This Week in Startups - AI Producers, Tesla Robotaxis & the Rise of the “Tiny Teams” Era | E2142

Episode Date: June 24, 2025

Today’s show:Tesla’s robotaxi launch, AI producers, job destruction, and smart toilets? In this episode of This Week in Startups, @Jason and @alex break down Tesla’s cautious rollout strategy, how AI is quietly replacing producers and employees, why second-movers like Tesla and Ramp often win, and how Throne Science is turning gut health into a billion-dollar market—with cameras in your toilet. Buckle up.Timestamps:(02:11) Introducing… PRODUCER CLAUDE!(04:21) Tesla’s Robotaxis are out and about! But are they SAFE?!(09:49) Squarespace - Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www.Squarespace.com/TWIST(19:35) Netsuite - Download the ebook Navigating Global Trade: 3 Insights for Leaders for free at https://www.netsuite.com/twist(24:07) Grabbing market share, and why there’s “no reason to rush”(29:59) INBOUND - Use code TWIST10 for 10% o your General Admission ticket at [https://www.inbound.com/register](https://www.inbound.com/register.) (Valid thru 7/31)(39:53) AI, Job Destruction, and what leaders aren’t saying out loud…(48:43) Which startups are doing the most with the smallest teams?(57:49) Mark Zuckerberg has entered FOUNDER MODE(01:05:48) How Throne Science is improving everyone’s gut healthSubscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.comCheck out the TWIST500: https://www.twist500.comSubscribe to This Week in Startups on Apple: https://rb.gy/v19fcpFollow Lon:X: https://x.com/lonsFollow Alex:X: https://x.com/alexLinkedIn: ⁠https://www.linkedin.com/in/alexwilhelmFollow Jason:X: https://twitter.com/JasonLinkedIn: https://www.linkedin.com/in/jasoncalacanisThank you to our partners:(09:49) Squarespace - Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www.Squarespace.com/TWIST(19:35) Netsuite - Download the ebook Navigating Global Trade: 3 Insights for Leaders for free at https://www.netsuite.com/twist(29:59) INBOUND - Use code TWIST10 for 10% o your General Admission ticket at [https://www.inbound.com/register](https://www.inbound.com/register.) (Valid thru 7/31)Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarlandCheck out Jason’s suite of newsletters: https://substack.com/@calacanisFollow TWiST:Twitter: https://twitter.com/TWiStartupsYouTube: https://www.youtube.com/thisweekinInstagram: https://www.instagram.com/thisweekinstartupsTikTok: https://www.tiktok.com/@thisweekinstartupsSubstack: https://twistartups.substack.comSubscribe to the Founder University Podcast: https://www.youtube.com/@founderuniversity1916

As an example, when a founder sends us an update, we have a tool that reads the updates, looks for the important data in it, like, what's their burn? What's their revenue? What's their growth? And then try to map those to fields in a database and write summaries. Founders will write us, God bless them, 1,000, 2,000 words, of which we have to scan that email and get the top five or six data points. And if those data points aren't in it, the next phase of that tool will be to reply to

the founder and say, hey, this is the launch AI bot, can you give us the last three months of revenue, the last three months of expenses, and then the last three months of burn, which is those two numbers subtracted from each other? And can you tell us to head count? Because we don't have head count in here. And can you tell us the cash in the bank? So we need to get those numbers to anticipate fundraising. And we are going to have the bot in the next iteration, not just take the update they send us by email typically or sometimes they'll send us a Notion page or a doc you send. You know, we're going to do it that way. So that is, I think, the big innovation here is that an outsider is going to do it. Be scared, folks. If your job is easy to do, be scared. I'm inviting

you to be scared. This weekend startups is brought to you by Squarespace. Turn your idea into a new website. Go to Squarespace.com slash twist for a free trial. When you're ready to launch, choose offer code Twist to save 10% off your first purchase of a website or domain. NetSuite. The business landscape is very chaotic right now. That's why you need NetSuite by Oracle. If your revenues are at least in the seven figures, download the free ebook, navigating global trade, three insights for leaders at netsuite.com slash twist. And Inbound. Connect with visionary leaders like Dario Amode and Amy Poehler at Inbound 2025, September 3rd through 5th in San Francisco, the epicenter of tech innovation, and transform your business strategy for the AI era.

Use code Twist 10 for 10% off your general admission ticket at Inbound.com slash register. Valid through July 31st. All right, everybody, welcome back to this week in startups. Three days a week, Monday, Wednesday, Friday. We discuss all the top tech news, all the private companies that we track at Twist 500.com. We have 500 private companies. I think we're at 350, Alex. Where are we at in total numbers there?

Do you want me to count the ones I'm adding today or not? Doesn't mean. I have 45 new names. So we're about 415. Perfect. So we're trying to track at twist500.com, the top 500 private companies so that we can have our new producer. We, you know, this is a big long game.

I've been trying to get this Twist 500 done for a year almost. And we now have producer Claude. Some of you know Anthropic has this incredible. large language model that does incredible things, agents, coding, it does everything. It's got a big open context window where you can put a bunch of data into it. So I was talking to my team. I said, I want to create a partnership, Alex, with Claude. And I want to have a producer Claude.

That goes through those 500 names that we have in Twist 500 and then looks for news, looks for sentiment, maybe companies we've missed, and then just gives us ideas. on how to make the show better for founders. My premise is pretty straightforward. I think the job of being a producer on a podcast can be done by AI better than a human. I think we'll be there this year. In the next 12 months, I think AI,

watching the show, looking at other news, reading comments from our listeners taking feedback, will be able to produce a better show than a human, why it's going to work 24 hours a day. Now, humans will still be in the loop, obviously you and I are still here for now. but who knows. So thank you to our friends in Anthropic.

