NEWSMAX

Massive massive news Tim Cook to step down at Apple this broke yesterday Garminator had the scoop of course is coming on the show later today but it's on the COVID of the Wall Street Journal today heavily predicted often debated. It's a time to reflect on Tim Cook's legacy and what's up next for John Ternus, the longtime insider we just say incredibly well executed incredibly smooth they sort of telegraphed it it wasn't a surprise that was already fully priced in yes I personally was hoping that the market would give Tim Cook A 21% salute yes where when the news went out it just immediately nukes 21 massive red candle let everyone know this is we don't like the sign of we love him it's a sign of we will miss him of course rebound immediately yes but I think that's something that the market collectively yes should try to do for yes more symbolism in the candles for for sure chartology is really the key thing 21% but of course.

I saw a post here from Bubble Boy. I want your reaction. Apple is about to become the mecca of hardware engineers around the world with John Ternus taking over at Apple. Is Apple not already the mecca? Is there actually somewhere to go but that is up in terms of hardware engineer recruiting. Do you see this as changing the culture in some meaningful way? I mean, they don't pay like these, you know, OpenAI's heart metas of the world. Apple has been pillaged by OpenAI and Meta and all these companies as of late, they are stripping apart Apple's hardware engineering division, hiring people from every team they can get their hands on, throwing very big offers at them. And so this has been a really big issue that Termis has been dealing with over the last year and change. Um, so. But Apple is, you know, the hardware mecca. They're the company that everyone wants to poach from. Yeah. And they're the company people go to to learn how to build consumer devices. So this is definitely. Yeah, I agree to a large extent with you, actually. All right. With Bubble Boy. Yeah, Bubble Boy, what I did. This might be somewhat separate, but just get me up to speed on the folding iPhone. What is the latest there? Announced in September, Turnus first big new product. Super exciting, super pumped. Yeah, we've talked about this. I'm sick of the candy bar phones. Been the same junk for 15, excuse me, 20 years now. Yeah, I want a foldable. I want a bigger screen. Yeah, I really hope John nee wants a newspaper sized phone. Well, they have those. I've seen those in China through the trifold. Right, but this is a bi fold. Don't get me started on the trifold. Okay, explain the trifold. Wait, why are they awful? It seems amazing. John wants pages of screens that he can turn flimsy and they break. Okay, you need a trifold and Apple like quality in 20 years. Because you know when they do a, when they, when Apple does a trifold, it'll be good. Okay. Okay. You know I open up a foldable phone right now. You open it up and you can hear the screen sort of creaking. Right. And then you have that big line in the middle and then it's like impossible to get your thumb in to open the thing. I hope Apple fix that. I don't want to hear a creak. For $2,000, I don't want to hear a creak. I don't want it to stand. Sound like I'm stepping on a wooden floor. Right. I want it to just open and I want it to open quickly. And nicely and it not be like I'm trying to lift the weight. Yeah. It's still going to be weird for video consumption, though, because I feel like we've done vertical video 9 by 16 and then 16 by 9 widescreen. But if you open up a foldable phone, you eventually get a square and that doesn't really make like a movie watching. No, Apple's is different. Apple's is like the new Huawei phone where it is iPad screen ratio. IPad screen ratio. When you open it. Okay. When you open it. Yeah. Okay. So still black bars. Any intel on. On. No, no, no black bars. Black. Yeah, sure. There'll be black bars when you rotated black bars on the top. If you're watching like a cinema film or even if you're scrolling Instagram, like, you won't necessarily get more view because for so long all the content production has been ultra widescreen. If you're making a Tarantino film and it's super cinematic, or if you're on TikTok and you're doing vertical video, then you're gonna have black bars on the side for the most part. But for so many other applications for Word documents and notes, TVPN will look great on it. Yeah. Something to look forward to. What do you think Turnus's new comp package looks like?

It you're just like back in high school or you're back as a kid is because we're physically wired for that. So that's real. I've always heard that. I've always heard that phrase smell is the sense that's most tied to memory. But I didn't know if it was just something you saw in like a T shirt or something. No, literally, neuroanatomically true. Target wall art? Yeah, yeah, it feels like target wall art. I don't know, it's just one of those things that you repeat. They say a smell from pop science. Thousand words. Okay, so that sounds like something that's extremely hard to reverse engineer. Do we have sensors? Because LLMs, it was so obvious that we had text that was already encoded into data, into ones and zeros. And so transforming that and encoding it, I mean, it was an incredible breakthrough. But it felt like the data was already in the computer. And I feel like that's not true for olfactory data, for smell data. But how are we. Do we need to digitize this before we do anything with it? How does digitization of smell work? Yeah, great question. So I was very fortunate to have those guys as my colleagues. I actually spun OSMO out of Google Brain. And so I was there when all that stuff got invented. And I ran the digital vaction team at Google Brain for about six years before we decided to make it a company through Lux and through gv. And you have it exactly right. Like the Internet had been been accumulating for a while, so we had all this text data, so we could basically slurp that down and start building models. We have chemical sensors, they're called mass spectrometers. There's other kinds of chemical sensors, MOX sensors, there's like a dozen. The history of sensors that can turn chemistry into Data is about 100 years old, maybe more. I mean, a lot of it was pushed forward in the Manhattan Project, actually. But what we've been missing is a map, right? So for sound, low to high frequency is a map which lets us build MP3 and speakers and microphones and Spotify, etc. And for color, RGB is a three dimensional map of color. And that lets us build CMOs, CCD, you know, cameras, etc. We haven't had the map for smell. And that's not crazy because there's three channels of color information in our eye, but we know there's over 300 channels of information in our nose. So in a way, we actually did need to wait for artificial intelligence to Mature in order to have the ability to extract a 300 dimensional map from dynamics data. And that's exactly what we did, starting with our first work at Google Brain. So you got to go get a crap ton of information, right? A bunch of molecules, what they smell like. We've since collected the largest AI data set for scent in the world. That's what drives olfactory intelligence. We have 5 million sniffs digitized, over a quarter million physical samples created. We've digitized about 6 billion fragrance molecules. So they call this like inside of the company. Because there's literally nothing on the Internet. The fragrance industry has done a phenomenal job keeping everything secret. So we built it all ourselves. Remarkable. Jordi, how are you going to make money? It's a good question. If, if, if you go. Actually let's. How can we make money on this? So does TVPN have a scent? Yes, and it's terrible. There's rubber smell in the studio. In the studio there's like thousands of cables and cables. And the cables, we do a good job hiding them, but we have so much gear going everywhere. It's a lot of rubber. And so we had to get these racetracks. Racetracks, they're called to cover all the cables. And it turns out these things smell terrible. Yeah, they're off gas. So we wanted to give our viewers. Yes. The, the full experience. I think we do not be like a scan that's sits on their desk aerosol. And it would just spray a rubber smell into the room. Yeah. So they could. Okay, we could be experience, we could capture it, but I think we should fix it. So. Yes. Really concretely, we raised our Series B. We put an additional 70 million in the bank with Two Sigma leading Lux. Love that it got the gong that was to underwrite building a fragrance factory. Okay. So we have a robot that's the size of a school bus that makes a new fragrance every 100 seconds. And what we do is we design and manufacture fragrances for brands. Oh yeah, that makes sense. So we, we use olfactory intelligence to design it. So super fast, data driven, basically, you know, perfect fit for the brand, for and for the consumer of that brand. And then we actually physically make it. And what leaves our factory is a steel drum of that fragrance while we build them for it. Yeah. We also will do end to end. So like if you want to actually make a physical bottle, we'll actually put the fragrance in the bottle for you so the full product comes out. So if you guys want to launch a TVP and cologne or something like that. We could design it for you. I mean, like, if you just smell like burnt rubber. No, no. Not doing burnt rubber. Rubber. There's some. It smells like disagreement. No, it needs to smell like. Like old $20 bills. Okay. From the 19. And mahogany, the official wood of business. We need it to smell like mahogany with definitely $20 bills. The smell of money. That's. Okay, cool. We've done this. We've done the Smell of Money one, which we demoed actually on the New York Stock Exchange floor, which is pretty cool, but no, we'll make something, send it to you. Talk about sensor miniaturization.

Action. John Ferrari's first electric car is priced at the low price of $650,000. An absolute steal. They're giving them away at that price. It's electric, right. So you don't have to deal with gasoline. You don't have to deal with. You save a lot on gas. You don't have to deal with all the noise that comes out of a V12. Yeah, no noise. So you save on gas. Yeah. Oh, well, yeah. I mean, if you're saving on gas and you're driving, I mean, if you. If oil keeps spiking and Gasoline goes to $1,000 a gallon and you're filling up every week, you could easily be spending millions of dollars a year on gasoline in a normal car. So there's potential cost savings here. So it's sort of a more you buy, the more you save situation. I think Ferrari Luce, there's really going to be a, like, what kind of Ferrari client are you? Moment. No, I think it'll be a statue. If you see someone with this, you know they can eat 300 grand of depre that. And you know that they are. That they are high on the list for the F90. Like, they are working their way up, buying in now, and when the F90 comes out in a decade, they're getting a call. Yeah, they're getting a call. Yeah. Very interested to see how this does in the market. And the good thing is, if you are excited about the Luce, but you're not excited about paying $650,000, you will have an opportunity to buy them for far less than that very, very quickly. Probably. Probably. But thank you for hanging out with us today. It's been an honor and a privilege.

