it is. I mean, it seems like building 10 missiles in a factory should be a pretty easy task. That's only 3,000 missiles out of your factory a year. So I'm just picking something easy. Let's even up it 100 hypersonic missiles a day in a factory. That should be possible. Yeah, 30,000, 40,000 a year from a factory. How many of those people need to be engineers, college educated?

• 27:08Speaker 00

  Most of our manufacturing base is going to be technicians and they do need to be highly skilled in certain areas, like some of the, some of the things that we're doing are dangerous if they're not performed correctly, like when you're dealing with energetics. So you, you do have to have highly specialized training and skills. But it's something that, you know, you can teach to anyone. It doesn't have to be like, they don't have to have a four year degree to understand how to handle this process safely and what makes it unsafe. Like, don't do this. And really you need, you need, when you're looking at manufacturing at rate, like you're really trying to control your process and quality. And so those procedures and processes, that's not like the technician's job to develop it.

• 27:53Speaker 00

  But they're an integral part of giving you feedback of like, hey, this is working.

• 27:56Jason Calacanis?

  So college educated, associate's degree, down to even high school educated, able to think logically and, and be thoughtful for an eight, 10 hour shift is enough. They're going to get paid 30, 40, 50 bucks an hour? What do you think?

• 28:11Speaker 00

  I don't like, if you don't have a college degree, but you have the right attitude and the ability to learn. I don't care. Like that, who cares?

• 28:20Jason Calacanis?

  This makes me feel a lot better talking to you and really getting into, again, just like first principles, which is just a fancy way of, you know, basically building the model from the bottom up. Well, we wish you great success. If you were an engineer and you wanted to work at the firm, where would we send the amazing engineers or developers, smart people who want to help build the defenses of the country so that we can have more peace and prosperity?

• 28:46Speaker 00

  castelion.com and we're also on LinkedIn under Castelion Corp. So please do find us and submit a resume.

• 28:53Jason Calacanis?

  All right. Continued success and we appreciate the effort you're doing and we do appreciate you coming on the show. Founders, let's talk about building your brand here in the real world. We know digital ads are amazing, but if you want to stand out in 2025, maybe you need to think a little bit bigger. And a great way to do that is out of home advertising. Yes, you know, those really impressive billboards you see everywhere, or even the beautiful murals that everybody loves. This is how you build a brand. And Adquick makes out of home advertising easy, measurable, and fast. They've taken the headaches out of OOH. You've probably heard OOH, out of home, by combining ad tech precision with real world impact. So now, planning,

• 29:36Jason Calacanis?

  deploying, and tracking your campaign is as easy as running digital ads. With Adquick, they're going to hyper target your campaign in minutes. Then they're going to use their data and technology to ensure your ads reach the right audience every single time. Plus, they're going to give you real time insights to measure performance and optimize your spend. So whether you're building brand awareness or looking to drive customer action, right? You want to acquire customers. You can see why Adquick is trusted by everyone, from the fastest growing startups to the S and P 500. Take your brand to the next level with Adquick, the smarter way to do OOH. And just for Twist listeners, Adquick is waiving their fee on your first campaign. That's an amazing offer. Get started today at adquick.com/twist.

• 30:20Jason Calacanis?

  That's A D Q U I C K dot com slash twist.

• 30:24Speaker 03

  Hey everybody, we got an awesome Twist 500 interview coming up. It's with a company called Crunchbase. Now, I'm sure you've heard of Crunchbase. It's the well-known online database of all things private markets, startups, venture capitalists, funds, all of that fun stuff. I'm adding them to the Twist 500 because they recently released a suite of AI-powered features that I think are honestly pretty awesome. I love data. I love charts. I love analytics. This is my jam, and I think it's going to help the company grow and possibly have a big exit. So it's Twist 500 material. That said, I used to work there a couple of years back in the day. I helped set up the Crunchbase News team, a project that I'm still very, very proud of. But because I was an employee, I didn't have options, which I partially exercised.