We've got a great partnership with them. They're helping us with all these tools. And we're going to call on producer Claude, you know, like once or twice a week to help us with the show. But really what I wanted to do is get attached to this Twist 500. We built with Coda. Great database, all the private companies. Okay.

So I have been tinkering with this. And I was playing with Claude, had a good time, was trying to ingest the whole Twist 500 and pull out all the news. And Jason, enough babble, enough introdial. because today is the day we can finally talk about the launch of Tesla's Robotaxi service. Yes, it went live yesterday in Austin, where you putatively live, even if you haven't been there in a minute. Yeah. I think this is a huge deal.

Elon Musk tweeted out super congratulations to the Tesla AI software and chip design teams on a successful RoboTaxi launch. He called it the accumulation of a decade of hard work. It says the AI chip and the software was all built from scratch inside of Tesla. Just first thoughts from you. I'm really curious what you're thinking. Well, you know, I watched it. I think it was a successful launch. It's super, super conservative, which is what I think any of us would want any entrant to self-driving to do, whether it's Zooks or whoever, because we did see, even when you are super conservative like Uber was many years ago and they had a tragic fatality where a safety driver was not paying attention, they were in fact, I can't. Wendy crush, like unbelievable negligence on the part of an individual. And then you had crews, which that program has been shut down. They were bought by GM. They dragged a passenger on the ground. This is a serious business. And Tesla's autopilot has been abused and FSD has been abused by many customers. There have been many accidents. Now, listen, there are many accidents in all cars.

But if we're going to have them unsupervised, we take the steering wheels out, we hit level four, autonomy, level five autonomy. It requires being conservative. Why? The second you make, a mistake, the public, the haters, the vigilant, the regulators, consumer sentiment is going to be on top of you. I would say 99 out of 100 rides will be flawless. One out of 100 will have a moment, just like Waymo did. That's why Waymo had safety drivers, I think, for five to 10 years. And they still do. Every time they go to a new place, they do safety. So I think there was one incident yesterday that was, you know, I wouldn't say like horrific, but it was, you know, one of these instances that, you know, anybody who has FSD will have, where, you know, FSD has a hard time making a decision, which is why they have a safety driver.

So we have to outline what constraints the Tesla team put on this. Number one, there is a safety monitor, not a driver. What's the difference? A safety driver would be in the left hands. seat, a monitor is in the right. The right hand driver has like three buttons on the dashboard. One of them stop and lane. One of them is pull over. I don't know what the third one is. And then on the right, there was a note that they had their finger, their thumb over the door handle the whole time, which I think is like an emergency button. So you'll see online a bunch of speculation,

like, why do these safety monitors in the right hand seat have that? Second thing they're doing, They only allowed like super fans of the product on there. So they gave early access to like all the people who tweet and do FSD videos already. So you got like a hometown crowd, smart move. And I think the safety drivers also, so somebody, the safety monitor is so somebody doesn't like jump in the front seat or do something stupid, grab the steering wheel, try to hit the pedals. I'm assuming that doesn't work if you grab the steering wheel or you do that. I'm not sure. I was curious about that.

So I went through the entire FAQ for this because I was curious, what are the rules. about just trying to help. And what they said is, you are absolutely not allowed to sit in the driver's seat. So I think they would really rather you not touch anything because I think even if you can't influence it directly, it moves the wheel on its own. And so you could grab it and stop that and probably screw with the AI. So I think it's more of like a just cordoned off. No one gets to sit the driver's seat, which I think is imminently reasonable.

And frankly, if you try to ruin your AI car by messing with it, you deserve crash. So, yeah. So, and I wonder in a Waymo what happens if you jump on the steering wheel, you know, and try to. Watch this. Producer Claude. Yeah. Can you figure out what happens for us in a self-driving car if you grab the steering wheel and give it a yank while it's self-driving?

Because I think that should be something that they've sorted out as an edge case. Oh, for producer Claude is added already. And it says here, if the passenger grabs a steering wheel while the car is autonomous mode, several safety mechanisms would likely activate. I guess this is its best estimation of what would happen. The system would detect the steering input likely disengage. That's what happens when you're driving an FSD and you're in the seat. Control, transback to manual driving the car would alert the driver, passenger with visual audio warnings.

Robotaxy, the system might allow limited steering input for emergencies. I guess this is Claude giving its best estimate of what would happen, but I don't see citations here. So more work to be done here. We need to figure out what should happen or what specifically. happens. And then Waymo has, yeah. Similar, similar rules. Now, just as a reminder to everybody, Jason's right. This is a relatively limited launch. It is still invite only. It is occurring in part of Austin. Rides are a $4.20 flat fee. Have to use the Tesla app, 6 a.m. to midnight.

No under 18s. And this is the coolest thing that I saw that I didn't know so far, Jason. you can change your destination right now in the app mid-ride, which is pretty cool. I didn't know that. But if it's a flat rate, they're not charging you per mile or per minute. So why not? All right, founders. Let's talk about your website. I know.

Disgratia. You're ashamed of your website. I know. Well, it's time to clean it up. Give your brand a quick refresh with Squarespace. That's the all-in-one platform that makes building a stunning, professional gorgeous website. Ridiculously easy. Doesn't matter if you're selling products. Doesn't matter if you're offering services. Or, you know, if you're just showcasing your portfolio, SwareSpace gives you everything

you need to grow. They've got this great new AI product that's called Blueprint. You got to try it. You just answer a few questions and you get a fully customized website in minutes, personalized layouts, on-brand visuals, and voila. You're done. I've been using this product for over a decade. Check out Squarespace.com slash twist for a free trial. And when you're ready to launch, go to Squarespace.com slash twist to get 10% off your first website or domain purchased. Once again, Squarespace.com slash twist. So I think, you know, a couple of videos would be great to show. There's one video of, you know, like an influencer taking a ride.