Up next we have Jake from Blue Energy. He's the co founder and CEO with a massive raise. The gong Breaker. Jake, how you doing? Welcome to the show. Hey guys. I'm doing well. I got a feeling you have the biggest number for us. I think you have the biggest number. Kick us off, how much have you raised and then we'll get into what you're going to do with it. But tell us about the financial situation for the company. We have announced a $380 million raise. Nuclear. And what will you be doing with all that money? Yeah, so our focus, we're unique amongst the field of nuclear players. Right now our focus is on building the world's first project financable nuclear power plant. So we're using this funding to actually put deposits down on long lead equipment as well as finishing out the engineering and development licensing on some of our first sites. Does that mean like less R and D risk or more in like the GE or Westinghouse territory? More like, you know, going with products that have been de risked, but it's very expensive and maybe the underwriting is different this time around. Or, or are you working on an entirely new reactor design somewhere in the supply chain? No, you had it exactly right. We are not a reactor designer, we're a developer. But our technology is the proprietary approach by which we go about the plant. So what bugged me and why I started the company was I have a lot of friends in the nuclear space designing really exciting new reactors. And then you have a lot of incumbents in the space who are working on, you know, kind of the same old technology that we've been operating safely for 70 years. Yeah, but nobody was focused on the core issue of how do we build nuclear on time and on budget. So. I grew up in a construction family. I used to be a draftsman for my father's architecture firm. So I just grew up around a lot of construction, did my nuclear engineering and physics degrees in the space and just felt like there hasn't been anyone really focused on the root cause issue. So what we're doing is we're borrowing best practices from LNG and offshore oil and gas and offshore wind to prefabricate everything at existing oil and gas fab yards and shipyards. And then we barge it all as a prefabricated system on the order of a thousand or two thousand tons to the operating site and then really basically are just installing it like giant Lego pieces. But what that allows us to do is bring a lot more debt financing to bear. So we're not taking a lot of reactor technology risk to start in the beginning. We're using mature light water reactor technology to start, but we'll happily work with the Gen 4 reactors as they. As they mature. Cool. Take me through. I feel like you've been wanting. Yeah, I've been wanting this for a long time. John's point has been copy paste, copy paste. Hey, we know how to do this. It works. We don't need to reinvent. I mean it's great if we want to reinvent the wheel. Yeah. We need the next gen tech, new reactors, but also more of us. Let's just build some. Yeah. So yeah. My question is about Vodil lessons from the Vodel project. What. What do you think they did well, that you want to copy, that you want to learn from. What do you. What if anything, do you want to do differently? Yeah. So to put some stats on Vogel, it, like so many other nuclear projects in the west, was ended up being about two to three times over budget and line schedule. But when you double click on where that cost was, you realize it wasn't actually in the reactor technology or the equipment. It was on the over 40% of it was just the construction overhead. So it was the cost of training and relocating 10,000 skilled workers to the site at Vogel. Think about the cost of training, relocating their families, retaining them once they're trained. Because data center projects are trying to steal that talent. You have to set up the nuclear quality assurance program in the field. So there's all this overhead. It's basically like building a small town. And. And then the hope is that you'd be able to move that traveling circus around from site to site. And then a third of the project costs was just capitalized interest on debt because it took over 10 years to build before it started generating revenue. So these are the two big problems we're trying to address is we are moving most of that work off site. We're keeping the workforce centralized at the Fabyard and the shipyard where they already are, so we can start to put nuclear into a learning curve and drive the cost down over time akin to what we've seen in wind, solar, batteries and gas turbines. You know what they did well was it was a mature light water reactor technology. It's a passively safe reactor technology. They, they really pushed the world forward a little bit in the licensing space and steel composite structures, which we're looking into as well. They actually had. This is not well known. They originally wanted to barge it in up the Savannah river, but they had to then dredge the Savannah. They would have had to dredge the Savannah river for miles. And that would have become part of their environmental impact statement. So they ended up. It became such a regulatory and permitting nightmare that they gave up and they said, all right, let's just truck it in. And then they had to truck it in and build a module assembly building on site. So they ended up doing all the welding model from Nightmare remodel. Very interesting. I. Yeah, I'm. I'm fascinated. Where's it. Where's the company base? Where are you guys based? We're in Chevy Chase, Maryland, Pretty close to NRC headquarters. And then we've also got a big presence in Edinburgh, Scotland, where there's a lot of offshore engineering talent, particularly from, like, kind of the history of shipbuilding and offshore oil and gas. Yeah. And then we've also got an office in Houston, also kind of offshore oil and gas capital world. Okay. I read a blog post about one of the potential problems or stumbling blocks that nuclear projects run into, and I want to reality check it with you. The thesis was basically that we have a reactor design. We have a. You know, there's infrastructure around that cement needs to get laid, Pipes need to go here and there. But oftentimes the regulation will change while the project is underway. And so in order to stay ahead of the changing regulation, you might have to, you know, jackhammer a bunch of concrete and move a pipe to a different route because the regulation has changed. Is that real? And is there anything that we have done or can do to get to a regime where the regulation is more locked and deterministic? So that I imagine you're not going to build this overnight, but if it takes you a couple of years, you know that the contracts that you put in place, the plans that you put in place today, will hold and you won't face a massive delay. Yeah. So that was another one of the big learnings from Vogel. Vogel was the first and still today, I think, the only project that did a combined operating and construction license, which means once they locked the blueprint, they really were not allowed to make changes to it during construction. So every time they encountered something and said, oh, we need, you know, the craft labor wound the rebar clockwise instead of counterclockwise. You know, they had to jackhammer it up or they had to go and reapprove it all. And there's just a. There's a long list of things like that that they encountered. So one of the things we're doing is we're following a Slightly different licensing process, the original licensing process of part 50, whereby we're going to incrementalize it so that we don't have to encounter that rework, that kind of regulatory triggered rework situation. But also because we're following this prefab approach and we're moving 80% of the CapEx into a fixed price contracting environment with these fab yards and shipyards. It forces us to go to something like 60% detailed design upfront and locking those designs, because that is what is going to be coming in prefab from the fab yard. So it's sort of baked into the strategy. But really this is about taking a lot of those lessons learned and making sure we don't make the mistakes of the past. But I'll also say we've never had a more supportive regulatory environment than we have right now. Like this is a critical juncture in the history of nuclear power that we can take advantage of. With where the NRC is presently at, what is the power output for the first reactor that you're targeting to bring online? So we are focused right now for the first project using light water, small modular reactors, which the power range of the units we're looking at range between 50 and roughly 300 megawatts per unit. Yeah, yeah. Each site we're targeting doing is going to have multiple units. So it's going to be a gigawatt to a gigawatt and a half per site. Because multi unit operations makes a lot of sense, is important and it helps drive down costs. And then are you already sharing a timeline? Do you have an optimistic scenario, a base case, a bear case? I'm sure you get asked this all the time is the worst question, but we talked to a lot of nuclear founders and we hear a lot of 2000 and 30s and there's a whole bunch of projects with hyperscalers and big tech companies that are looking at 2032, 2035. It's exciting. Better, you know, 2032 than never. But do you have anything to share on, like the timeline of rolling out new nuclear capacity in America? Yeah, we'll be announcing things very soon, but what I can share on the dates and this is actually another unique thing we're doing. Part of our strategy is we're pursuing this thing we call gas to nuclear conversion. So we're actually going to be building half the nuclear plant right away. So the whole nuclear steam turbine system set up for nuclear steam conditions and quality and we're going to fire it early with two combustion turbines is a Two on one combined cycle. So it's kind of a Frankenstein combined cycle. We've actually gotten the NRC to buy off on this methodology through a Topco report recently. So that allows us to actually project finance half the capex for a first smr and then we will build the reactor and splice in the steam and switch it over from gas steam to nuclear steam. So that actually accelerates our commercial operation date with confidence. So we're looking at generating first power in 2030, 2031, mostly driven by gas turbine delivery dates today. And then our first nuclear commercial operation, we're looking at 2032. Okay, so is that switch out possible at any legacy natural gas infrastructure site in America currently? Because that seems like an environmentalist dream. I'm not ready to say yes. I think this opens up a whole new world of fossil to nuclear convergence, which we think is an important precedent to set. Yeah, it seems huge if possible, but I imagine that it's not exactly USB C on both sides, but hopefully one day we can build the adapter. That's right. Yeah. What we're really focused on is there's a lot of announcements out there, a lot of sometimes noise of what there's a lot of exciting things happen in the nuclear sector. We think we've got the first project financeable nuclear project and the first one that's going to power it'll be a new build that powers a new AI data center. So we're excited about that and we think our timeline is credible, is aggressive but credible and defendable. And we've got the right set of partners around it to make it happen. That's amazing. Don't use the word data center. We're using the word supercomputer. Supercomputer. They're supercomputers. They're super computers. No one likes data center. Everyone likes a supercomputer. Yeah. Anyway, thank you so much for taking the time to come chat. Great to meet you, Jake. I'm sure you'll be back on soon. Good luck with everything, really. We really appreciate your approach. It feels like you're making plays and I like how pragmatic but innovative the approach is at the same time. It's great stuff. Thank you. Appreciate the time. I'll talk to you soon.

We have Carolina Aguilar from Inbrain Neuroelectronics building the first in human study of graphene brain interfaces. What's going on? Welcome to the show. How are you? Thank you. Very good. Thank you for having me. Please, since it's the first time on the show, introduce yourself and the company a little bit. Yes, my name is Carolina Aguilar. A lot of people call me Carola. I am the CEO and the co founder of Embrandio Electronics. And we are a graphene based brain computer interface therapeutics company that actually is developing the most intelligent interface between the neural system and AI to restore health for billions. Okay, so walk me through brain computer interfaces and the decision tree that got you to graphene specifically. I'm familiar with like the first decision is probably invasive versus non invasive. We've talked to a number of founders that have taken either approach. How did you confront that first question? Yes, well, I call them implantable and non implantable systems. Yeah. And in our case, we are an implantable company. We believe that the real signal processing that is going within the neural system is actually deeper in the brain. And to listen carefully to what it says, what the neural system says, and being able to decode it, but also modulate it, we need to be close to those neurons and interact with those neurons firsthand. Okay, so when I hear modulate, it sounds like not only processing information that's coming out of the brain, but also potentially writing information back into the brain. Is that the long term vision? Yes. The magic of graphene is actually about reading and writing very effectively at micrometric precision within the brain. I think that's why we took, let's say, higher risk to get an advanced material into this funnel because we see that the benefit is incredibly impactful. Okay, what are the most near term commercial applications? Yeah, so the Morgan Stanley report stated the market in 400 billion. And we thought that we needed to bring a platform with three product verticals to actually penetrate such a big market. We are creating three products. One is, let's say, not implantable. Actually it's a semicronic platform like the modern Utah array. It's like 100 contacts of graphene that can read and write. We went into two more and epilepsy resection at the beginning. And that one is pretty close to commercialization. We're almost there. The second product is the implantable platform for the brain. This is a implant on the brain for Parkinson's disease. We didn't do assistive BCI because we saw 1.8 billion market that is suboptimal, that we could actually displace very easily with this technology. We decided to go therapeutics into Parkinson. The third one is the same platform, but instead of, let's say, brain sensor, we connect a vagus nerve sensor that is actually able to decode all the fibers that go into the different organs. So we have a therapeutic target for each of the organs just by targeting that nerve in the neck. What about the actual implantation process we followed from Neuralink? They had to build a whole robot just to drill into the skull. It's incredibly high precision. Are surgeons capable of implanting this at this stage, or will there need to be other robotic devices that are developed to actually deploy this technology safely? It's an excellent question. I'm coming from medtronic. I spent 10 years in neuromodulation and another three in diabetes. And I think in the future, when micro robotics are ready, we will have a very close relationship between our interfaces and microrobots that probably can deliver this implantation in 30 minutes. But today, when there is not micro robots and we are not Elon Musk, we decided to actually have our platform ready for the current surgical workflows that today exist. So we are not changing much from the neuromodulation workflows. And it's an easy procedure. Two hours. One or two hours is enough. In the case of the neck is 45 minutes. Wow. Wow, that's very impressive. Well, congratulations on all the progress and thank you for the work that you do and thank you for stopping by the show. Yeah, great to meet you. Have a great rest of your day. Cheers. Well,

But up next, we have Spyros from Resolve AI raising a massive round to build AI that runs production systems. Let's bring in Spyros. How are you doing? Hello, guys. Good to be here. Welcome to the show. Sorry we're running a little bit late. Kick us off with an introduction on yourself and the company. I'm one of the founders and the CEO of Resolve AI. We're building agents that can help you debug and run production things. Think of it as the counterpart to coding agents that produce all this code. And our agents are there to support you. Okay. Is your customer always? Deeply in the throes of vibe, coding has rolled out agent decoding across many organizations. Who is the target customer? Do they have to already be deep in the agentic coding wave to really get the value here? They don't have to, but the two are correlated. Like anybody who runs a large software system has this problem. The only solution we had so far is humans manually solving it. Right. Using the tools, being on call. Of course. Now, AI allows us to automate all of this, but I would say this is true. It was true before. Now, with all the AI generated code, it becomes a necessity. Right. So we see strong correlation between the two often. Yeah. And what are customers coming to you asking, is it. I want the code that's written. We're writing way more lines of code. We want it to be more readable, we want it to be more secure, or we want it to be more performant or all of the above. The way to think about it is like, for anybody who's delivering their business through software. Yeah. Look at some of the customers. Coinbase. Sure. Salesforce, MongoDB. Right. To them, reliability is of paramount importance. If anything goes wrong and affects customers, it's a big problem. Yeah. So Resolve becomes necessarily the first level of defense that captures any problem that happens in production that can affect end users. Gives you a resolution and a fix, let's say, so you can accelerate that loop. Right. And it doesn't take too much human effort. But more importantly, it doesn't cause impact to customers. What is like, I mean, the company is now over $1.5 billion in valuation. What has been like, the key to growth? Is it just product led growth? Do you have a big sales team? How are you actually scaling the business as you scale evaluation? Yeah. So this is a very big problem. Right. Anybody who has delivered business software is facing this issue. And whether you're a cto, who pays for, let's say, developers to focus on reliability or whether you're an individual that has solved this problem, you'd rather have AI do it for you. So we've seen like huge amount of demand from day one, since we launched the company a bit more than a year ago. And we've seen it coming from both big and small companies. We primarily focused on larger enterprises because we think there is a lot more complexity given the complexity of the software and most of the growth, I guess most of, let's say the demand comes inbound to us because it's a well understood problem. And of course we have both a product led approach, let's say, but also sales led approach as we work with large customers. Yeah, in some ways the naive approach would be okay, just point a typical AI agent at the code base and just tell me where the fault lines are. But I imagine there's some special sauce in the engineering to understand knock on effects that can happen across a large code base. Are you actively working around context windows or creating a special harness to understand these problems that can come up before they do? Yes. So think of it like we have a production id basically, right? The same thing you have for your code, we have it for all your production systems. Production involves code, involves, let's say telemetry logs, metrics like tools, like datadogs, splunk. It involves aws, Right. So you have to deal with all of these, not just code. And then we also are training our own models now to improve, let's say at the state of the art, let's say. Right. How far you can go? Far enough, let's say with a good harness and a lot of work, let's say on the agency front. But now, and we just announced together, we're funding that we're building a lab to focus on actually training our own models for this domain. Sure, sure. What goes into getting relevant data or actually nailing a specific model for this? Because I imagine that you have some great clients, they probably don't want you training on their data at the same time, if you just grab some open source code, it might not be as complex as the Coinbase Mono repo or whatever they have going on over there. So how do you create enough training data to justify a special model? What is important here to understand is the training doesn't happen on code per se. What happens on is actually the action a human takes to perform a task for the most part. And we're talking about very long, let's say planning tasks here. It might take many, many iterations. Looking at code, looking at datadog, looking at infrastructure, and this, generally speaking, this is not in a training set of models. So and software, let's say in general is both deep and wide domain. Right. So I think if you actually focus on building a model for the types of problems we're trying to automate and how you run in the back production, I think you can have a lot of gains both in performance cost, but even like quality of outcomes. Right. And that's our goal. And I would say, you know, the big labs make it sound, make it look like it's impossible for anyone else to build a model, but I don't think that's the case and you know, that's what we're seeing ourselves with our investments. Yeah. How do you put together such a low dilution round? Yeah, yeah, tell us about the round. I want to hit the gong, the 40 on one and a half billion. So it is an extension. We just did essentially the a. We just, we just did the A at $1 billion like two months ago. Right. And I would say Resolve. There we go. Sorry. Resolve is essentially in many ways created this market. Right. Like AI for production. Sure. And I think it's well understood by investors. It's also proven given the customers we have. Yeah. So. And we're also a very ambitious company. Right. Like we are obviously trying to build the agents and the models for this domain and we have a lot of traction. So I mean, as simple as that. Right. Like there's nothing you can do to create a load dilution out other than be very successful in my opinion these days. That's great answer, great answer. The step one, be successful. I love it. Step one, focus on building a. Focus on building a business. Right. Like this. This is my first startup as a found. Yeah. Yeah. I made this mistake many times before. Right. Of thinking that raising money is success. It's not. It follows real success on a product. Yep. Yeah. No, that's 100%. Right. I love it. Well, thank you so much. Congratulations on the new round. Yeah, great having you on and great having. Excited to watch you guys come. Thank you. We'll talk to you soon. Have a good day. Goodbye.