• 31:07Speaker 03

  So now I actually own stock in the company. Normally, I try to avoid any sort of conflict of interest because I don't like them as a journalist. But as Jason says, no conflict, no interest. So I figured, let's just be honest, let's add them to the Twist 500. And so in that vein, I have Jager McConnell, the CEO of Crunchbase, and my former boss, Jagger. Hey, how you doing?

• 31:27Speaker 04

  How's it going? Good to see you.

• 31:29Speaker 03

  So I mean, first of all, it's been, it's been a little while. And since, you know, we were in touch, Crunchbase has watched the AI wave explode. And when I think about Crunchbase, this, you know, amazing repository of private market information and the importance in AI of having proprietary data, it seems like a match made in heaven. So can you just take us back to when Crunchbase was like, oh, we're gonna go in this direction?

• 31:56Speaker 04

  Sure. And I remember very clearly because I was using ChatGPT and I was like, oh, this is so cool. And everyone had a sort of like moment of like, this is going to change everything. And I was like, oh, this is going to change everything. And the challenge was, I said, well, we're great. We're a data company. And everyone was saying, oh, congratulations, data is the new oil, all of this. I was like, yeah, kinda. Except once our data goes into the LLM, it's gonna never come out again. And then the oil is being made somewhere else that we're not a company anymore. So it actually felt like an existential threat to me, where I said, well, we can't just rely On historical data now, like historical data, in fact, and this isn't just for Crunchbase.

• 32:38Speaker 04

  Every company in the world that relies on historical data or makes a business model off of it is now completely host, in my opinion. And the reason is because once it goes in, you're done. And the insights that AI is going to do on top of the the data is going to be way better than anything you're going to be able to do as a company. So what do you do about it? So that was where that was the wait, what do we do? And I was like, well, there is data that we have that no one else has. And can we use that somehow? Those are two things. It's every edit that's ever been made in Crunchbase over all time, right? You can't crawl your way there and find that. And then the other piece is, um, do you have our engagement data? Right?

• 33:19Speaker 04

  What can I do with engagement data anonymized to sort of figure out what's going to happen? Um, next in Crunchbase? So can you go and start?

• 33:28Speaker 03

  So historical data is looking back. It's knowing that Crunchbase raised $106 million, thought it's history. It's everything that's kind of in the rear view mirror. But your point is that Crunchbase knows so much about how people have interacted with that data that it provides almost like meta context on top of the raw information itself.

• 33:46Speaker 04

  That's right. But if you, if you go and look at a profile now, you'll see the history of the company. But what you don't see is what, how this profile changed over the last 17 years to get to where it is? And then how, how's traffic flow changed on that profile of time? Are there more investors or less investors looking at this profile now than it was before? Um, or new corporate people or recruiters, whatever happens to be? Or is the entrepreneur engaging with these profiles more or not? All of that is this engagement data and historical edit data. Um, that when we looked at it and we again said, well, the whole this unstructured data, what do we do? We just pumped it into chat GPT and sort of some of these, uh, open source ML models. Can we figure out that it can predict patterns?

• 34:31Speaker 04

  And the answer was a hard yes, which is great.

• 34:34Speaker 03

  So this really feels like a transition at the company away from we have a ton of data. You can give us some money to access it, pull it out via the API, do whatever you want to do with it too. We are now going to, instead of offering access to the data, offer intelligence that the data gives rise to, thanks to ML models put on top of it. Is that fair?

• 34:53Speaker 04

  That's 100% correct. In a way, like ChatGPT is like this. It took all the historical data of the public web and it's now trying to predict the next word that makes sense. Right? And we're doing the same sort of thing in a smaller context of Crunchbase, saying all of the data we have unstructured. Can we go and figure out what is going to happen next in a company? That that first prediction of whether it be funding or acquisition or whatever it happens to be.

• 35:18Speaker 03

  So how many different things are you predicting now with the new system you've put in place? Which came out earlier this week, I believe, actually. And frankly, Jagger, how accurate are you in these predictions thus far?