Sped up. I'll talk over it when you play it. It's, I know the area, if you can get it playing. There is an area. in Austin, which is called South of Congress, or South Congress, just South Congress, I guess, would be the way to say it. Below Lady Bird Lake, it's a hip part of the city.

Lots of, like, Soho houses down there, lots of great restaurants. It's not the main part of the city. It's not the downtown area with all the tall buildings. This is like the hipster area. You might think of it like Soho or Brooklyn. Sure. And here you can see cars driving around quite flawlessly,

and they're on the right, obviously. is the safety monitor, not a driver. So this is kind of like halfway between, I guess, how Waymo launched or cruise or some of the other services. But anybody who has FSD hardware for will know this experience, which is, you know, like I said, nine out of 10 rides, hundreds of miles,

it doesn't have an intervention, is the latest statistic. And then it lets you do a rating, and then it offers you to do a tip, And then if you try to tip the robotaxies, it says, just kidding. So that's, okay, that's funny. I like that. I was going to say, like, are they so cash strapped? They need to have tips going into the central self-driving pot?

That's a good joke. I like that. Do you want to watch the short clip, Jason, of the mistake? Sure. Yeah, I think this is how you're going to be judged. So this is a sort of important lessons for founders. You're going to get, you could do a thousand things right in your startup. And you're going to get judged by,

Your worst moment. It's unfair, but it is what it is. You know, if you're a restaurant, you seat somebody late, that's the first person who writes your help review. If you're a SaaS software company and somebody's using a really old computer and they lose, you know, they don't save their work as they go or they got a terrible internet connection. You're responsible because they lost their document or something. Just the nature of technology. So assume that you have to be responsible. But, you know, you can hit play here.

and what you'll see is it's supposed to make the next left turn. And here it realizes, oh, I was making this left turn. That's the wrong one. Let me go try to make the next left turn. It doesn't get into the right lane on the right side of the double yellow line. It stays in the opposite lane. So it's going in the opposite lane of traffic. That's obviously not good. So this is why you have a safety monitor. The question is what happened here during this, which is a significant mistake. This would be a significant disengagement. And I know this because I drive this since the beginning, and anybody who drives it, even the biggest Tesla fan, will say there's a disengagement every couple of hundred miles. There's tons of statistics on this. And that's why this is a beta trial. It's a

closed beta trial because they know that they're going to have these every, I would estimate a couple, low hundreds of miles, I think is probably the statistic. I know when I drive it on my hardware three, I get probably two of these an hour when I drive it on hardware one, hardware four, the latest one I have, I have another model Y with the latest. I would say it happens once an hour so far,

so once every 50 miles or so. That's not bad. That's not bad. Well, it's not enough to be a robo taxi and it's amazing for, you know, augmented, you know, self-driving to help me. So in other words, we're not in a position

where you can sleep in the car. No, no, we're not. What will happen is these will be documented. And I wonder if the safety monitor intervene there with that button on the right, if that is in fact what's happening. And you can look up online people talking about the right hand door passenger. And you'll see pictures and discussions of that online if somebody wants to do that. I'm going to leave that to producer Claude.

I want to show you something, Jason, from the Robotxy launch that I thought was incredibly cool. So from an AI engineer on the Robotaxy team, there was this image that was shared. shows the team with the launch. And if you're on the audio version, it's kind of like the NASA room where they have all the screens and computers. In this case, though, no telemetry from a rocket Jason. Instead, it's a lot of information about the Robotaxy stuff.

But here's the fun thing. They may have shared a little bit more than they meant to because if you look at the screens, there's a really interesting bit of data. And I took a screenshot of that and zoomed in for us. And as you can see, at this time, they had done 112 rides, 499 miles and then 35 is this mystery number because I cannot read what users in the system.

Yeah. Users in the system. 35 users, three rides each, four rides each. Okay. Makes sense to me in first day. I'm going to go with users. Members or active unique users for the day. Ah, I like that.

I thought it might be disengagement, but I think active users probably makes more sense because folks were taking several. There was the one guy who took 11 rides. Yeah, I think most of them took two or three, it seems. like, and it was open at night. It's open from six to midnight. I'm going to go with, yeah. And so the average ride was four miles. Average person does four, maybe three. So that's, that wouldn't, that would happen for me. The other thing you see there is there were, I think, of those 10 monitors or eight monitors, however there were, maybe four or five of them were actual cars and the five or six cameras around the car. So if you look,

bottom left, the second row top, third row top, fourth row top, and then the fifth row, so there's 10 monitors, bottom row. And those are six cameras, I guess, front and sides, back and sides of every car. So you can watch every, if they do have five cars on the road, I think they have more, you're watching every single one. There was another picture of the engineers and they had steering wheels in front of some monitors. Now, I don't know if that was for a remote takeover of cars and they use a steering wheel, or that was like just maybe they had steering wheels and they were doing testing previously. I think that was from a Chinese self-driving. It may have been a buy-do as Apollo Go.

Oh, no. People were talking about it being for Tesla. Oh, oh, okay. Well, then I fully misunderstood that. I would have pulled it, but I saw that as Apollo Go, and I was like, well, that's not quite the story. Yeah, so they have remote drivers. So the question is, in that video, when it's trying to make a decision and you see

the line of what road to go on, flip back and forth, I was trying to determine, I wonder if that's the remote driver taking over. and if the hand on the right, so if somebody has that screenshot, Lon, you can just search for it on X, the button on the right, the door open button, if you search for door open button, you'll probably find it. The safety drivers, the whole time there's somebody in the car, have their thumbs over the right hand door. Now, if you're driving a Tesla and you press the door open button, it will not open. So the theory that people are speculating here, and I'm sure Tesla will confirm it at some point, is the safety drivers are intended to have their hand over that button.