Everything was correct and then regenerate the image on top of that. Of course we should go to the timeline because there's some people that are having fun with the image generation. Someone made presidents in Elden Ring. There's Joe Biden, which is a Dark Souls style boss that you can fight, and fdr, Lord of the New Deal. And these images, it just looks remarkable if you ever played any of the Elden Ring or Dark Souls games, because of course the text is flawless and then you can fight Richard Nixon in front of the Watergate. And this looks like a mod that I think people would play if it actually existed. Blake Robins said the world is now ready for the rumored OpenAI image model. People are creating Google Street View images that just look perfect and Grand Theft Auto 5 loading screens.

Up next, we have Alex from osmo building olfactory intelligence. We've talked about this before. Can AI smell? That's our current benchmark for AGI. We say if, you know, they talk about white collar work. We see sommeliers as white collar workers. Unless you can smell, it's not AGI. And artificial intelligence is falling short. But it's your first time on the show. I would love an introduction on yourself and the company because I'm fascinated by this topic. Let's talk about it. Your sommelier comment and also what you talked about with Max Hodak's been on my mind. My name is Alex Wilchko. I'm founder and CEO of osmo. We're giving computers a sense of smell. I've been working on this problem for 20, exactly 20 years or so. First as an academic, did my PhD in olfactory neuroscience at Harvard and trained under Bob D. Who trained with Richard Axel, got the Nobel Prize for discovering the receptors of smell. And my AI mentor, trained with Jeff Hinton, who got the Nobel Prize for deep learning. And I'm the one weirdo that's like, we've been waiting for you to. We've been waiting for the entire history of the show. There's been great prophecies of your arrival for hundreds of years. So excited. Okay. I'm so pumped to be here. So. So should we start with maybe like, Olfactory Science 101? Can you set the ground on, like, how does smell even work? What are the important sort of like, building blocks that we should know and then we can build up to the next generation and how AI is being applied 100%. So the chemical slice of reality, all the stuff that's data in the air, we can detect that our sense of smell is literally our brain leaving our skull. So when you smell a molecule, whether it's a tree or it's a meal or a drink, like, the physical pieces of that thing enter into your nose and touch a piece of tissue about the size of a postage stamp. And that's your brain, right? So, like, you were in physical communion with that thing. That information gets turned into neural data, which actually skips all of the normal way stations for the other senses and goes right to your centers of memory, the hippocampus and emotion, the amygdala. So our sense of smell is very primal in that regard. So it's like, it's the reason why when you smell something, you get dragged into a memory and you cannot stop it. You're just like back in high school or you're back as a kid is because we're physically wired for that. So that's real. I've always heard that. I've always heard that phrase smell is the sense that's most tied to memory. But I didn't know if it was just something you saw in like a T shirt or something. No, literally, neuroanatomically true. Target wall art. Yeah, yeah, it feels like target wall art. I don't know, it's just one of those things that you repeat, they say. A spell from pop science. Thousand words. Okay, so that sounds like something that's extremely hard to reverse engineer. Do we have sensors? Because LLMs, it was so obvious that we had text that was already encoded into data, into ones and zeros. And so transforming that and encoding it, I mean, it was an incredible breakthrough. But it felt like the data was already in the computer. And I feel like that's not true for olfactory data, for smell data. But how are we. Do we need to digitize this before we do anything with it? How does digitization of smell work? Yeah, great question. So I was very fortunate to have those guys as my colleagues. I actually spun OSMO out of Google Brain. And so I was there when all that stuff got invented. And I ran a digital vaction team at Google Brain for about six years before we decided to make it a company through Lux and through gv. And you have it exactly right. Like the Internet had been. Been accumulating for a while, so we had all this text data, so we could basically slurp that down and start building, building models. We have chemical sensors. They're called mass spectrometers. There's other kinds of chemical sensors, MOX sensors, There's like a dozen. The history of sensors that can turn chemistry into Data is about 100 years old, maybe more. I mean, a lot of it was pushed forward in the Manhattan Project, actually. But what we've been missing is a map, right? So for sound, low to high frequency is a map, which lets us build MP3 and speakers and microphones and Spotify, et cetera. And for color, RGB is a three dimensional map of color. And that lets us build cmos, ccd, you know, cameras, et cetera. We haven't had the map for smell. And that's not crazy because there's three channels of color information in our eye, but we know there's over 300 channels of information in our nose. So in a way, we actually did need to wait for artificial intelligence to mature in order to have the ability to extract a 300 dimensional map from data. And that's exactly what we did, starting with our first work at Google Brain. So you got to go get a crap ton of information, right? A bunch of molecules, what they smell like. We've since collected the largest AI dataset for scent in the world. That's what drives olfactory intelligence. We have 5 million sniffs digitized, over a quarter million physical samples created. We've digitized about 6 billion fragrance molecules. So all this is like inside of the company. Because there's literally nothing on the Internet. The fragrance industry has done a phenomenal job keeping everything secret. So we built it all ourselves. Remarkable. Jordi, how are you going to make money on it? So if you go. Actually, let's. How can we make money on this? So does TVPN have a scent? Yes. And it's terrible. There's rubber smell in the studio. In the studio, there's like thousands of cables and cables. And the cables, we do a good job hiding them, but we have so much gear going everywhere. It's a lot of rubber, a lot of. And so we had to get these racetracks. Racetracks, they're called, to cover all the cables. And it turns out these things smell terrible. Yeah, they're off gas. So we wanted to give our viewers. Yes. The. The full experience. I think we do not. It would be like a can that sits on their desk. Aerosol. And it would just spray a rubber smell into the room. Yeah. So they could. Okay, we could be experienced, we could capture it, but I think we should fix it. So. Yes. Really concretely, we raised our Series B. We put an additional 70 million in the bank with Two Sigma leading Lux. Love that it got the gong that was to underwrite building a fragrance factory. Okay. So we have a robot that's the size of a school bus that makes a new fragrance every hundred seconds. And what we do is we design and manufacture fragrances for brands. Oh yeah, that makes sense. And so we. We use olfactory intelligence to design it. So super fast, data driven, basically, you know, perfect fit for the brand and for the consumer of that brand. And then we actually physically make it. And what leaves our factory is a steel drum of that fragrance. When we build them for it. Yeah. We also will do end to end. So like, if you want to actually make a physical bottle, we'll actually put the fragrance in the bott. The full product comes out. So if you guys want to launch a TVPN or something like that, we could design it for you. I mean, like, if you tell me the Prompt right now. It smell like burnt rubber. No, no, I'm not doing burnt rubber. Rubber. There's some. It smells like disagreement. No, it smell like, like old $20 bills. Okay. From the 19 and mahogany, the official wood of business. We need, we need it to smell like mahogany with $20 bills. The smell of money. Okay, cool. We've done the smell of money one which we demoed actually on the New York Stock Exchange floor which is pretty cool. That's amazing. That's amazing. But no, I'll send you, we'll make something, send it to you. Talk about sensor miniaturization. My phone has three cameras and no smelling sensor. Can we swap one of these out? When you say mass spec I imagine a device the size of a living room. I imagine that they are getting the dishwasher. Dishwasher. Size of the dishwasher. Is there a path to actually shrinking that down to something that's more portable? So yeah, in the same there's like many kinds of cameras for it. So the one in the Hubble telescope not getting smaller. So if you need resolution it's got, it's going to be big but you can make trade offs and like when the thing that's reading the data instead of it being a person, it's an algorithm you can actually make really intelligent trade offs which is what we've done. So we actually have a sensor right now the size of two shoeboxes and I kind of use that metric trick aptly because we've actually used it to smell fake shoes. So if you're buying a pair of like 500 Air Jordans the real smell different from the fakes, we can actually pick that up. That's crazy. It turns out we can. Yeah. Yep. The counterfeiters use cheaper glues. Turns out. And the, the other thing that's interesting is we can actually tell the factory of origin of the shoe 93% of the time. So the smell is a fingerprint. So we're already miniaturizing these devices. Look, the path to get from two shoeboxes to one shoebox is pretty clear. Yeah, we're working on that. To go to something that's like the size of the AirPods case, there's going to be some like part four engineering required to have it be a component that fits in your phone. There's some breakthroughs like I can't quite see through the fog yet but there's nothing like look, our noses do it so there's nothing that mother nature is saying I think impossible. But we just got A lot of work to do. Yeah, that makes a lot of sense. What about taste? How closely is taste linked? Talk me through the sommelier example. Yeah. So flavor is everything that happens in your mouth. You know, that's a sensory experience of food. Taste is not as like 10 of that. It's like what happens on your tongue. Like, you ever eat a jelly bean? It like plug your nose and you just actually can detect very little of what's going on there. It's because 90% of what you experience is actually called retronasal olfaction, where when you're biting, we're biting on something, there's a chimney effect in the kind of almost the steam of what you're eating goes back through your nose and you smell it. Excuse me. And then there's also the texture and everything in your mouth. So we've done tests and our OI models, this is from a while ago. We haven't revisited it. We're really focused on fragrance right now. But our OI models actually work on flavor surprisingly well. And so the whole world of flavor is there for us when Moretti. But we're really focused on this particular business. I've seen a couple of these sort of. I don't want to call them niche, but like vertical AI projects that are not fully generalizable. There's a DNA model also from Google or DeepMind, and it feels like they're starting to get on scaling curves, on scaling laws. Are you at a point where you feel like, oh, if I 10x the computer, 100x the compute that goes into some. I believe Alex is ready for a 1 gigawatt data center. He can be trusted with that. I would trust you. How. How universal do you think scaling laws are? Is there a scaling law here? Is it data based? Is it compute based? Both. How are you thinking the bitter lesson is real? The better lesson super real. I always think about technology as S curves, right. And like, what's driving you up that S curve? And then how can you hop on the next one? Our current S curve is data, which is why we're maniacally focused on, like, generating a ton of data. Like, we have a giant fragrance robot that spits out a ton of fragrances. We have mass specs running 24 7. We have sensory panels, both domestically of a building of that just smell all day abroad. And we ship them. Great. Some stuff to smell. And that's how we get to 5 million sniffs. Right? So data, data, data. The. The size of the models is not the limiting factor. Right? Now and it will be at some point and then switch to the other S curve. Yeah, yeah. Because you don't just have like the open Internet to scrape because there's not like an existing data set. Makes it totally double edged sword. Right. So we had to make it all right, which is really hard. But also nobody else has it because we had to make it all and had to learn a ton of stuff in order to do that at scale efficiently and all that stuff. Yeah. Where is the, where is the business today? I mean, you've raised money. It seems like there's, you know, monetization opportunities for sure. Are you fully in commercialization? Are you still in research? Is it half and half? Like how do you think about raising more money over time and just growing the business? Yeah. So we're always going like we started with like this curiosity driven drive to figure out how to digitize smell, which is like a pretty wacky thing to do. So we're always going to be trying to push the edge here. But look, we have a factory, we manufacture fragrance for brands. We did this commercial kind of R and D to commercial transition last summer and we're kind of almost at the end of that. And we built a manufacturing organization, we built a sales organization. We have some really amazing partnerships with some big brands and we're making fragrances for brands. You can go into Target, buy a product that has our fragrance in it today. And so we're scaling this part of our business. We're still placing bets on the future though. Right. So I think we've got really the tiger by the tail in this. It's a whole other conversation. Sometimes you come to the factory in New Jersey and see how it operates. But like the fragrance industry is wild. We've got a lot of work to do. There are a lot of opportunities, so we're focused on that. Amazing. Well, congratulations and thank you for the work that you do. We think it's so important and one of the most interesting companies we've ever, we've talked about on the show. Yeah. True science fiction. Awesome. I love it. We're trying to make science fiction into science tech. But like open invitation to come see how it all gets made. It's pretty crazy in person. So come to the Willy Wonka Chocolate Factory where all this stuff happens. I would love to. Thanks so much. So great to meet you. Come back on soon. Yeah, we'll talk to you soon. Have a good rest of your day.