• 35:32Speaker 04

  Yeah, it's the first question is easy. The second one is harder to answer. The first one, we did 18 insights and predictions. So it's, it's, we're shifting the focus of it's not just historical now. It's what's happening now at the company and what's going to happen. And that's insights and predictions by definition. On the predictions side, you've got funding prediction, acquisition prediction. Are they going to close? Is the business going to close? Are they going to do layoffs? Are they growing? So there's a number of these different angles. And some of these things we're not going to put onto the website. Like you're not going to go and see like, are they going to do a layoff? Well, that is an interesting prediction that we're using more on the risk side, which is more on the API side.

• 36:15Speaker 04

  So if you're an API customer, you might have access to like the riskiness of companies.

• 36:20Speaker 03

  But you know, as a person who has made most of his money working for other people, I think I would love to know if a major layoff is coming as a signal. So are you not putting that one in public just in case you accidentally worry people about their employment prospects?

• 36:34Speaker 04

  Yeah, I mean, the the reality is, is like, none of these predictions are gonna be 100% per perfect. And that's when you get to the accuracy conversation a second.

• 36:40Speaker 03

  Yeah.

• 36:41Speaker 04

  But there but some of them are not perfect and we don't want to unnecessarily worry people. When we say it's a 60% chance of of a layoff, like, what does that mean exactly?

• 36:51Speaker 03

  Right.

• 36:51Speaker 04

  But for some of our customers, you know, if you're thinking about like, know your customer, that increasing risk sort of like a credit score. Kind of says, well, maybe I should take a pause and maybe do a little bit more due diligence here and dig in deeper. And that's more the intent of it.

• 37:05Speaker 03

  So these are much more directional versus absolute predictions about what the company is going to do. 60% chance of layoffs is useful if you're selling per seat SaaS. And you might think, oh, they're reducing headcount, not a great place to go do a sales call. But it's not like, oh, I have an 84% chance of being laid off in the next 74 days.

• 37:24Speaker 04

  Yes, that's correct. That and especially when you add that time scale to any of these predictions, I guess, very tricky. So our fundraising prediction, we have an incredibly high level of precision and recall on whether or not a funny round is going to occur. When you say, well, when is it going to occur? Now it's a different, that's a different question. If you're saying, well, Jagger, tomorrow, what funny rounds are going to happen? My precision and recall go very close to zero. But what's interesting is it's way closer to 100% than you guessing on your own. Because we've got more signal than you do. It's still a very, very small number, right?

• 38:00Speaker 03

  So, you know, people that I talk to still sometimes have kind of an old Crunchbase model in their heads. And I don't know how long it's going to take to educate people that it's been growing and changing and improving for a long time now. Thanks, by the way, I use Crunchbase all the time. But if you think of Crunchbase as a wiki back in the day, it's cool because everyone got to participate. And it sounds like the new Crunchbase with AI technologies is still predicated in a way on how people show up, access and interact with the data. So it's still, in a way, community powered is my read of the situation.

• 38:33Speaker 04

  Yeah, certainly the 80 million people using Crunchbase is an important part of the how Crunchbase works. But yeah, one of my biggest frustrations is when I go and say, you know, someone says, oh, I know exactly how questions work. I go, why don't you tell me? They're like, well, it's a Wiki. And I'm like, no, well, it used to be in 2014.

• 38:51Speaker 03

  Right, that's 11 years ago now. But 2014 is a useful data point because you joined in 2015 and took over as CEO when it was spun out as a private company. For those folks who don't know, Emergent, Mayfield, Omers, and then most recently, Jagger, you raised from, as I scroll through my notes frantically, you can bail me out. Arrangement growth. I had arrangement growth in my head and I'm like, I know that's not correct. But I bring that up because it's a lot of capital. You guys have raised 106.5 million, according to Crunchbase.com. What did that last $15 million unlock for the company? Because that was in 2022. So in and around the point when everyone started to talk more about AI.