If they have their hand over that button, then they can very quickly, without touching the screen and having to look for the button, they can just do it instinctually, keep their eyes on the road, press that button and stop or pull over, or maybe ask a remote driver to take over. So here you see on the right, the safety driver has their thumb over the door release button. So they could easily have reprogrammed in Robotaxy for this.

test that right button. If they did that, that would be incredibly clever to click it when there's an issue, have it alert a remote driver who's waiting, and have them take over the car. By the way, remote intervention, Waymo has it, Zooks has it, all the self-driving robots everybody has it. So the question is, when do, the question now becomes, if you really want to understand what's going on here, is when will this be good enough to take out the safety monitor? I'm going to guess six months, 12 months, somewhere between six and 12. 12 months, the safety monitor can come out. And then in some markets, they might say, you know what, safety monitor needs to be safety driver, moving from the right-hand seat to the left-hand seat.

And the local regulation will be local regulation, and there are going to be local regulations that say, my guess is you have to have LIDAR. That's the standard. There'll be other jurisdictions that say, hey, Tesla's approach is just fine. And as a startup, you, when you're releasing a new product, have to deal with customers, press, and regulators. All of these folks are going to be giving you a real thorough examination here. The business landscape is so chaotic right now. You've got tariffs, you got supply chains. Everything just keeps shifting constantly.

Your company needs to adapt in real time to all these changes. So to do that, you need total visibility. That's why you need NetSuite by Oracle. AI powered business management suite. It's trusted by over 42,000 businesses. You have the visibility and control you need to make quick decision. For example, automated payment tracking so you can see which clients are up to date and who's overdue. Revenue recognition, automatically tracking everything to make compliance a snap. Lead to customer conversion and CRM followups so you can make your funnel tight. And with AI embedded throughout, you can automate a lot of those everyday tasks.

freeing your team to the more important higher-level work, one system giving you full control so you can tame the chaos. And all of the startups we know that have gotten to scale use NetSuite. So here's your call to action. If your revenues are at least in the seven figures, download the free e-book, navigating global trade, free insights for leaders. At netsuite.com slash twist, that's netsuite.com slash twist.

The press, I think, is mixed on this. They, they in some ways are going to be skeptical and some number of them don't like Elon because of politics. The regulators, I think, are going to think about safety. Some of them might be politicized as well, but generally, they are risk averse. And then customers, there'll be some customers who are not fans of Elon, and then some customers who are super fans. So the data we're getting right now is super fans and a jurisdiction that is pro self-driving and, pro innovation and a press that's a mixed bag. So what you have to do when you sort all this out is examine the videos for yourself. And if you examine, you watch these videos, understanding they were taken by super fans, it's still very impressive. They're able to do the majority of the rights, overwhelming of majority of rights without an intervention. That doesn't mean you can put 100,000 of these on the road, because I know this, having been an early investor in Uber, you might have heard, regulators play a big role in all this. Regulators play a big role. But one more data point, Jason, that you didn't mention. And it's our favorite one. What does the stock market think

about this? Well, the stock market is pretty freaking excited. Tesla's up like 10% today. Keep in mind, guys, for a stock trading around the trillion dollar mark, that means $100 billion in fresh value. That's a lot of, that's an enormous amount of money. That's a third of Open AI in a day, because this went well. So there was quite a lot writing on it. And I think it went better than I expected. I mean, this is hard. This is a hard challenge.

There's been no collisions. No one's gotten hurt. One or two small mistakes. But I mean, like, I mean, we didn't expect it to be perfect from day one. Because then if it was perfect, why have all the, uh, the most, yeah, the most cynical take you could have is, oh, my God, on the first day. And I saw this from like super cynical people. Oh, on the first day, it missed a turn and drove into oncoming traffic.

Technically true. But it's, this is not without. a drive, without a safety, without safety monitors slash drivers. And so, you know, you're going to have these. Waymo must have had these when they had safety drivers. You just didn't hear about it. And by the way, I've been in Uber's and lifts where they've done, you know, much more dangerous things than that one instance. You could be like, oh my God, on the first day, they already had one of these. This is going to be the nature of it. Every city they go to, they're going to have hundreds of these interventions that'll be necessary. The question is,

how quickly can you get rid of the edge cases? Yeah. And then the big one's going to be the LIDAR debate. I predict that the LIDAR debate now becomes real. If there are more instances of interventions during fog, rain, nighttime, whatever, fog, and, you know, things where LIDAR might be able to see better, then cameras, that's going to reinvigorate that debate. Regulators are going to get involved in that debate.

I think it's going to be 50-50. Some jurisdictions are going to say you have to have lighter. Just a prediction. And I don't know which ones it will be, but I think the ones that are most conservative will say put some LIDAR in it and, you know, that'll be an easy concession for Tesla to make, I think.

I think it'll be conservative in terms of like their approach to self-driving, but also just weather. Like I bet you where there's a lot of snow, a lot of ice, a lot of fog. Like if you're in Minnesota in the winter, I hope you have every type of radar known to man because it's a mess. I mean, it's like, you know, 15 below zero and snowing.

So there I would like everything. In Austin, camera sound fine. So I wonder if there'll be some just sort of geographical differences as well that come into play. Question for startups though, Jason, because I'm thinking a lot about Tesla here, joining the market a little bit late, kind of like how Ramp did compared to Brex. And Ramp was dismissed originally as kind of a clone, a copycat, a Johnny Come Lately. and it has absolutely stormed the gate, and now I would say ramp is ahead of Brex.