Destroy as much equity value as possible by sounds like discrediting this obscene metric of car or at least the way it's being used today so we can all get back to like building real companies. So that's, that's, that's, that's what I'm trying to get out there. Sure. I mean where this came from is I think I just notice more and more founders and investors telling me things about air reporting. You know, mainly the public reporting but also some of the internal reporting that was just getting more and more skewed. And yeah, there's all these headlines, headlines being published about, you know, ARR records being broken. Sure. And you know, when the laws of physics are being broken you have to ask is it is AI breaking the laws of physics or you know, might there be some other kind of illusion going on as well? And I think, I think it's a bit of both. We have really high growth, awesome companies being built, but when you have really high growth, you know, issues can kind of fester and hide underneath. So yeah, I'd heard a lot of more and more stories of people using this, this metric of C error. Often using this metric when they're talking to press about, you know, their revenue and then gaming it in some pretty, pretty obscene ways. So maybe I just. Yeah. The tweet. So you say the Setup company signs three year enterprise deals. Year one is discounted, say 1 million. Year two steps up, 2 million. Year three is full price. They report 3 million as ARR even though they're only collecting $1 million this year. That's a big deal. The worst part, the customer has an opt out option at 12 months. It's not actually a three year contract. So they're basically like taking the three year number, pulling it into the present even though it's not a, it's not, it's a contract that, that the customer can, can get out of interest and they're not actually on, on the hook. So it's not really. That's rough. Yeah. Just react to that, I guess. Yeah. So I think that's a specific real example that I heard of in the wild from an insider of how these error metrics were being gamed to create some amazing revenue charts. But I would say there's a broad category of issues that I can talk about a few of them that after the tweet went viral I got a huge response of other founders and investors saying that they were saying the same thing and some other examples of the type of gaming that's going on. Yeah, because, because you get one person in a category that starts doing this and then the other people like we have to report the same way. Have to start reporting the same way and cycle. Yeah. And it start, and it starts pretty innocent. You know. C error for folks that don't know C error is contracted error. So it allows you to count revenue that's not live yet. So maybe you're doing like a nine month implementation or you have a one year short, short term stuff like hey, this, this, this contract is going, you know, this, this customer is actually going to be going live next quarter. But we've signed it and we're just going through the implementation process. Yeah, and, but, but yeah, there's nothing that, there's no like law that says you can't say like we're going to extend the sort of like timeline dramatically. It just is not a very grounded way to run your business. Exactly, exactly. Yeah. And I think it's innocent, like three months extra credit credit arguably useful but it's a very easy metric to game especially if you miss those obligations. So I think because we've kind of normalized the forward deployed engineer which we used to call professional services and so now you have these really complex implementations where you might be promising a customer like hey, we'll build this feature and once we build this feature then we'll start billing you what happens if you don't build that feature? So one of the issues you see is companies stacking all these commitments of they'll, they'll switch on billing once they deliver X with their forward deployed engineers. And then what happens if they miss that or what happens if that gets delayed? That and then the reporting it upfront as error publicly but they're not actually at the point where they're. The error is live. So yeah, that's another category of issue. And then there's you know, people reporting pilots, you know, just three month pilots as error and they're free, free pilots that, you know I was talking to an investor yesterday which just sees that all the time from early stage companies like coming out of accelerators saying they have like a millionaire and they look under the hood and it's just all pilots that haven't converted yet. So there's a host of different, you know, issues with the metric. And then the other one is the step up contract where yeah, you're stepping up, year one is 25% of the cost, year two is a little higher, year three is higher. Then people are either amortizing that back over the period to get a higher average or even taking that year three amount. Like you said at the beginning, there's a bunch of patterns that are happening. The other thing is there's early opt outs. You can have early opt outs in these long term contracts, but there's all, we're a contract company, so there's a million ways that a contract can be terminated. Seen a few, Seen a few contracts. Yeah, yeah, yeah. So, yeah, yeah, I think, I think it's a really ungrounded metric and people should stop using it to report their enterprise. AI companies should stop using it to report their error publicly. I think no one should take it seriously, except maybe internally for some projections. You know, it's not, not a good, good metric. What is the gold star example of using air? Correct.

Anyway, I believe we have our next guest in the waiting room. Scott Stephenson from Spell Book. He's the co founder and CEO. Scott, how are you doing? Doing great, doing great. How are you guys? We're good. Welcome to the show. How are you Doing good. Yeah, thanks for having me. Can you. I mean, I want to go into the contracted ARR debate, but let's get the update on Spellbook. How are things going? Where's the company at? How are you feeling? Going very well. We had a killer Q1, crushed our stretch targets last year. We have over 4400 customers on board in 80 countries now. 80 countries? Yeah. We're the most used AI contract review tool in the world. Why so international so early? It's been inbound. Yeah, we've had a ton of interest inbound. So yeah, it really, you know, the choice is accept the customers or turn them away. Yeah, you know, we chose to accept them. Yeah. Is the product sort of multilingual by default? I would say yes. Like there a lot of it is driven by AI models. AI models have some ability to deal with different. Actually pretty good ability to deal with different languages. And then we're able to supplement the models with legislation and norms from many different jurisdictions since we're a legal product. Yeah. So yeah, where else are the key integration points? Like what's the hard work to like bring a country on board or even bring a new flow on board or expand the capabilities of the product as models are just sort of getting better every month by default in the background? Yeah, I mean I think for us that the bit, you know, we try to be two years ahead of the market and build things that are two years ahead of what anyone else is building in legal air elsewhere. And we've consistently done that. We built the very first gen AI product for lawyers back in the summer of 2022. This is like before ChatGPT. Well, a lot of like Claude for Word just came out that was kind of what we launched like four years ago. So we're pretty pretty far ahead from that now. What we really focus on is building unique workflows that are not just chat. I think if you're building something chat shaped, it's very difficult to make that defensible because there's going to be some really good general AI products for just generic chat based, based work. What we focus on at Spellbook is really rails for high volumes of contracts and contract workflows. So we sell to like Fortune 10, Fortune 100 companies and really companies of all sizes who are processing Hundreds of thousands of contracts, not just with the legal team, but with their sales team, the procurement team. We're in manufacturing, shipping, all of these different verticals and we kind of build these end to end rails that allow these contracts to move quickly and safely through organizations. And there's a lot that can slip through the cracks when you're dealing with these high volumes of contracts. A lot of mistakes are made. So, you know, we give like every legal team a second set of eyes on like these massive flows of contracts going through the organization. Yeah. What drew you to cargate? Why are other. Is it, is it legal AI companies that you feel like are getting a little bit dicey on this front or, you know. Yeah, I mean, how do we get here? Yeah. So I've got some interesting examples to cite, but I think it's an enterprise AI problem. And I'll say first, like my goal here with this tweet and what I'm doing is to destroy as much equity value as possible by discrediting this obscene metric of car, or at least the way it's being used today, so we can all get back to building real companies. So that's what I'm trying to get out there. Sure. I mean where this came from is I think I just noticed more and more founders and investors telling me things about ARR reporting. Mainly the public reporting, but also some of the internal reporting that was just getting more and more skewed. And yeah, there's all these headlines being published about ARR records being broken. And when the laws of physics are being broken, you have to ask, is AI breaking the laws of physics? Or you know, might there be some other kind of illusion going on as well? And I think, I think it's a bit of both. We have really high growth, awesome companies being built, but when you have really high growth, you know, issues can kind of fester and hide underneath. So yeah, heard a lot of more and more stories of people using this, this metric of C error. Often using this metric when they're talking to press about, you know, the revenue and then gaming it in some pretty, pretty obscene ways. So maybe I just. Yeah, the tweet. So you say the Setup company signs three year enterprise deals. Year one is discounted, say 1 million year two steps up, 2 million. Year three is full price. They report 3 million as they are, even though they're only collecting $1 million this year. That's a big deal. The worst part, the customer has an opt out option at 12 months. It's not actually a three year contract. So they're basically like taking the three year number, pulling it into the present, even though it's a contract that the customer can get out of. Interesting. And they're not actually on the hook. So it's not really. It's rough. Yeah. Just react to that, I guess. Yeah. So I think that's a specific real example that I heard of in the wild from an insider of how these AR metrics were being gamed to create some amazing revenue charts. But I would say there's a broad category of issues that I can talk about. A few of them that after the tweet went viral I got a huge response of other founders and investors saying that they were saying the same thing and some other examples of the types of gaming that's going on. Yeah. Because you get one person in a category that starts doing this, then the other people like we have to report the same way. Have to start reporting the same way. And it cycle. Yeah. And it start. And it starts pretty innocent. You know CAR for folks that don't know CAR is contracted ARR. So it allows you to count revenue that's not live yet. So maybe you're doing like a nine month implementation or you have a one year short term stuff like hey, this contract is going, you know, this, this customer is actually going to be going live next quarter. But we've signed it and we're just going through the implementation process. But there's nothing that, there's no law that says you can't say we're going to extend the timeline dramatically. It just is not a very grounded way to run your business. Exactly, exactly. I think it's innocent. Three months extra credit arguably useful but it's a very easy metric to game especially if you miss those obligations. So I think because we, we kind of normalize, you know, the forward deployed engineer, which you know we used to call professional services. And so now you have these really complex implementations where you might be promising a customer like hey, we'll build this feature and once we build this feature then we'll start billing you. What happens if you don't build that feature? So one of the issues you see is companies stacking all these commitments of those they'll switch on billing once they deliver X with their forward deployed engineers. And then what happens if they miss that or what happens if that gets delayed? That and then the reporting it upfront as air are publicly but they're not actually at the point where they're. The error is live. So yeah, that's another category of issue. And then there's you know, people reporting pilots, you know, just three month pilots as error and they're free, free pilots that, you know, I was talking to an investor yesterday which just sees that all the time from early stage companies like coming out of accelerator saying they have like a millionaire and they look under the hood and it's just all pilots that haven't converted yet. So there's a host of different issues with the metric. And then the other one is the step up contract where yeah, you're stepping up. You know, year one is, you know, 25% of the cost, year two is a little higher, year three is higher. And then people are either amortizing that back over the period to get a higher average or even taking like that year three amount like you said at the beginning. So yeah, there's a bunch of patterns that are happening. The other thing is like there's early opt out. So like, you know, you can have early opt outs in these long term contracts and. But there's all, I mean we're a contract company so there's a million ways that a contract can be terminated. Seen a few contracts. Yeah, yeah, yeah. So yeah, yeah, I think, I think it's a really ungrounded metric and people should stop using it to report their enterprise. AI companies should stop using it to report their error publicly. I think no one should take it seriously except maybe internally for some projections. It's not a good metric. What is the gold star example of using ARR correctly? Because it's very easy once a company's public to just say, okay, let's just go off of GAAP revenue for the year. And what did you actually book this quarter? There's a whole revenue recognition policy. It feels like there's some benefit to tracking ARR month by month if you're a high growth startup. But what is overrated? Companies should just report their daily annualized run rate or hourly. Yeah, and that goes in the debate of annualized versus annual. Right. So how have you processed sort of the better cases or like, like what is the responsible way to report a revenue metric in 2026? Yeah, I mean I think it depends on the company and the shape of the company and whether it's usage based billing or seat based billing, which you still have. Lots, lots of both. Definitely don't do like hourly, you know, annualized run rate based on the hour. That's not good. I think the, the main thing I would say is it should be live. It's like what revenue is actually live right now for like what Customers are you actually billing and are actually paying you? So calculating run rate based on, you know, the month of revenue that you have coming from customers that are actually paying you, that you're actually billing, I think that's okay. I think, you know, annual recurring revenue based on live customers that you're actually billing that are actually using your service, I think that's pretty good. I think once you start stretching into people who will pay you or, you know, might pay you, that's where things start to. I mean, it can just be so easily gamed. And anything that can be gamed will be gamed. Yeah. Yeah. Are you optimistic that anything will change or do we need to see a massive correction and, and a dark. And dark. The dark ages, like, I mean, post 2022, I mean, I would like to see a steep correction and then back, back, back to building. You know, we'll see if we can make that happen. You know, my reach is only so far, but, you know, I've, I've spoken to a lot of reporters in the past, like 48 hours who are like, I'm always going to ask now, like when the company tells me their ARR, are they talking live error? Are they talking this like long term committed ARR, that might come. So I hope at least the journalists are going to be a little bit more savvy and ask more questions before they report on these numbers. Yeah, I want to ask about who suffers, but in terms of ARR. Yeah, there's almost something where you should just report your last month's revenue instead of doing the times 12 thing and then if people want to multiply it by 12, they can, but at least you're just reporting, hey, last month this is what the stripe account. You can also just say by Q3, we will be at X ARR or. Exactly. That's a different way of saying. That's better than saying we are at 10 million. Yeah, yeah, exactly. Much better. Who. Who suffers here? Is it. Is it purely investors? Because I feel like a good venture capitalist. Their joke job is to dig into the contracts during due diligence to set prices. And if they want to pay 1000 times ARR because they think it's 100 times Carr, that's their risk profile. I would maybe be careful, but that's their job. Or is there a risk that employees see a headline number and think that the business is more stable than it is and they join and then they're rugged. How do you think this affects. Who needs to watch out for this? Basically, yeah. I mean, I think, I think investors are generally good investors are generally very aware of the difference between Sierra and error and aware of the, the widening gap between these metrics. And like in most board Dax, you see two metrics on the press. You only, you know, you usually see crr but it's called error in a board deck. You see both metrics. So, so you know, investors are quite aware but, but I don't think it's victimless at all. I think yeah, employees are signing up for companies and as you know in like a high growth startup people are committing a ton of blood and sweat to be successful based on, you know, part of it is based on the growth of their equity. And if it turn and they might think they're, they're, they're multiple, you know, they might read the headline number, they may not know the number that's actually in the board deck. They might read the headline, you know, error in the press release and they might base, base their decision to join a company based on these headline revenue numbers which are really not grounded in reality whatsoever. Like, and by that I mean I have literal examples, confirmed examples of you know, the press number being three to five times higher than the actual live ARR number. So like that's a huge difference if you think about a multiple. Yeah. Between a public company that's trading at 10x revenue multiple or something and then you get your offer from a company and it seems like they're at 10x but they're actually at 50x. Like that is very material for how you should think about valuing that stock that you're working for. Makes a ton of sense. There's the customers. Like customers are trying to figure out which company is most mature, least mature. And then there's the, you know, like the whole competitive landscape. It's like if one person, if one company starts doing this, all companies have to start doing it and it just creates. Yeah. I mean we could start doing contracted viewership. So we signed three year deals with people in the audience that requires them to tune into every year. They have an opt out after a month. Yeah. If they don't like it after a month. Still advertising based on contract of viewership. Exactly. Contract for the rest of the year. I watch every day, every hour. I mean, I mean, I guess that sort of does happen for YouTube channels that, I mean no one really does this but there was a time when YouTube channels were sort of valued on like the subscriber number as opposed to the average view number. And of course there are some channels where every video gets a million views and they only have 100,000 subscribers for whatever reason. And then there's vice versa, where someone's been doing. You see this on, like, old legacy media accounts on X, where they'll have, like, 30 million followers, and then the post will get like, three likes, and it's like, those are two wildly different metrics. Like, that happens all the time. But this is the name of the game in Silicon Valley, the metrics game. Everyone's finding an edge somewhere. Well, thanks for keeping everyone honest and good reporting. Good luck fighting the good fight out there and spreading the good word. Don't get too sucked into all this. I got a business to build. Yeah, you can. We promise you can come back on in three years with your. With your honest ARR. And take a good victory lap. No, no. Become an investigative journalist. Pivot to investigative journalism. Blow the doors wide open on this. Wow. This goes deeper than I thought. Good. Good to see you. Have a good one. Thanks, guys. We'll talk to you soon.