• 39:33Speaker 04

  Yeah, that's right. Honestly, the market shifted pretty dramatically at the end of that year. You might remember, a lot of the data companies out there took a beating as prospecting became less important to the sales prospectus of the world. So we also took that opportunity to do this pivot. So ChatGPT came out. This is when we had a sort of aha moment. And we said, let's not blow all this capital just going and trying to sell. Let's go and actually build something that is very materially different than what's in the market today. So that's why we went and started focusing on this new pivot towards predicting the future rather than just doing better historical data.

• 40:07Speaker 03

  A lot of people are using proprietary closed source models. And I was just kind of curious, what did Crunchbase pick as its kind of like model paradigm? Are you guys using, you know, Metas Lama models? Or have you built something internally? What's the underlying brain for all the new AI stuff?

• 40:23Speaker 04

  Yeah, I mean, it's a combination of a bunch of tech, honestly, and some of it is our own stuff. We use open source like TensorFlow to go and do a lot of the ML side. We certainly are using OpenAI to go and handle a lot of the unstructured data. Just in our testing, it seemed like the best. And but we're still to the way that we can kind of move in and out depending on what's what's better and what's cheaper, honestly. So sort of that combination. So like when Deep Think came out, we're certainly like, hmm, that's interesting. They're like, hmm, maybe not. So there's an ongoing conversation, but the nice is I think every every software and data company that uses this stuff needs to be thinking about how to how to make it interchangeable so you can serve go where the wind blows if something becomes better.

• 41:07Speaker 03

  Yeah, I think the phrase is model agnostic.

• 41:10Speaker 04

  I'm sorry.

• 41:11Speaker 03

  Yeah. Does AI raining compute that whole that but the kind of bucket of of of cost? Does that now take up a much larger portion of Crunchbase's OpEx than like the old AWS bills back in the day did? I'm trying to get a handle for like, what is it like to change your company towards this kind of an AI first model? And how does that shake up your profit and loss statement, frankly?

• 41:35Speaker 04

  Yeah, it absolutely does. As a short answer, certainly, it's a huge line item that didn't exist two years ago. And, you know, we're watching the the people using our stretch base scout today to sort of understand how much that bill might be going up, you know? Um, so it's it's an interesting pricing time. And how do we think about passing those prices on to our customers ultimately, right? Because at the end of the day, I can't just go negative and and and become an unprofitable company again. So it's how do how do we balance the costs? And obviously, as we see the prices going down in the sort of LLM world, that also helps us out. So there's a lot of what you see right now in terms of these numbers, but it's a it is hard to sort of plan for.

• 42:20Speaker 04

  Certainly, I think that's a big challenge for CFOs.

• 42:22Speaker 03

  Well, it's kind of cool because I feel like right now, if you build something that uses modern AI techniques that is too expensive, you should do it because in 12 months, it'll cost 10% as much. And if you capture that market, you can. I can see a reason for the cash burn in many circumstances. This is not 2021. People are not just taking bricks of cash and heaving them out of windows.

• 42:44Speaker 04

  So, right. As long as it's like the challenge, though, again, is like you can do that, but then you're completely vulnerable to someone saying, go make the exact same thing to an AI agent that goes and builds it later. So, so again, at the end of the day, like, and what is the most valuable thing? It's the data that no one can have access to. Even if you have this circle data behind a paywall, I think you're still vulnerable. Because I think there's going to be AI in a feature that goes and social engineers its way in and gets its data. Like, we go and say, AI, go and have all the data there is in the world, and it says at any cost, like, that AI is going to go make phone calls. AI is really go.

• 43:21Speaker 03

  Yes. It's going to create six different email addresses, rotate its IPs, use three different agents from four different models and six different companies to go, and it's going to get around it. I wasn't going to bring up copyright, but I mean, I think it does play here. I know Crunchbase does some work to prevent scraping, if memory serves, but I presume that that's gone, you know, 100x in the AI era. So, how have you guys managed to defend the fort, if you will? And have AI companies come to you and said, Jagger, we'll give you, you know, 20 million a year for the whole data setup, we can just ingest it into our model?