So being second is not the worst thing in the world. Tesla's come here after Waymo already reached scale. What's the best playbook for founders who are coming into market, not first but second, and want to grab market share and really not fall prey to incumbency bias in the market? It's a good question. I think if your product or solution is better, you can trump it why and, you know, compare the two. you don't have to, and you can beat people on price, performance, speed, you know, many different vectors depending on, you know, choice. So if you're Uber-Eats coming after DoorDash or Postmates

coming after DoorDash, whatever it is, or DoorDash coming after Postmates, having an exclusive, being in a region, they're not. So Lyft's specific strategy versus Uber in the early days was to go to the cities where Uber wasn't. So why even bother competing? I think, that's actually might be a really power move by Tesla is to say, where is, where isn't Waymo? A Waymo isn't in these Florida cities. Great. Florida is a freedom state. What are the freedom states, you know, perhaps right leaning and which ones have, you know, don't have Waymo's yet. That might be the zig where they zag and be the first mover in those markets. Rhode Island. Rhode Island, sure. I think the Northeast will.

be one of the last. I know. Yeah, anything, anything with snow, I think, is going last. Anything with roads that are not a grid system. So roads that were created in the 17th, 18th century versus the 19th, 20th century. Dang it. You know, which would be Europe. You know, like, you want this thing in the back streets of some southern France town or in Spain, you know, twisting around. Like, those would be the final bosses. Yeah. And then a town with inclement weather. and old roads, if you took those vectors, when was the city designed? Right?

Is it an old city or a new city? And then perfect weather versus not perfect weather. It's why California, Arizona, Texas, a lot of these cities expanded in the modern era with grid systems, planned communities, wide roads, there's plenty of space, and it rains 10 days a year. And there is a chance these will pull over

and not be available in certain weather conditions. Just like when you have a blizzard in New York, don't expect to get a Uber because they don't have snow tires. And they're just not available. And everybody in New York has adapted to that. If it's a snowstorm and you take the subway, that's it. Your choices are limited. I have a message here for the CEO of Tesla,

which is if you can put a rocket up and catch it again, you can make this work in Rhode Island. Come on down to Providence. I'll be your first tester. I would love this. Someone, please come to my show. city, Jason? The other big picture, I think, here to look at is there's no reason to rush. Sometimes people, startups and founders get a little bit edgy about losing market share and having

to catch up and they think they're in a race with a competitor when, in fact, you have to actually really assess where the race is. The race we're talking about here is car ownership versus non-car ownership. It's not a self-driving race. So let's just pause and understand the game you're in. There's a really interesting concept of finite games versus endless games, right? Finite versus endless. A finite game, there's a winner and a loser, right? And you play and it's miserable and you don't enjoy it until after it's over and you've either won or lost. And when you win or lose, then you have, you know, this moment in time where you feel great and then, you know, afterwards, it kind of sucks.

You know, other games like, you know, when I go skiing, skiing is an endless, open-ended game. It's not a finite game. I can ski many different mountains around the world. I'm not skiing against anybody. And it's the enjoyment of the game. Here, if you think you're in a race, it's going to be miserable. If you're, you know, it's Waymo versus Uber and Lyft or, you know, Waymo and Uber versus Tesla versus Zooks, what you're really in a race here in this infinite game is car ownership.

And in that way, everybody's on the same team. Zooks, Uber, Lyft, Tesla, they're actually all on the same team. And right now that team has one point. The other team has 99 points. So now we get into the finite game. It's an infinite game. People can take an infinite number of rides and will if the price keeps going. down. So if you frame it like that, then it's car ownership. And Tesla becomes a very interesting

company, right, because they are still selling cars. So car ownership matters to them, but they also see the future and they're disrupting themselves. So now you get to the concept of the innovator's dilemma. The more successful Tesla is, the more people need to buy their cars. Inbound, 2025 is almost upon us featuring a lineup of visionary leaders and personalities who are going to going to transform your entire approach to doing business. There are lots of big names in this lineup, including Dario Emote of the breakout AI company and Twist 500 member Anthropic. We use Claude every day. This company is making amazing strides. And you just know, Dario's got some great insights to share at the event. So that's going to be an awesome keynote. And this is how you level up

as a founder by learning from the people who came before you and soaking up all that experience and just downloading all that wisdom. We talk about the all the time. There's no substitute for being in person. The era of work from home and from remote hiding in your bedroom, that's over. It's time to get back out there. So if you want to connect with visionary leaders and personalities like Dario, Victor Riparbelli, Darmeshaw, my friend, and more, you have a chance at inbound 2025. That's right. The epicenter of tech innovation is happening this year from September 3rd to 5th in San Francisco, California. Use the code Twist 10 at checkout to get 10 sent off general emission tickets at inbound.com slash register. That's inbound.com slash register.

So one side of their business, the balance sheet of buying cars with young people might go down because riding in cars is going up, riding and ride sharing. I think we're going to get to 20% of rides will be done in a robo taxi, a Zook, a Waymo, at Uber, a Volkswagen, a Tesla. within 10 to 15 years. I think it could be 20% of rides. And if that happens, that means you need much less cars. Many fewer cars will be needed.