Number six is a. I only going to say very little about this, but a security camera. Okay. Interesting. Well, at least the Apple Vision Pro has one key fan. No, I see. Do you think the Lamp is a. Is a predecessor for a humanoid? Do you think Apple would ever do a humanoid? I do, I do, but I think it's going to be a decade if they do and they're going to wait. Yeah, I feel like it says, it says a lot that they're not talking about it that you don't know about an internal humanoid project yet. Yeah. Oh, I do, but. But they're exploring humanoids, but they, they're not working on it full throttle. But they have a large robotics initiative. They're working on AI robotics technology and they're also working on robotics hardware. John Ternus actually took control over the robotics hardware team about a year ago. He took it from the AI chief that Apple, you know, got rid of a couple weeks ago, John Giannandrea. But they're also looking at, they're actually building, it's really cool, a gigantic manufacturing arm or a gigantic robotic arm that they want to use in manufacturing, but also used in Apple retail stores to grab, you know, products off the shelf in the back room and whatnot and bring it into the store. That's probably five years away. But they're looking at robotics from a manufacturing standpoint, from a retail standpoint, and also, most importantly, from a consumer standpoint. They've also been exploring a mobile robot, something like an Amazon Astro. But I don't think that's probably going to see the light of day. That's fun. I feel like Apple has a perfect brand. Talk about, talk about Turner's challenge with supply chain, broadly, what you think he's going to be focused on over the next five years? I don't think he will be. I don't think he will be. I think just like. But is that not. You're saying, like, just basically he's done.

Hard decision. Yeah, that was the quote that went through the Journal. But I don't know, it seems like it didn't matter because he got the job. Well, got the job two years ago and I think he's going to do a hell of a job in this new role. I am quite optimistic for Apple in the long term with Turnus at the helm. He has product sensibilities that Tim Cook simply doesn't have. He has product decision making ability that Tim Cook certainly had but wouldn't utilize because he himself knows that product based decisions is not where he could have the most impact. Just like Tim Cook really oversaw the operations part of Apple as the CEO and left product development to other members of the executive team. My expectation is that John Ternus will be intimately involved with the product side of the organization as he was in his prior role to CEO and will leave the operational side to people like Sabi Khan and Priya, the people who run the operations division and at Apple, supply chain, manufacturing, procurement, AppleCare, you name it. And so he's going to pick his spots and his spots is hardware and product development. There's a reason that when he chose his successor for the hardware engineering organization, he chose Tom Maryab. Not an innovator, but an incredible execution guy when it comes to hardware engineering, product quality. He did that because his belief is that he will still be intimately involved and sort of be that product visionary for Apple in this new chief executive position. Yeah, when I remember Steve Jobs, I think of Jobs as an innovator, as a visionary, as someone who both did.

Around in a new AI platform that they're going to be using to improve product development processes and overall quality. Interesting. Okay, I saw a post here from Bubble Boy. I want your reaction. Apple is about to become the mecca of hardware engineers around the world with John Ternus taking over at Apple. Is Apple not already the mecca? Is there actually somewhere to go but that is up in terms of hardware engineer recruiting. Do you see this as changing the culture in some meaningful way? I mean, they don't pay like these, you know, OpenAI's hearts and metas of the world. Apple has been pillaged by OpenAI and Meta and all these companies as of late, they are stripping apart Apple's hardware engineering division, hiring people from every team they can get their hands on, throwing very big offers at them. And so this has been a really big issue that Termis has been dealing with over the last year and change. So. But Apple is, you know, the hardware mecca. They're the company that everyone wants to poach from. Yeah. And they're the company that people go to to learn how to build consumer devices. So this is definitely. Yeah, I agree to a large extent with you, actually. All right. With Bubble Boy. Yeah, Bubble Boy. What I did this be might.

I want this now. Yeah, I think both are going to be really important experiences. And obviously we have like, you know, kind of great like smaller and faster models like the mini that are great for like more synchronous interactions, like within Airtable, like if you go to Airtable or you use ChatGPT with like one of the smaller models, you get that like faster kind of almost like more like real time experience. But I do think like a really important class of work that will come to dominate. Like every Frontier company or company trying to reinvent themselves to be frontier is like figuring out how to operate in this new modality of like, you know, it's like the best developers today don't go and like sit there in front of their IDE and like synchronously like talk to the agent. You have like 30, you know, separate branches that are each being worked on by a different agent. And like you can have the agents continue to like update, you know, the branch based on human and other agent feedback, right? So you have like comments back or like, you know, run like tests, etc. And I think this whole idea of like, look, it's going to take like hours for that entire loop to complete, right? Like, agent pushes some changes, the changes get feedback from other agents or humans. Agent responds to that. Like that whole loop could be hours, not just minutes. So you're not going to sit there and watch it one at a time. But the powerful thing about this is each one is still actually operating faster than a human engineer could have back in the day. When I think about the speed with which our early team at airtable could build features and we had a very good team, one agent on one branch can do the work of maybe three humans back in the day operating probably in three times the time, right? So it's literally like a 10x kind of leverage factor just for one agent. But the best engineers are now able to multitask and kind of basically say, look, I'm going to oversee my own little team of like 20 to 30 agents working concurrently. And so I think it requires like, it's almost like everybody needs to graduate from being an IC to like an IC manager of agents. Meaning like in every function, like if you're like a VC analyst, your job should no longer be to go and synchronously research one company. It's like you're going to go and research like 30 companies and do them all faster, better and higher quality, right? Like, than, than what you could before. And so it, I think that's the greatest leap that is going to be challenging for a lot of people in a lot of roles to make the leap on because it's, it's a totally different mentality to, like, how you operate and what your role is than for what are you pushing the.

Part like be able to leverage the flexibility of our product almost like in a Palantir like way to show the customer, like, we can actually solve this really deep problem for you quickly. Is there any sort of like PLG Motion or Land and Expand that happens in the Fortune 100? Like, yeah, because there's some small team inside of Coca Cola or something that's using airtable and then you're able to use that as a demo or jump off point. That does happen. Yeah. I mean, it's like there are some companies where you just can't even get your foot in the door without the top down. A lot of big banks, like, we just, we were firewalled out until we got some. Oh, interesting. So they can't even. They literally block you, right? Okay. Your IP is like blocked. Okay. Like it might be like hard wall, like request access. Okay. So, you know, there are some companies where you have to come in top down, but like they're, you know, I would say 70 plus percent of our current enterprise accounts, including the ones that are like now like five plus million in revenue, like originated from teams within the company organically adopting airtable. Right. Sometimes it was kind of like shadow it. They just figured it out on their own and they just like they showed real value from using the tool. Right. Like they would build some real operation on it and say, like, well, I've been waiting for like it to deliver me this like old bespoke solution or some like crappy other vendor for like two years now. I got impatient and just built the thing myself. And that's like a big, I think, you know, I think of like the enterprise landscape now is like, you know, there's plenty of dollars in enterprise, right? Even now, like, you know, just shifting from like traffic traditional software to now AI. But like, there's plenty of dollars, right? The budget's there. But like, the question is you're not the only one going after it. And so like, what's your kind of asymmetric wedge to get in there and like take those dollars, right? And if you're a big company, like a Salesforce, maybe it's like we already have the distribution. We have like the customer data in there. We're gonna go and attack adjacencies if you're airtable. We don't have like the scale of like a ServiceNow or like an SAP or Salesforce. What we did have is like the usability of the product. So like the PLP was like kind of the entry point. And then also like, even when we pitch to other people in the company that hadn't used airtable. They had, you know, probably heard about it from a friend. Like, maybe the CMOs, like, you know, partner, like, uses Airtable in their company. Or we can go in and just show them, like, a really compelling demo quickly, talk about AI, what?

While Apple built the biggest consumer hardware company on the planet and thrived in the public market. And I was doing some digging on this as well. I pulled up. It's easy to look at like well, zoom Back to the Mag 7. All of the Mag 7 have done fantastically well over the past 15 years during Tim Cook's tenure. Did he do anything special? Is he in a different category in some way? And maybe not when you look backwards from the current Mag 7, but if you go back to 2011 when Tim Cook took the reins and you look at what were the biggest tech companies there then and how have they performed, it does look like outperformance because Apple was at 377 billion. That was big, huge. The biggest company at the time. Then Microsoft 218 what their peers were at the time where he took the helm. And to be clear, Apple, Microsoft, Google and Amazon were clearly there. Faang was the term at the time. Facebook, Amazon, Apple, Netflix, Google. For some reason Microsoft didn't make the cut in that acronym. That has since changed of course, but there were a lot of companies that didn't go on as significant of runs. You have IBM, Oracle, Intel, Cisco, Qualcomm and HP, all that were in the the top 10. They were sort of the mag 10 of 2011 and now they are not in the mag 7, although many of them have done very well, so it seems. This was our take from a long time. We like to harp on the failure of Apple intelligence and how Siri is ineffective sometimes and the FaceTime interface is odd and the new iPhotos app is hard to use. But where it matters, we give them credit on genmoji. No, where it matters is did they navigate tariffs, did they navigate supply chain, did they navigate the transition to Apple Silicon delivering a great product consistently that doesn't break. Like we have ordered so many Apple devices throughout building tvpn and there was a time when you would get a new consumer product and it would just be oh, it's a bad one, I got a bad one and I gotta take it back. And that's never happened. The quality control is flawless and navigating a very, very difficult Chip export Act From Biden in 2022 all the way to the Trump tariffs, to different political swings back and forth and back and forth. And Tim Cook has just done a great job of like keeping the wheels on the train going down the track and I think that should be celebrated even though this new AI features bound to be under, under appreciated because he wasn't, he wasn't the visionary that, that, that Steve was. But he also never. I don't think he ever wanted to be seen in that way. But the consistent operational excellence over almost two decades is almost unprecedented at this scale. It is. Yeah. Just the way you put it. Right. The same experience that I had getting a new Apple Computer as like a teenager I have today. It is actually almost remarkable how similar the experience is. You open this wonderful box, you get a great device that works for a long time, and you still get that today. So the consistency and. Yeah, I think he will just get more. When people kind of process the run over time, he will just get more and more respect. Yeah. And a lot of the.