• 43:54Speaker 04

  Yeah, we certainly have had some of those sorts of offers. Like, we are, we are avoiding those. We, of course, do our best to block crawlers. We use a lot of tools to do this, like Trimer X is probably one of the best known, where they're going in and blocking, and that's our whole core competency, but they're still not 100%. And that is why you need to be careful about what data you put out on the internet. I will never put engagement data, obviously, onto the, onto the internet. So, no one will have access to that. But we, if we are using forward-looking stuff, you can crawl that all day long. Predictions change every moment, so you'd have to be crawling us constantly. And there's, and at least today, there's no way to, like, simultaneously crawl every single page that Crunchbase has and pull it all down simultaneously, like, our server's going to handle that anyway.

• 44:38Speaker 03

  Well, yeah, that's a DDoS attack.

• 44:40Speaker 04

  Yeah, so we, exactly. So, we've got, we've got plenty of protection there in the sense that it isn't technically possible. So, the, the, the predictions are so dynamic and so live that we think we're safe there.

• 44:53Speaker 03

  Okay, so there's kind of three major things that Crunchbase just rolled out. We've talked a little bit about the predictive company profiles. Essentially, here's the company, here's their information, and then here's what Crunchbase thinks is going to happen next. And here's kind of the state of the business. There's also a private market homepage. Now, I've gotten to play with this a little bit, but for folks who haven't seen it, Jagger, can you just tell them what that is?

• 45:14Speaker 04

  Yeah, so basically, the idea here is, is there's a lot going on in Crunchbase that you don't know about, right? You, you, now we depend on you doing a search or looking at a specific company. You're totally missing other things that you might really care about. So let's make a homepage that essentially has a feed of all the cool stuff that's happening. Whether that be predictions, insights, what's trending, all of that's available now. And the part that I actually like more is the for you section, where you can go and specify, these are the industries I care about. These are the predictions and insights that I care about. So if you say, hey, I want to know every time a key hire gets made at a certain type of company, you can go and do that. It will just pop up on your feed or a new prediction for funding, whatever it happens to be. And you can click into it, see some details and see the details that are driving that below it.

• 45:55Speaker 03

  That feels very much like an analog to CNBC to me, but with slightly different data. Cause you know, private markets are real time, live, everyone can see the information. But I go to CNBC to tell me, what do I need to know from this massive ocean of data? And it feels like in a machine learning context, you're doing that with Crunchbase now with this new homepage.

• 46:14Speaker 04

  Yeah, I think, I think it's right. Like internally, what we're saying is like, what's the TikTok equivalent for Crunchbase? You know, like how can we create a sort of stream of interesting stuff? And that's like the dream. I don't think we're there yet, to be clear, but maybe someday.

• 46:29Speaker 03

  So I'm going to see you doing a dance on my, my private market homepage. I see. Okay.

• 46:34Speaker 04

  AI generated version of me. That's right.

• 46:37Speaker 03

  So that way it can stay on beat. There's another thing, though. You mentioned it earlier, but I haven't gotten to touch on it in particular. It's a Crunchbase Scout, which is, from my experience, kind of like an AI agent that I send forth to do tasks. And I presume you've stuck with the traditional Crunchbase dog branding because it fetches things and brings them back.

• 46:56Speaker 04

  It's your, your associate that can help you do things. It's going to get better and better over time. Right now, you can say things like make a chart. Comparing funding of these two different companies, it'll go and do that for you. Or the other cool thing about it, it has all the information from the public web as well. So if you want to merge those two things together, what companies may be affected by policy changes, you know, like those sorts of things can start to happen. It can go reach out, do its own searches against the news and then integrate that with the Crunchbase data that it has access to. And I think there's a lot of opportunity hiding in there.

• 47:27Speaker 03

  Everything you just described is kind of what I expected the words to be that came out of your mouth. But to me, just a black box behind the scenes in terms of how many different technologies had to be put together to make that happen. Did it take a long time to get the first version of it that was, that felt right? Like, I'm just curious, because you're a product guy. So I'm kind of curious, like the process to get this from like, okay, we're going to do this to now it's good enough that we can begin testing.