Many fewer will need to be produced, which means that an environmental basis, you know, I think we make 50, I think I make 70 million cars a year globally. And I'm pretty sure, it was a good job for producer blood. I'm pretty sure, you know, Toyota, and Volkswagen or, you know, your main players there

and the Chinese are coming on strong. I think they're exporting about five million cars a year now, which- 94 million cars made per year as of 2020. Okay, okay, great. Wow. I have all data. It was 70 million was my old number. So yeah, let's call it, we're, you know,

getting close to 100 million cars a year. You're, for each percentage point of ride sharing, you know, you're talking about 10 million cars. It's gonna be a lot of cars need to be dedicated to this, but at some point, there are, enough. So if we all of a sudden could redirect all hundred million towards ride chairing, I think it would be the end of car ownership. But we can't. There'll probably be one million or two million of these created per year out of that hundred million. And then it will get to three or four

and five and six million per year. And we'll be off to the races. Volkswagen made an announcement last week that they're going to sell ID buzzes with self-drive, full self-driving to anybody who wants to buy them. This is a real shot across the bow of Tesla, the entire industry. And then Waymo announced that they're giving their software to Toyota. And you'll be able to buy a Toyota Robotaxi. So now you'll have really three players you can commercially buy as a consumer a self-driving car from. So you could own it and sleep in it. Very interesting how this chess board's going to go. But again, the great part about this game is everybody who is on the side of non-car ownership has an opportunity to win big. Let's move on, Jason. You wanted to talk about this

really interesting job posting from our friends over at Cora. But if you don't know what Cora is, it's a well-known online question and answer service, Jason. But the second act of the company most recently is a thing called Po. It's a service that allows you to interact with a great number of different chatbots. You just can pay between five and $250 a month for Poe to get more credits to use more AI models. Seems to be doing quite well. But we're curious today about a new job posting from Adam Dangelo. And this is for a new role that he says is, quote, a single engineer who will use AI to automate manual work across the company and increase employee productivity. You thought this was incredibly interesting. Why did it catch your eye?

I think somebody's going to figure out how to manage an entire company by AI. In other words, AI CEO. So here we are with producer Claude introducing it this week. Hey, this is an important role. How can we train the AI to be a real-time producer? In fact, you know, my long-term goal for producer AI is to have it listening to the show. A year from now, I'd like producer Claude to be able to interject and maybe put data points on the screen without us asking. So it anticipates. So I say, what's the market cap of Volkswagen? All of a sudden, just producer

Claude comes along in the bottom and puts it there. Eventually, it could be an avatar. It could talk to you, but maybe not breaking in, but just slow and then fast. So when somebody like Adam, who is deep in the AI game, and Cora, my understanding is they're doing a ton of licensing and training data. They're in a really good place. And we saw a scale AI bought by Facebook doing like a training data company. I think Cora kind of falls into that category as well. When you see the founder saying, I need somebody to examine every single function or role and have it, you know, attacked by AI,

relentlessly ripped apart and automated. This goes back to the blog post I wrote, automate, delegate, deprecate. And then I wrote a blog post this weekend on Sunday, I shipped it on Saturday or Sunday. about just, hey, we don't talk about job destruction anymore. So here when you look at the responsibilities, develop, and maintain inner tools and systems to automate existing work

and increase employee productivity, that is the high order bit. What that means is, what are the employees working on that are chores, that are not the main thing that we can take away from them, away from them. And instead of having employees do this, you have an outsider, an agitator coming in and saying, I looked at what you do, and I've automated it. So an individual will never automate themselves. Why, there's going to be some natural fear that they'll automate themselves out of a job.

They'll be, they're too busy to do it. They'll be they lack objectivity. They think that there is something special about what they do, right? And so the unique thing about this is that they are an outsider, watching people work, and eliminating work, taking things off their plate. So I anticipated this. I've been talking about it. And when you see CEOs do this and then you see the reaction, I don't know how many millions of views that tweet got, that post on X got, but I'm going to guess millions, like two or three

million people probably viewed that. Because I saw a number of other CEOs quote retweeting it like I did and saying everybody should have this position internally. And in fact, last week I asked the team to start thinking about a growth and automation position here at the firm. So instead, because not everybody knows how to build these tools or likes building them to just look at every function and figure out how to grow that function and make that function more automated.

As an example, when a founder sends us an update, we have a tool that reads the updates, looks for the important data in it like, what's their burn, what's their revenue, what's their growth, and then try to map those to fields in a database and write some. is founders will write us, God bless them, a thousand, two thousand words, of which we have to scan that email and get the top five or six data points. And if those data points aren't in it, the next phase of that tool will be to reply to the founder and say, hey, this is the launch AI bot. Can you give us the last three months of revenue, the last three months of expenses, and then the last three months of burn, which is those two numbers subtracted from each other?

and can you tell us to head count? Because we don't have head count in here. And can you tell us the cash in the bank? So we need to get those numbers to anticipate fundraising. And we are going to have the bot in the next iteration, not just take the update they send us by email typically, or sometimes they'll send us a Notion page or a DocuSend. You know, we're going to do it that way. So that is, I think, the big innovation here is.

that an outsider is going to do it. Be scared, folks. If your job is easy to do, be scared. I'm inviting you to be scared. So to answer your earlier question, Jason, 1.4 million views on the tweet. So that's a lot on X. Yeah. And also, I just thought your point about not everyone knows how to use these systems was actually really dead on because one thing that surprised me when I was reading the requirements for this job at Cora, they want five plus years of, quote, experience in full stack element with strong skills in Python, React, and JavaScript, and they want to experience creating LLM-backed tools. So this is not a job for someone who is just an AI enthusiast.

This is for an AI engineer, which I think is where we are today, probably in a year. It'll be easier to fill with a lower technical requirement. But the thing that I'm really curious about is what stage of a startup is the right one to bring someone like this on? Because if you're two people, you don't need one. If you're 2,000 people, you should have one. So, Jason, just in your thinking, when is the startup big enough that it should have someone dedicated internally to beating back at cruft and delay?

The founders should know how to do this. And when you get to, you know, scale, maybe 10, 20, 30 people, yeah, having somebody do this sounds like the best possible use of your capital. Because what you have to ask is, how much more efficient will this person make the organization per year. I'm going to guess they make the organization 10% better per month, but I'll put it at but 2% better a month.

Can an AI optimizing person make the organization 2% better a month is your calculation? 2% compounded for 12 months. That's obviously a bit more than 24%. I don't know what it is. A great one for Claude to do for me off the top of head. 27%.