We have our next guest, Mark Gurman, the Germinator himself, in the waiting room. Let's bring him in to the TVP on Ultra Dome. Mark Gurman, how are you doing? Tired. How are you doing? Tired. I can imagine this has been months in the works. You predicted this many times, but also on our show, how did this come together? Did this match your timeline? Were you surprised by this particular Monday that it was announced or. Walk us through the scoop. Yeah, and get the, get the scoop. Get the scoop ready. This is the scoop for you. The golden scoop. Oh, I've never. I've never seen that. Yeah, it's a new prop in the studio. You'll have to see next time in person. The OpenAI deal is already doing work for you guys. There you go. So here's the deal. There's a reason I published my profile of John Ternus just a few weeks ago. Right. This was all coming together. Things really ramped up internally at Apple on this. At the end of last year, things have been in motion. The plan was to announce it after the 50 year anniversary celebrations. And it almost felt like the 50 year celebrations were not just, you know, about Apple's 50 years, but sort of a goodbye celebration to Tim Cook and his legacy at the company. So it all came together over. Over several months. This really started about two years ago when Tim Cook identified John Ternus as the next one. Ternus had been being prepared for this role probably for over five years at this point since they put him on the executive team and he became SVP of hardware engineering. But this started in early 2024 and at the time I wrote that, that was the first time I read that he would be the next one. Yeah. How do you. Have you been able to process. I know you've published a few memos. Have you been able to ascertain anything about the internal response? Are Apple employees excited about this? It feels like it's been managed from a communications perspective very carefully and so it shouldn't have been a surprise to anyone. But are Apple employees generally excited about this? It's. It seems like there's a lot of cause for optimism, but I'm always interested to hear. Yeah, there was that one article a couple months ago where clearly they were getting quotes from former employees that were like kind of taking potshots at him, basically saying like, he's never made a hard decision. Yeah, that was the quote that went into the Journal. But I don't know, it seems like it didn't matter because he got the job. Well, got the job two years ago and I think he's going to do a hell of a job in this new role. I am quite optimistic for Apple in the long term With Turnus at the helm. He has product sensibilities that Tim Cook simply doesn't have. He has product decision making ability that Tim Cook certainly had but wouldn't utilize because he himself knows that product based decisions is not where he could have the most impact. Just like Tim Cook really oversaw the operations part of Apple as the CEO and left product development to other members of the executive team. My expectation is that John Ternus will be intimately involved with the product side of the organization as he was in his prior role to CEO and will leave the operational side to people like Sabi Khan and Priya, the people who run the operations division at Apple. Supply chain, manufacturing, procurement, Apple care, you name it. And so he's going to pick his spots and his spots, his hardware and product development. There's a reason that when he chose his successor for the hardware engineering organization, he chose Tom Marieb. Not an innovator, but an incredible execution guy when it comes to hardware engineering product quality. He did that because his belief is that he will still be intimately involved and sort of be that product visionary for Apple in this new chief executive position. Yeah, does. When I remember Steve Jobs, I think of Jobs as an innovator, as a visionary, as someone who both did Pixar and the iPhone. You know, so many different projects, a lot of them wildly successful. Tim Cook felt like a focusing of that a little bit, but you still had the car Vision Pro. There's a few different projects going on. Is this more, is this the most focused Apple has ever been and will ever be? Or do you think that there's. Do you think Turner has like some aces up his sleeve where he might want to take a wild swing at something? I think with Tim or which on Turners is going to have to do is stay the course in your iPhone, iPad, Mac, Apple Watch, AirPods upgrades. But at the same time is going to need to do a better job bringing out new product categories that Tim Cook has done. If you look at Tim Cook's legacy in terms of major new products, it really was on the services side. The AirPods and Apple Watch, you know, those were both really developed by management teams, engineers and people who came from the Steve Jobs era. That's not a slight. But my point being is that we really haven't seen anything wholly new that is also successful since 2016 with the AirPods and the Apple Watch. At the end of 2014, the Vision Pro has obviously been a Tim Cook product, a Tim Cook priority. And it's been sort of a flop, at least for now. I know Apple has a very long decade long spatial computing roadmap. They eventually want to get to AR glasses. They'll have display list glasses to compete with Meta several months from now into 2027. But he needs to get cracking. There are six major Apple products in development right now. Six major new product categories. AirPods, smart glasses, pendant Pendant, Smart display. Is that the lamp or the kitchen thing? No, no. Lamp is number five. Smart display is different. The tabletop robot. So that's the lamp, the moving lamp. And then number six is a only going to say very little about this but a security camera. Okay, interesting. Well, at least the Apple Vision Pro has one key fan. No, I see. Do you think the lamp is a predecessor for a humanoid? Do you think Apple would ever do a humanoid? I do, I do, but I think it's going to be a decade if they do and they're going to wait. Yeah, I feel like it says, it says a lot that they're not talking about it, that you don't know about an internal humanoid project yet. Yeah. Oh, I do, but. But they're exploring humanoids. They're not working on it full throttle, but they have a large robotics initiative. They're working on AI robotics technology and they're also working on robotics hardware. John Ternus actually took control over the robotics hardware team about a year ago. He took it from the AI chief that Apple, you know, got rid of a couple of weeks ago. John Gianna. Yeah, but they're also looking at, they're actually building, it's really cool, a gigantic manufacturing arm or a gigantic robotic arm that they want to use in manufacturing, but also used in Apple retail stores to grab, you know, products off the shelf in the back room and whatnot and bring it into the store. That's probably five years away. But they're looking at robotics from a manufacturing standpoint, from a retail standpoint and also most importantly from a consumer standpoint. They've also been exploring a mobile robot, something like an Amazon Astro, but I don't think that's probably going to see the light of day. That's fun. I feel like Apple has the perfect brand. Talk about, talk about Ternus's challenge with supply chain broadly what you think he's going to be focused on over the next five years. I don't think he will be. I don't think he will be. I think Just like that. But is that not. You're saying like just basically he's delegating it broadly. Ignore that. It's a kind of a key risk to the business to have, you know, I mean his team with the hands in all hands meeting with Apple employees this morning. He was pretty clear that Tim Cook didn't do everything. Tim Cook chose his spots and Turner said that he's going to pick his spots as well as we know turn Tim spots was operations, finance and sales and he delegated every everything else. My sense is that Turnus is going to turn this mandate. Turnus was hired because they believe that he is going to be able to bring Apple back to the forefront of product device innovation. Okay. They already have the best in class, operations, finance, salespeople. They don't need Turnus to do that. They need Turnus to keep his eye on the prize, which is products. Yeah. And what do people point to when they say that there's a risk to Apple staying on the frontier of product development? I saw the Android phone that has the privacy screen that you toggle on and off. That looked like kind of a cool feature. There's folding phones that they're working on. But are any of these features that exist in other phones? It feels like they haven't actually gotten a groundswell and started pulling iPhone users away from the ecosystem. But are there features that people are worried about or what? It's not here yet. Yeah, okay. It's not here yet. Nothing you've seen is the risk. The risk is whatever the hell Meta and OpenAI and Hark and all these companies eventually come out with the risk is one of those companies doing something really cool and jettisoning Apple from that perspective. But we all know that nobody has done, quote unquote cool stuff yet to steal away iPhone users. Nobody is ditching their iPhone for Android. In fact, the switching is going in the other direction, despite the fact that Apple is supposedly the most innovative company in the world and has the least innovative AI technology. Yeah, but consumers care about value and things like the MacBook Neo really deliver that value. Brand colors, value system, privacy. Ternus was only senior VP of hardware engineering at Apple for five years. It's a short tenure to be an SVP of a division at Apple. In the reason because he's. Oh, there you go. Now he's been at Apple 25 years. I'm kidding. We use that ironically. No, I know, but the point I'm trying to make is that he still has a legacy and Ternus legacy is making Apple hardware more performant in terms of speed and battery life and higher quality. He's really focused on the durability and longevity and the reliability of Apple products and it's meaningful. I think that the person that they chose to be Turnus replacement in hardware engineering, Tom Harryeb from Intel, is a product quality and reliability expert rather than a product design person. Yeah. What was the thinking? I mean I remember we talked about this how I got the new iPhone and it has immediately been dinged. What was the thinking on making it visible? But it's better for heat or better for wireless connectivity even though you can't get the color to adhere to the. To the material as much so it scratches off. Is that the, is that the current trade off? Yeah, yeah. There's trade offs with every material. Like titanium was light, it looked cool, you could beat blasted it, you know, interesting. I mean it looked interesting and it gave them a good marketing point like, oh, come buy a titanium phone. Like anyone cares about the material of their phone but it had really bad properties related to heat. Aluminum, which we've known for 20 years, is an excellent material to build consumer electronics out of. So they went back to the basics. You know, they were really talking about at the end of last year splitting the line between ultra thin with the iPhone air and pushing the iPhone pro to the right as much as possible by making it more performant. My expectation is they're really doubling down on this. Their goal is really to just squeeze as much performance and power in these iPhone pros as possible. And for everyone who needs less power, you can get the thinner and lighter iPhone air and I think you're going to continue to see Turnus push in that direction, making the MacBook Pro as amazing and most performant as it can be in pushing everyone else to the MacBook Neo and the MacBook Air. And I think his legacy on performance in product quality is. Is really important thing to remember. Yeah. Has Turnus ever talked publicly about AI in any capacity? He talked about AI in his all hands meeting with employees this morning. He said that I'm gonna check it out. No, just hold on, hold on. I don't want to give you inaccurate. Yeah. Fake news. Okay. Yeah, just hold on. Bear with me. I think I. Internal memos. No, no, no, no, no. He said that he's especially excited to be stepping into this role at this moment because I am telling you we are about to change the world once again. He said Apple has an incredible roadmap ahead and that I'm not exaggerating. When I say this is the most exciting time to be building products and services at Apple in my entire career, AI, There you go. Is going to create almost unlimited potential. We're going to be able to keep unlocking possibilities that, that are going to create entirely new opportunities for our products and services. And I'm so excited about what that's going to mean for our users. Earlier this month he reorganized Apple's hardware engineering division around a new AI platform that they're going to be using to improve product development processes and overall quality. Interesting. Okay, I saw a post here from Bubble Boy. I want your reaction. Apple is about to become the mecca of hardware engineers around the world with John Ternus taking over at Apple. Is Apple not already the mecca? Is there actually somewhere to go but that is up in terms of hardware engineer recruiting. Do you see this as changing the culture in some meaningful way? I mean they don't pay like these, you know, OpenAI's hearts and metas of the world. Apple has been pillaged by OpenAI and Meta and all these companies as of late. They are stripping apart Apple's hardware engineering division, hiring people from every team they can get their hands on, throwing very big offers at them. And so this has been a really big issue that Termis has been dealing with over the last year and change. So. But Apple is, you know, the hardware mecca. They're the company that everyone wants to poach from. Yeah. And they're the company that people go to to learn how to build consumer devices. So this is definitely. Yeah, I agree to a large extent with you actually. All right. With Bubble Boy. Yeah. Bubble Boy. This might be somewhat separate, but just get me up to speed on the folding iPhone. What is the latest there? Announced in September, Turnus first big new product. Super exciting, super pumped. We've talked about this. I'm sick of the candy bar phones. Been the same junk for 15, excuse me, 20 years now. I want a foldable. I want a bigger screen. Yeah. I really hope John nee wants a newspaper sized phone. Well, they have those. I've seen those in China through the trifold. Right, but this is a bi fold. Don't get me started on the trifold. Okay, explain the trifold. Wait, why are they awful? It seems amazing. John wants pages of screens that he can turn flimsy and they break. Okay, you need a trifold at Apple like quality in 20 years because you know when they do a, when they, when Apple does a trifold, it'll be good. Okay, okay. You know, I open up a foldable phone Right now you open it up and you can hear the screens sort of creaking. Right. And then you have that big line in the middle, and then it's like impossible to get your thumb in to open the thing. Yeah. I hope Apple fix that. I don't want to hear a creak. For $2,000, I don't want to hear a creak. I don't want it to stand. Sound. Sound like I'm stepping on, you know, a wooden floor. Right. I want it to just open and I want it to open quickly and nicely and it not be like I'm trying to lift the weight. Yeah. It's still going to be weird for video consumption, though, because I feel like we've done vertical video 9 by 16 and then 16 by 9 widescreen. But if you open up a foldable phone, you eventually get a square. And that doesn't really make like a movie watching. No, Apple's is different. Apple's is like the new Huawei phone where it is. IPad screen ratio. IPad screen ratio. When you open it. Okay. When you open it. Yeah. Okay. So still black bars. Any intel on. No, no, no black bars. Black. Yeah, sure. There'll be black bars when you rotate it. Black bars on the top. If you're watching like a cinema film or even if you're scrolling Instagram, like, you won't necessarily get more view because for so long, all the content production has been ultra widescreen. If you're making a Tarantino film and it's super cinematic, or if you're on TikTok and you're doing vertical video, then you're. Then you're gonna have black bars on the side for. For the most part. But for so many other applications for Word Documents and Notes, TVPN will look great on it. Yeah. Something to look forward to. What do you think Ternus's new comp package looks like? You know, we were. We. We almost marched on Cupertino multiple times because of Tim Cook. To get Tim Cook, you're going to handcuff yourself to the spaceship. Yeah, exactly. Exactly. I would. I'm just guessing. I'm just guessing. I think a million shares over 10 years, that's pretty big. And wait, can I just tell you why I think that. Yeah, why? Because that's what they gave Tim Cook when he was named CEO. A million over 10 years. So I would assume it's the same, but again, I don't know. Entry level researcher salary, but it's a good start. If you can. If he can. Tim Cook is getting Tim Cook. You know, Tim Cook was getting 100 million a year, and then everyone flipped out except you guys. And so he had to cut his Pay to like 40 million. And then when things died down, he's like, all right, I'll take. I'll take my 70, 80 million. I slept peacefully for, for those years. And then you should see my sleep score once. We were really. We were really the strongest supporters of the, the Tim Cook pay package. But I guess, I'd guess a million. I was just like a baseball player. The same amount as a guy leading, you know, a 4 trillion dollar company. Make it make sense. Yeah. Maybe it'll be 500, 000 shares. Maybe it'll be 500,000 shares. I. I don't know. But I know that they gave Tim Cook a million. We got to get those numbers up. You got to get those numbers. It's time to really. You're all in on furnace already. We should preemptively march. We're bullish on both. We love both here. Well, now you get. You get both. Now I know we do. Yeah. Why? Why, why is 65 a retirement age for the CEO of Apple? Like we were talking about Warren Buffett. He was able to manage a, you know, trillion dollar organization well into his 90s. Is it a more physically demanding job? Is he traveling more? Is it. Is his hand ringing from shaking hands in D.C. why not have another 10 years if you're in that seat? I don't think the hand situation has much to do with shaking other people's hands. Why is he not there another 10 years? Well, he needs to give the new guy Runway. Okay. I'm sure there are some. I'll just tell you what, Tim Cook. I'm not going to get into it, but what I'm going to do is tell you why Tim Cook said he's stepping down. He said he's stepping down because it's the right time. And there's an intersection of John Turner is being ready, Apple's finances being in a very strong place, and Apple's future roadmap being in a very strong place in terms of the real reason why he's stepping down. Now, you can read some of my prior articles taking a deeper look at the situation. Okay. Yeah, Makes sense. Cool. I love seeing you. I love talking to you. Thank you for coming on. Congratulations. Get some sleep. Great to see you, Mark. Keep up the amazing work. We'll see you soon. We'll see you in 15 years for Termis Junior. Can't wait. We'll see you. Goodbye.