• 47:52Speaker 04

  It's probably one of the most complicated parts of the other stuff that we've launched to make it feel right. It's, there's a lot of technology involved. And, and if you think about it, it's like, how do you scope the conversation to the stuff that is in Crunchbase? I don't want people asking, who's our favorite baseball team and having like strong answers there. So, uh, so how do we scope it to Crunchbase? How do we give it access to the Crunchbase data so to merge those two things together? It's a very complicated problem. And how do you know when to use the public web and when not to use the public web? And, and you can't rely on AI just to figure that out for you. So you have to sort of put those rules in place around it. What do you give it access to? What do you not give it access to?

• 48:35Speaker 04

  And then even like, does it cost involved, right? Every time someone is asking a question, it's costing Crunchbase dollars.

• 48:42Speaker 03

  So wait, not, not dollars, please tell me it's cents.

• 48:45Speaker 04

  Any pennies and dollars.

• 48:47Speaker 03

  I was like, dude, I did a lot of testing. Should I send you a check like?

• 48:53Speaker 04

  But it is maybe more expensive than you think it is. So it is, it is a, um, so how do we, how do we make that efficient and still keep the user experience good? I mean, it's something that we're, we're thinking about. And of course, um, now that it's live, you know, like we're learning every minute about how people are using it, what's not working right. And we're just saying make it better. And that refinement is actually where the most amount of work is going to be put in.

• 49:16Speaker 03

  Going back in time. So the round from 2022, the series D, you guys were talking about having, you know, over 60,000 customers, thousands of SMBs, and you had dropped some really interesting notes about how the company was performing, saying that in the first half of 2022, I believe, you'd added like nine million in net new ARR with only $2 million in burn. Earlier in this conversation, you said you don't want to become unprofitable. So how has growth been? And it does seem quite a lot like you shot for, we're going to get in the black because the market's uncertain. So can you just fill me in with like, I don't know, the last couple of years of financial progress?

• 49:53Speaker 04

  Yeah, sure. So, so again, we got hit with that sort of pain stick at the end of 2022 as well. Growth slowed down pretty, pretty quickly. And because we were, we were really pushing hard and selling to sales prospecting. Like that was really our focus at the time. In 2023, we made a hard decision to do some layoffs. And they're, and we, what we did was we pulled back on go to market. We said, look, we need to go build AI. We need to go and build this in a direction that's going to take dollars. We don't have the money to do everything simultaneously. So we have to go and pull back and go to market. So we did, did that layoff in the middle of 2023. And then in, and so from there, our growth slowed down. And it was actually a business decision to do that. We said, look, growth isn't the thing that's important right now. The thing that's important

• 50:38Speaker 04

  is this tech. You've got to be profitable while you're in that mode. Like if you're going to make that cut, you can't just go and burn the cash until you build the thing.

• 50:47Speaker 03

  So if, if slow growth, then profit and then faster growth later. Oh, okay. That doesn't sound unreasonable.

• 50:55Speaker 04

  It's not unreasonable. So with this launch, we certainly are burning more, right? We want to make sure there's a big splash. We want to make sure that everyone has seen and is aware of this big change for Crunchbase. And we now are reinvesting in growth again. So what we're doing in a more measured way, so that we can make sure that we don't get ahead of our skis. So we're going to try to stay as close to even a positive as we can. And, and as we dip down, and then later this year, we should hopefully fly right back.

• 51:26Speaker 04

  Back to where we were.

• 51:27Speaker 03

  I'm putting you on the spot just a little bit. Coming off of some time when you were not focused on growth. How fast can you hope to reignite that for the company this year? Is that like a, we're going to go 15% or is that like, fuck it, we're going to go 50% this year and go hog wild? I just don't have a good vibe for like, what does a startup that's going back into growth? Set for expectations that will be aggressive, but possible.