Perfect. So you're 27%. okay, so if you had five people, you'd be in the black. If you had 10 people, that's only 10% of your revenue to make everybody else 27% more effective. Oh, Lord, you get that 17% spread if they were, in fact, 10% of your spend. And it might be you're spending on other things besides just salaries, but on a salary basis, it would be 10%. It might be 5% of your overall spend or 7%. You know, usually staffing is two-thirds of a startup's expense or any companies in tech.

So this is a big deal. I think this is the future. And I think job displacement and doing more with less is a trend that is accelerating. When Andy Jassy wrote his note last week, did you talk about that on Friday? We did.

We did. I got so many podcasts I'm doing. I did Twitter on Sunday and talked about it all day. All in it came up. And then I wrote this blog, this substack this week on my J-Cow from All-In list. I have a J-Cow

from Allenlist, where I'm talking about the subjects from All In. And then I have the startups one, which is just, I think, calicanus.substack.com, which is where I'll talk about startup stuff. So, you know, maybe politics and think pieces on one and just straight up startup news on the other. But I cross-posted it. And, you know, the thing in the industry is we're just not talking about the job destruction, displacement, automation, however you want to frame it. You can frame it how you like.

What you can't do is deny it's happening. And I felt the need to write this piece because people are specifically telling me to stop talking about this. Whenever somebody tells me to stop talking about it, I do the opposite. I lean in. So when people are like, hey, don't talk about Trump and tariffs. Don't talk about job displacement. Don't talk about crypto and scams.

We're all making money. Then I talk about it more. That's just who I am. So, you know, I talked about crypto and the crime and, you know, all that stuff 10 years ago. And for the last 10 years, I talked about tether, human trafficking, whatever the issues are around these issues. I'm going to talk about this one. If you don't like it, tune into another podcast. They'll take you.

Yeah. You can just go, yeah, listen to another podcast and they'll lie to you. The truth is, this is going to be the fastest job destruction in human history. The fastest job destruction in human history. So what one must do in this situation is just, understand that's happening and think about the opportunity. The opportunity is there will be many, many white-collar individuals and many blue-collar individuals, and I'll say entry rung in, you know, jobs, people who would have taken the entry-level jobs, the bottom of the ladder, jobs,

dishwasher, greater at Walmart, the least qualifications necessary, the lowest-paying jobs in the world picking strawberries, driving an Uber, whatever the least amount of skill you need and the least credentials, those jobs go first. At the same time, management and white collar jobs that are not highly technical or highly human, they're going to go as well. And so here we are, folks. The opportunity is for startups that use that surplus. There will be a surplus of humans available. The white collar ones should start companies. The blue collar ones should also. start companies. If you're a blue-collar worker and you can get 20 people together, you know what's not going to go away? Like home repair, handyman jobs, gardening. Like those jobs are very intricate

and difficult and nuanced and you can learn anything. Long story short, huge opportunity coming for all this surplus. Absolutely. Just one last question on this because I was really trying to think about the future of this. So how do you measure ROI inside of a company that's trying to to automate tasks? Is it just like how much slower your hiring ramp is? People are freed up to do more. Is it your software spend you might be cutting out? How do you make sure that you're doing this intelligently to actually move the company forward versus being performative about it? Previous cost versus current cost. Previous time to complete task versus current time to complete task. And you really have to be monitoring those things. As an example, when we would have an application for funding or we would

get like an update from a founder who we'd already invested in, these were 30 to 60 minute. endeavors. Now, with AI summarizing a deck, AI summarizing an email, AI putting the data from those two things into the database to start, you take out about 80% of the work. Because a lot of it was manually saying, this is a SaaS company, this is a marketplace, this is a fintech company, this is a company that's raising a series A, this is raising a seat. A lot of times they'll just tell you, I'm raising a seed, I'm raising a series A. Where's the crunch base link for this? Where's the LinkedIn for these founders? A lot of this stuff can be done with AI,

and so how much time were you spending on it previously? What are you spending on it now? And then times the number of times you do that task. We have 20,000 applications for funding. If you save 20 minutes per application, that's 40,000 minutes a year. That's a lot. You know, yeah, that's almost 1,000 hours, 800 hours,

whatever it is, like 800 hours. The average employee probably works 1,600 hours. technically they should be working in 2000, but we all know how this works. Some people work more, some people work less, people have vacations, people drink a cup of coffee and hang out the water cooler. Ballpark, that might be like half a position. When you replace half a position, then now I have redeployed in our venture fund that time. And I'm finding we have more time to talk to founders. So we started something where we're doing pods of portfolio companies. So I took 11 people, divided it by the number of active investments we have, which is maybe 200,

50. Each person in the company has a 250 person pod, I'm sorry, 25 person pod, 20, 25 founders. They invite them every month to hang out in that pod and just give an update to their fellow founders and themselves. And we're having, I think right now, five to 10 of them show up. And we just did the first one. If the next one, 10 to 15 show up and we grow from there, just more times of us having an opportunity to understand where the founders out at. And we were previously, you know, be looking online for that information, et cetera. So you can just level up, level up, level up, redeploy that time. But I don't think we'll ever hire another, I don't think we're going to hire people

in relation to our portfolio size anymore. Okay, so this chart from Apollo, share of firms answering yes to the question that they have used AI tools in the past two weeks. So this is a question asked ostensibly to CEOs. did their firm use AI tools in the past two week? Only 9% of them said they are using AI tools. So this to me is an incredibly low number. I would ask the question,

if you go to a startup, you say how many AI tools are you using? Man, if I asked one of our founders, how many AI tools? How many AI tools did you use in the last two weeks, Alex? How many AI tools have you used? Five to 10, probably. on the opposite side of that.