But yeah, without further ado. Sure. Let's bring in, let's bring in Howie. Luke Airtable. Howie, how are you doing? Welcome to the show. Thank you so much for coming on down to the TVPN UltraDome. For those who might be living under a supercomputer, not a data center, introduce yourself, tell us who you are. All right, Howie Liu, co founder and CEO of airtable, now also maker of Hyper Agent, part of Airtable. Cool. I've been doing this for 12, 13 years now. 13 years. Overnight success. Give us the backstory. Take us from college through early career to the first, the founding moment. Yeah, yeah. So in high school I kind of got into programming. Like my dad had this C book, left it in this corner of the house one summer and I was super bored cuz I learned it. I mean this was like 2003. Yeah, it was like pre Python for the most part. Yeah, I mean Python was around but people didn't really use it. It wasn't the jumping off point. Java definitely like early days for even like web apps like Rails exist, like all that stuff. So learned C, thought it was kind of cool and then started thinking about like how do I turn this into like a real career? Because it just, it was a lot more fun than like classes. And like I went to Duke, took some like mechanical engineering classes, but on the side basically learned how to do web app programming. Like first with php, then like Rails and stuff. And I stumbled on Y Combinator. Actually like pretty early on it was like maybe 06. Literally the first class. No, because I remember like I saw Looped someone's first company and I was doing research, like I wanted to do a similar type of company or product and I was like, no, you know, I was a nobody in college, didn't know anything. And through that like found out about Loopt, I was like, damn it, somebody's already got this, this, this idea. And then learned about YC and like Sequoia. And so that kind of became my first inroad into like just learning about that whole world of like startups and tech. Eventually after college applied to YC with my first company which was big. Basically it was called E Tax, like contacts with an E. Okay. It was like a personal CRM product. Yeah, exactly. Oh yeah, yeah. They were like, oh, this is a big problem. Everybody has this problem. Yeah, we'll fund it right away. Am I correct in my take has always been the people that clamor for a personal CRM really just don't realize that their friends or people that they do business with. They should either just use a real CRM or just. And just be friends. I mean, it's. Yeah, I think it's like a very unique target audience for whom it's a very high pain point. So I think there's like a market there, but like, it's a very like, power user prosumer audience. And the punchline of it though was that like, after a year of work, we like worked on this thing, raised some money, you know, hired like a couple people, and then we kind of realized, like, I sort of realized, like, I think it's a more niche market than we set out to go after. And we had some like, different acquirers come knocking. Like, Salesforce was one of them, but like also big, like consumer Internet companies who want to just buy us for talent. Sure. And you know, to me it was like. And what this is like we were winter 2010, that batch and the acquisition talks were like 2011, basically late, late 2011. And you know, we kind of got this point where I realized, like, I want to work on a really big problem, like a meta problem. Not like, here's one small niche for like some people, but instead, like, what's like the underlying problem, which is, you know, you could actually build this whole CRM with like an app platform. Right. Like, you really want something that's a lot more just configurable and customizable. And so we took an acquisition by Salesforce, worked there, and like, for me, the big light bulb moment was, you know, Salesforce is one big data. What's Benioff like on all hands? I mean, electric on the. Well, there weren't that many, like all hands were quite infrequent at Salesforce at the time. But like, I went to like, their. Their big sales kickoff that year. I mean, Mark is like a very smart guy and also a very like, commanding presence. Like, he's a physically. Like you would if you met him, like on the street. You didn't. Didn't know who he was. Like, you'd probably think he was like a linebacker in the NFL. Like, he's massive and he just like, he exudes charisma. Like, even in like a quiet, small room. Like, he'd take meetings in his house. Like, I go over with like, you know, some of the other. Like, he's the final boss. I mean, but like, not only like, he's like, he's going head to head with him, But he's like, he's just got so such a presence even when he's not like, like booming, you know, out loud, like on a stage, like when he's quiet room, the dolphin sound. Oh, I don't know about that one, but I didn't get to see that part. But that's the whole genesis of Salesforce. Apparently he came up with the idea while he was swimming with dolphins. I guess that's what. What all the Hawaii motifs are for. Yep, yep. He loves. It was a fun time. I mean, honestly, it was a really fun company. Like, you know, it was like for being in enterprise software, it was like one of the more fun experiences. Like, you know, people were kind of super laid back. It's like all aloha and like or what. But. But learned a lot, you know. And I think like, for me the big aha was like, wow. Like all of enterprise software is basically just like a database with like some app logic and interfaces on top. Right. And like, that's basically all that Oracle is used for. That's basically what SAP is. That's what Salesforce is. And if you could create like a way simpler version of that, like that's super intuitive, like, that's. That might be a big market. And yeah, that was basically the genesis of airtable is like, I want to go and like, basically PL gfi before that even was a term like this category. Sure, sure. So, yeah, what was the. What was the initial, like, hunting for team, raising money, building an mvp. Like, what was the first step? I mean, the second time around? Like, so this was my second company then, airtable, you know, I wanted to do things a little bit differently than the first time. Like, the first time was kind of like just go and like, apply to yc, get in, do whatever it takes to get some traction. Like, it literally felt like this roller coaster, right? Like every week it was like, launch, get some, you know, signups, go and like, raise money. And airtable is a lot more premeditated. Like, we spent two and a half years building the product before even launching. Wow. It was actually weirdly, a very parallel timeline to Figma. So like both, like two and a half years around the same time, launched around the same time. Very like PLG in both cases. And I think we both kind of exploited like, you know, the advent of like rich browser experiences, like for the first time. So like you and multiplayer. Yeah, you couldn't build like a rich real time, like single page app experience before maybe like 20, like 1112, like, and really it became like, you know, kind of really legit in like 201415 with like V8 becoming like really mainstream and dominant. Like, you know, just like the performance of the browser became there. Right. So we built this product like, you know, the premise was, let's make it really, really simple for like, anyone, like a small business owner, like, you know, podcasters or even like people within a larger, larger company to build their own app or database. And they're like, you know, FileMaker, Microsoft Access. Like, some of these products existed back in the day, but never made the transition to the web. So we kind of built it. Yeah. My career started shortly after you guys kind of like came onto the scene. And my first ever business, we signed up for Airtable, like probably day one. Yeah. Still use it. How many? Like eight years later? Something like that. So just running, like it's been core infrastructure every single day. That's awesome. Yeah. And like evolved. And it turns out, like, databases are pretty sticky. Right. Like, think about all the Oracle installs and like, just random, like large enterprises that are just still chugging away. Like you've got your system of record in there and like, built a lot of like, customization. Oracle database as a revenue line within Oracle is growing. Revenue top line is growing for the Oracle database. Not their AI stuff is a separate thing. Very different valuations, but it is growing, which is, I think, a narrative violation that I think a lot of people talk about, like, the early go to market. I mean, you said plg, but, like, are you sending this to like, startup friends? Are you trying to sell this into Salesforce on day one? Like, how are you thinking about, like, enterprise versus mid market versus startups versus, like prosumer? There's like so many different routes you can go. Yeah. So this is like 2013 ish. Right. And like, at the time there weren't that many. I mean, there wasn't like really a PLG like, thing. Yeah. I mean, Slack had, I think, just come out when we launched in 2015, so they had launched a little bit before. Dropbox and maybe Evernote were kind of the best, like, PLG pioneers. Yeah. And they were both very like, consumer prosumer first. So, like solo, like, individual user first. And. And then Drew Houston has the funniest riff on. I don't think he calls it plg, but he calls it like the web growth. The Web 2.0 Growth Playbook, which is like going viral. But he takes it a lot further and he's like, so you want to sneeze on as many people as possible. And he refers to that as, like, if you send someone a file. Yeah. That you've sneezed on them and then they might create an account. It's just like a much more like visceral way viral market. I can't even get this well. Yeah, no, I mean, take the thing about. So my first company is an ad network. So we would. A company would come and say, I want to advertise $100,000 budget and then the company would put together a dashboard of potential buys and then the person would go through and so it was inherently viral. Every customer that would work with us had to log into Airtable and use a product. So that was just happening at massive scale. Yeah, yeah. And I think that that type of like you have some like data set. Like, you know, maybe it's for your ad inventory or whatever, maybe it's for like your CRM or whatever. You need to collaborate with it. Like it's a very fundamental construct in just like how knowledge work is done. Right. So I think like the lesson learned for me or like that the principle applied was like, can you go after something that's so foundational that like it's always going to be around? Right. Like, I think with the personal CRM thing I kind of felt the like turbulence of like, is this in vogue right now? Is it not? Like. And I really wanted to go after something that felt like it's going to be around for decades and like, what's more eternal than like people need like databases, database that you can do stuff with the past 40 years of computing, it's probably going to be around for the next 40. And like, even now with agents, it's like the database layer actually becomes more important. Right? Yeah, you don't want just like a bunch of ephemeral context windows, like for agents. Like they need to like store and collaborate on data along with humans. So you know, we kind of pick that as the vantage point. And a lot of the early customers were like startup founders, like small business owners. But like, interestingly, we had like, we had written this like fake business plan. It was like basically like a vision deck more than an actual business plan. But like we had said, you know, conjectured, like we're going to have to go after like a long tail of like the kind of prosumer SMB audience. Basically like Dropbox. Yeah. And I think what was really surprising is it turned out to be a little bit more like Slack, where we got the most virality within larger companies. Like there'd be a big like media company or like even like a scaled startup, like a WeWork or something that would run all of Their operations very quickly early on on Airtable and then just grow with the company. Right. And so like, I mean wework was one of our early customers, had like, like probably 10,000 people. Like, you know, when they were at their peak, like, it basically was like used by every almost employee there. And like a lot of their operations, building operations, et cetera, were just built on Airtable by default. And I kind of learned the. The value of like having this like data gravity. Like once you get enough data into a product like Airtable, like, like it just kind of retains really well within the company and gains more and more usage. Yeah. How do you think until the company. Well, you index against the industry that you're in. So I want to get to all the good part right now and all that stuff. But walk us through. Since this is your first time on the show, you went from being one of the hottest companies in tech during the whole no code boom, the like PLG boom zerp. Like it must have just been, you know, I mean, an insane experience. And then like there's kind of this reset in late 2022, 2023. How, how has it been kind of like building out of that trough and then like, have you. I'm assuming it sounds like you've been like very re. Energized by this new, this new opportunity. I mean, one of the baby benefits of like not being an overnight success because we took like two and a half years to build the product. Sure. Even like from 2015 to 2017, 18, I would say, like we were getting like a steady compounding of growth, but it wasn't like Slack or like Dropbox where it just overnight became super easy. Right. Like, it felt like we had to really grind. We had to think about, like, how do we need to like improve the product and increase the, you know, kind of like shareability and the scalability of it. So it's kind of a grind for like at least the first five years. And 2018, when we got our first unicorn round, is kind of the first year where it felt like it was starting to get easy. Right? Yeah. So 2018 to 2021, like very fun, easy years. But also everything is so good. This is just printed in the world. Unlimited money. And like, you know, we got to raise like a big set of rounds. Like how would you like 100x revenue or. Yeah, I mean, yeah. And just like the absolute scale of funding was like huge compared to prior art. Right. Like now, I mean you can raise like 100 billion. You know, like if, if you're opening eye. But like at the time like you know we raised, you know our first, our first unicorn round was like $100 million round and we raised like another like couple hundred and then like a few hundred more. And then like our big round was our Series F which was like kind of at the peak of like the markets raised 700 plus million in that round and 11 billion, 11 billion valuation. And you know we still have like all of that money on the balance sheet and we're now like cash flow positive. So I think like, you know, it's kind of a fun, fun, fun time to like, you know, kind of get to like ride that wave and then you know. But like I always, I think for myself knew like you know you have to build like a durable business, right? And so like valuations are going to like rise and fall. It's just going to be like macro. But like you know, ultimately either we build a great enduring business or we don't. If we don't then like you, you know, you could be like a flash in the pan, right? So I think we were always like trying to focus and I tried to focus on like what do we actually need to do to like you know, compound growth, like go after the enterprise. Obviously at the time especially like it was clear that was like the move, right? You get plg, but eventually you have to go into the enterprise and win like these big multimillion dollar contracts like become a really sticky system within these larger companies. And we did that and like we're still doing that. We have like a bunch of the Fortune 500 like running really critical operations on Airtable. Whether it's like content production at a big media company or