• 51:51Speaker 04

  Yeah. And it is a fantastic, especially trying to predict that and put that in a model for 2025. It gets very, very challenging. Our prediction model isn't that good yet, unfortunately. But the, the, the good news is that, you know, we've, we're, we feel pretty comfortable that we're going to have double digit growth. The question is, is that's a like, which part? Which side of that scale? You know, it's a wide range.

• 52:13Speaker 03

  11 or 99?

• 52:15Speaker 04

  That's right. But we're, we're feeling pretty comfortable that we can reinvigorate growth. Like we've got plenty of leads. The question is, is how, how much will people pay for predictions that they've never seen before and that are better than they, they, they probably believe it to be. With, with the precision refund numbers that we have is, is, it's a little like, it's hard to get your head around and just that, how good we're at. Like, how do we even do this? You know, and that's, so we spent a lot of time explaining even how we do the forecasts or the predictions. So some people can understand and start to believe a little bit that maybe we are on to something here.

• 52:49Speaker 03

  Well, it's a little bit more complicated than the old crunch based score, which back in the day, the algorithm was like, I mean, let's call it basic.

• 52:57Speaker 04

  You know, it was basic.

• 53:00Speaker 03

  And it wasn't, it wasn't trying to be more. It was a very basic system that was stood up and worked for a while. But I mean, it was easy because you're like, it's three things. I go, whatever it was versus now we're applying much more complicated AI tools to a growing data set.

• 53:14Speaker 04

  There's thousands of feature vectors that we're looking at across a company to go and make predictions. Or, or there's, we're thinking about, you know, how does investor flow go to a profile, right? And, and, and how recent is that from an edit that entrepreneurs made? And is it about time for them to go and raise funding? Do the, the entrepreneurs look back at those same investors? They're both searching for each other and finding each other on the platform like, oh, That is just on the funding, funding predictions and the acquisition predictions. And there's a lot of power that's hiding here to make the accuracy or really the precision recall of, of, of, and the funny example, 95%, 99%, um, just it, it's just, it's wild to sort of see like, so when we, when we got the results, we're like, no one's going to believe how good

• 53:57Speaker 04

  this is. That's, that's pretty exciting.

• 54:02Speaker 03

  Well, it's good to have that versus the opposite, which is no one's going to believe how bad this is. Oh my Lord. Now what are we going to do? One last one before I go, Jagger. You know, we're talking a lot about M and A, a lot about IPOs and Crunchbase is by definition a late stage company now. And I'm kind of curious about just how you're thinking about building versus selling versus eventually going public. Like what's, what's the vibe from your view on what's the next kind of like major financial step for Crunchbase?

• 54:30Speaker 04

  Yeah. I mean, I, I go to my Crunchbase profile and I refresh the acquisition prediction all the time. So I see what the system states. Um, you know, it's right now, I think it says we're probable that we're going to get acquired and we're unlikely to go IPO. No, that's probably accurate. That's probably accurate. Um, so you know, I think I like, I'm going to just leave it to my Crunchbase profile. I want to do the talking for me at this point.

• 54:53Speaker 03

  Um, well, I, I, I looked up, I, I don't do it often, but I did log into my card account and I was like, yeah, it's still there. So, you know, feel free to pay for my children's private school because I learned what that costs recently and I wanted to cry and vomit. So, um, literally wishing you all the best. And with my journalist hat back on, thank you very much for coming on, Jagger. I appreciate it. And for folks who want to try all this stuff out, um, where should they go on the great wild internet?

• 55:19Speaker 04

  Crunchbase.ai.

• 55:21Speaker 03

  That's the place. All right, Jagger. Well, I appreciate it, man. I'll talk to you soon and best of luck this year.

• 55:26Speaker 04

  All right. Talk soon. Take care.

• 55:27Jason Calacanis?

  We'll see you all next week. We had a full amazing week next week. Wednesday, I think we're having Vlad from Robinhood and Raul from Superhuman. Is that correct? Producer Matty?

• 55:40Jason Calacanis?

  Think, oh, wow. There it is. We got a, got a yes chef. We have two amazing guests on Wednesday and then news, uh, will be there on Monday and Friday. Have a great weekend, everybody. Bye bye.