like you know like fund operations at a company like you know like financial services company. So these are like the like almost like modern erp. Did you hire a different set of individuals to work on that that were already connected and knew that flow or was it something where like your best sales reps just sort of got bigger and bigger and leveled up? It was a, it was a little of both. I mean I think it's a different muscle. Like I think Rolodex selling is even at that time like you know, not that effective. Like I think like just knowing somebody at a big company like doesn't even if you're like you know, very senior and they're very senior like doesn't actually help that much. Like we've had some reps come in like they've had like a decade long relationship with like you know, the CMO of XYZ company. Sure. And I think that gets you like a phone call, gets you like a meeting. Yeah. But ultimately like buyers are wising up. Right. They have been for quite some time where it's not just like, oh, I know this guy, I'm going to like or you know, gal like I'm going to buy this like product from them. Like you actually have to like show them why this is going to help their, you know, help them in their job. Right. And help the company. And so I think it became much more about like transitioning from like oh, people can just use it on their own and they'll figure it out on their own to like starting to do more of a consultative sale, like come in and say like okay, how can we solve like a really big problem for you? Yeah. And maybe like for one company it's like how do I consolidate like my end to end operations for like how we do all of our brand planning, launching new products, all that. And that's kind of like it's like one part consulting, one part like just thinking about like a big enterprise scale solution and then one part like be able to leverage the flexibility of our product almost like in a Palantir like way to show the customer like we can actually solve this really deep problem for you quickly. Is there any sort of like PLG Motion or Land and Expand that happens in the Fortune 100? Like yeah, because there's some small team inside of Coca Cola or something that's using airtable and then you're able to use that as a demo or jump point. That does happen. Yeah. I mean it's like there are some companies where you just can't even get your foot in the door without the top down. A lot of big banks, like we just, we were firewalled out until we got some. Oh, interesting. So they can't even, they literally block you. Right. Okay. Your IP is like blocked. Okay. It might be like hard wall, like request access. So you know, there are some companies where you have to come in top down but like they're you know, I would say 70 plus percent of our current enterprise accounts, including the ones that are like now like five plus million in revenue, like originated from teams within the company organically adopting airtable. Right. Sometimes it was kind of like shadow it. They just figured it out on their own and they just like they showed real value from using the tool. Right. Like they would build some real operation on it and say like, well I've been waiting for like it to deliver me this like old bespoke solution or some like crappy other vendor for like two years now. I got impatient and just built the thing myself. And that's like a big, I think, you know, I think of like the enterprise landscape now is like, you know, there's plenty of dollars in enterprise, right? Even now, like, you know, just shifting from like traditional software to now AI, but like there's plenty of dollars, right? The budget's there, but like the question is you're not the only one going after it. And so like, what's your kind of asymmetric wedge to get in there and like take those dollars, right? And if you're a big company, like a Salesforce, maybe it's like we already have the distribution, we have like the customer data in there. We're going to go and attack adjacencies. If you're airtable, we don't have like the scale of like a, you know, ServiceNow or like an SAP or Salesforce. What we did have is like the usability of the product. So like the PLP was like kind of the entry point. And then also like even when we pitched to other people in the company that hadn't used airtable, they had, you know, probably heard about it from a friend. Like maybe the CMOs, like, you know, partner, like uses airtable in their company or we can go in and just show them like a really compelling demo quickly. Talk about AI. What are customers demanding? What have you rolled out? Where does AI fit in well? Where does AI take a back seat? Yeah, I mean, I think it's crazy because it's like we've seen so many layers of disruption happening almost in parallel, right? You think about desktop to mobile, it was a single form factor change, kind of easy, almost execute on. That was the one that I experienced at Salesforce. The big thing at the time was like Mark would tell every team, like, show me the mobile UI first before you show me like the desktop ui, right, like the mobile first. Right. And you know, it was like the right move and also kind of a simple move. Now it's like you've got at one level, like obviously every product should have like AI in it. So, you know, we have the obvious stuff, like you can now talk to airtable's assistant, like copilot style and have it do stuff on your behalf. In the product we have what we call field agents, which are kind of like the ability to map, reduce AI calls against like all of your data. So you have like 20,000 customer records and run like, you know, AI agentic like you know, kind of tool calls like search and like research about the company. Like synthesize, like hydrated bio. Yeah, like exactly that kind of stuff. One row. Yeah. And like you know, we do all that stuff but to me like the more interesting, you know, kind of disruptions underneath that are like one like you know, are people. Do people even want to come into your interface anymore? And this. Yeah, that's what I was going to ask like why you care about like storing the data in, in a. In a sort of safe, secure. Yeah. Force. They just went like headless recently. Like. Yeah. Is there a plan for that or how are you going to. Yeah, I think it's like, I think the right move is like hybrid headless. Right. Like, I think, I think the whole look like if you wanted just like a backend database you could use like Postgres, like Supabase. Right. Like you know it's. And you know there's like PHP my admin equivalents, like modern day ones. Right. Like Prisma has its own version of it like that are okay or they're good. Right. Like, but like I think what most people actually want, especially in a business context is like you want like the database but you want to have like proper permissioning, you want to have proper collaboration and most importantly like you don't want to exclusively interface with the data through like an agent. Right. Like you want to do that a lot of the time but like it's really helpful to actually go in and see the, like see the actual data. Like I think of it as the equivalent of you know, even though agentically you can like generate all your code and you should as a frontier developer. Like does that mean you never want to inspect any lines of code ever? Like no. Like you still want to see like a diff of like all the actual like code files change, whether that's in your IDE or in GitHub or whatever. And so I think the equivalent here is like you want to be able to drop down into a really nice interface and we've done some work around like kind of figuring out what's the best blend of the two. Right. So like with ChatGPT for instance, we have a kind of a first class integration where you can go in through ChatGPT and like interact with your data in airtable. Say like hey, pull me like all the customers that are like waiting for an outreach for me and, and pre draft like outreach messages. But then it can basically compose like a fragment of a view within the ChatGPT interface. So like you can actually see like airtable. Sure. But like, you know, kind of a part of the interface. So it's not completely headless. It's almost like you get to pull out like pieces of its face at the right time on demand. And I think that's a really important kind of like UX form factor. Yeah. How are you thinking about speed in the context of AI? I feel like the models keep getting smarter, but they also keep getting slower basically. And while I'm extremely confident that I could point a deep research agent at a massive airtable with 20,000 rows and get very good results, a lot of times I'm just in my email and I want to find one thing very quickly and that feels like it has yet to be, you know, AI ified or at least like LLM ified. It's very much, it's very, it's very much like, okay, well I should probably just fall back to SQL query or just some Boolean logic or just like vanilla search because I want this now. Yeah, I think both are going to be really important experiences. And obviously we have like, you know, kind of great like smaller and faster models. Sure. Like the mini, you know. Yeah, mini, you know, that, that are great for like more synchronous interactions and like within Airtable, like if you go to airtable or you use chatgpt with like one of the smaller models, you get that like faster kind of almost like more like real time experience. But I do think like a really important class of work that will come to dominate like every frontier company or company trying to reinvent themselves to be frontier is like figuring out how to operate in this new modality of like, you know, it's like the best developers today don't go and like sit there in front of their IDE and like synchronously like talk to the agent. You have like 30, you know, separate branches that are each being worked on by a different agent. And like you can have the agents continue to like update, you know, the branch based on human and other agent feedback. Right. So you have like comments back or like, you know, run like tests, etc. And I think this whole idea of like, look, it's going to take like hours for that entire loop to complete. Right. Like, agent pushes some changes, the changes get feedback from other agents or humans. Agent responds to that. Like that whole loop could be hours, not just like minutes. So you're not going to sit there and like watch it one at a time. But the powerful thing about this is like each one is still actually operating faster than like a human engineer could have like back in the day, right? Like when I think about like the speed with which like our early team at Airtable could build features and we had a very good team, like one agent on one branch can you know, do the work of like maybe three humans back in the day operating probably in like three times the time, right? So it's like literally like a 10x kind of leverage factor just for one agent. But the best engineers are now able to multitask and kind of basically say look, I'm going to oversee my own little team of like 20 to 30 agents working concurrently. And so I think it requires like, it's almost like everybody needs to graduate from being an IC to like an IC manager of agents. Meaning like in every function, like if you're like a VC analyst, your job should no longer be to go and synchronously research one company. It's like you're going to go and research like 3:30 companies and do them all faster, better and higher quality, right, like than, than what you could before. And so it, I think that's the greatest leap that is going to be challenging for a lot of people in a lot of roles to make the leap on because it's, it's a totally different mentality to like how you operate and what your role is than for what are you pushing the, the team to achieve. So a lot, you know, I think like there's basically three different levels of self disruption we're trying to do at airtable, right? One is the core product itself. How do we reimagine that for an increasingly agent led future? So all the headless hybrid type stuff we talked about and the best testament to that is do we see massively growing basically tool call volume from ChatGPT, Claude, any other agent products are people using Airtable more and more agentically and is it working smoothly for them? So that's like priority one. The second though is like I think we have to like really transform how we operate internally, right? Like you know, clearly like the companies, the best companies in the future are not just going to hire like massive armies of people to do everything, right? Like they're going to hire like people who can really effectively leverage agents, right. It's so obvious that's happening in engineering where like, you know, if you could hire one engineer who could be fully agentically leveraged, you get more output from than like 30 kind of traditional engineers doing traditional engineering. So that's kind of one internal thing. But then the third Is like I'm a strong believer that like you have to go and skate to where the puck is going, like index against like the big tidal wave coming, right. Like Amazon did this like back in the day against like the growth of the Internet. Right. Like you know, Bezos picked books and like you know, E commerce because like he thought that would be the best way to index against the growth of the Internet. And so for us like we get that through airtable and like kind of hybrid headless airtable. But we're also placing a big bet on Hyper Agent because Hyper Agent is basically like taking all of the like excitement of frontier agents. Like that is openclaw and like YOLO agents that can just like have access to your data and tools and do stuff like really, really like long running stuff, not like 10 second tasks but like 10 hour tasks but for non coders. And we want to do it in like a business friendly way. Right. So like you can go and like do this, deploy it into your company, run like agents across your entire company. And so that's kind of like a bet on if we believed airtable, you know, 10 years ago was like the most meta problem, like the largest problem we could work on, which is like software arguably is like was the biggest and fastest growing industry at the time. Right. And like how do we go after that entire category and index against that? How do we now go against agents and say like we want to build like the best agents platform that any business and any person can come in and use and just start building agents with. Right. And deploying them into their company. Right. And so if we do that really well then like we get to doubly win both as the data layer, but also kind of have a bet on the, the agent wave. Yeah. Last question. Should children learn C? Definitely not. No. I think. Is that because of C or because of the AI era? I think. Well, I mean I think both. Like, I think, I think the fundamentals of good technical architecture are going to be the most important thing. But that has like, I think the abstraction that really matters now for creating value is raising up right. Like it used to be at one point like you know, Bill Gates wrote like some of his first programs in like literally like machine code. Yeah. And like would punch it into his like PDP 10 and like clearly you could be a great startup founder or be a great software engineer and make lots of money like without having to go down to that level. And so I think now with agents like the bar has raised yet again where like what you really need is, like, good product, business, and, like, tech architecture sensibilities. Like, how should this system work? Like, where should the different, like, levels of, you know, kind of responsibility belong? And, like, if you can get really good at that, then you have super leverage. If you are just kind of like, like, learning the literal kind of, like, lines of code and how to write them that, you know, a lot of engineers were poor, I think that's going to be increasingly below the frontier line of, like, agents can just do it, like, equally or better to humans. Yeah, that makes a lot of sense. Well, thank you so awesome guys coming on the show. Thanks for having me. Have a fantastic rest of your day. Great hanging. Talk soon.