Yeah. I was going to say five at a minimum. Yeah. And so I think I've used six or seven in the past week easily that I've dipped in and out of. So this is incredibly low, but I think these might be legacy firms. I think for startups, and that is the advantage. That's the advantage for sure.

This is definitely the pool of $1.2 million is a large number. All I'm trying to say is that as we think about how far startups are taking this, the rest of the industry is way behind them. So there's a lot of gains coming. And a lot of also, I think, startup revenue to be made. there's a lot of people in law seats to sell. So it's very, very encouraging. Now, on the small team size point, would you like to see Jason a list of the companies that have made the most money with the smallest staff? Okay, most money, smallest staff. So this is back to efficiency.

Yes, I would love to see that. All right. So let me just pull that up for you. This is a website called the Tiny Teams Hall of Fame. And I saw this and I thought to myself, Jason's going to freaking love it. So here we are, Jason, as you can see, this is a tiled list of companies with essentially how big their team is and how much revenue they have generated, mostly measured in ARR, and some of the data here is a little bit dated. For example, we know cursors at 500 million ARR, not 200, but it goes to show just how many companies are building quite a lot with very, very few people. Magnific, a startup that does AI images, 10 million ARR, two people. Mercor, a company we've talked about on the show, 30 people, 50 million ARR.

This is no longer just a proposal. This is now a reality. And I'm blown away by this data. Pretty incredible if you were to look at Lovable. We had them on the program. 17 million ARR, 15 people, a million per person. I think a million per person is a pretty predictable, you know, bogey now. 200 million ARR, 20 people at Cursor, if that is true, that's 10 million per person.

You know, if the average person all in is 150, 200K per year, depending on benefits and how, you know, seniority. Because remember, AI employees are paid much more. So we've seen this, you know, there might be one or two people at Cursor getting paid a million dollars a year who are AI scientists. Then you might have AI developers getting paid 250. you might have a product manager getting paid 150, then you might have a bunch of people getting paid 50, 75K, who are at other data labeling, you know, less developer tech-centric jobs.

But one to 20 million seems to be the average. This would be great in a database to see the average and the trends. And that will be the interesting thing is, where will these companies be over time? It's unnecessary to make 10 million per employee, obviously. So then you have to ask yourself, is that company, if I was on the board of that company, I'd be saying, why do we have so few people? Why don't we have 20 people in Japan making a Japanese language version of this and go down to

five million per employee? And in fact, why don't we also have a sales team in Germany and a German language version and have five people there? And what are the other markets? Why do we not have people in India? And also, why, you know, don't we have a training course? for young people and an educational effort where we have five people working on bringing this technology to, if it's so great, to universities. And so what you just saw happen in my mind was capitalism at its best. Great capitalist, great entrepreneurs. I'm not saying I am one, but they will look at, oh, my God, how profitable can this company be?

And then when the profits come in, then you say, oh, wait a second, I have product, market fit on profitable. How can I increase the velocity of that top line? We know that the bottom line can be profitable. That's what happened with DoorDash, Uber, and some of these other companies. They were losing money. Robin Hood was probably losing money every time they got a new customer. They were investing. And then suddenly it flips. You go out of the J curve and money starts raining down. So this is going to be a great moment in capitalism where you have the luxury. If you're getting profitable so early, you're going to have the luxury to say, what would we do if we could hire three more smart people in Germany? What would they do?

Okay, we'd have a, we would start hosting a weekly in the German language cursor webinar, and we'd have an in-person, you know, cursor con. And we'd take cursor con around the world, and we'd send the same team to different cities and have them train people and buy them bagels and locks and have a great time. So being a money printing business, you know, is great. But if you make, I mean, I can't believe I'm saying this, but if you're making, if your earnings are too great, then you're going to have to give dividends to your shareholders, buyback your shares, incredibly boring. I would, these companies should be asking, how can we grow faster? You just said music to my ears.

Share buybacks incredibly boring. Agreed. Now, I ran into this, Jason, because I was reading a couple of articles, one from Bloomberg, one from the Times. And both of these actually are talking about this trend. of startup founders building quite a lot with very few people. So I think the broader media landscape and the public are catching on to it. Who wrote this story?

The New York Times. And this was from Bloomberg. Yeah. What was the New York Times story? I missed that. When was that? The New York Times was from February 20th, 2025. It was entitled, A.I.

is changing how Silicon Valley build startups. And then Bloomberg's is from June 20th. Silicon Valley's tiny team era is here. Okay. But here's the cool thing. Jeffrey Busgang from Flybridge Capital teaches a class at Harvard. and he was telling his class that they should prompt chat GPT to, quote, act as a co-founder to develop their startup ideas. And this struck me as slightly worrying, but I just wanted to get a vibe check from you.

If someone came to you and said, Jason, I'm a co-founder, sorry, I'm a founder. I'm not a solo founder, though. I also have my AI co-founder. Would you laugh them out of the room or take that seriously? I would take it deadly seriously. because it shows like some visionary lunacy that, you know, when you see something odd, weird, peculiar, peculiar is like maybe a good word. The weird stuff is the most interesting to me.

So I would say, tell me more. And then people hiring agents, remember there was a company charging 15,000, like build an agent to do this. Like, that could be the future. The future might be, hey, we've got this. Claude producer and we wanted to do really cool stuff. We need somebody to write that code and to, you know, write the prompts for us and to, you know, iterate on it. So there's somebody out there who is a specialist that Claude and Zoom and audio and video and podcasting and they just keep building tools. And, you know, that becomes a startup. And then they just sell you the agent.

And then your agent goes from an upgrade like buying droids. So these might, there might be like Jawa's out there. doing this kind of stuff. I like it. Sure, why not have an AI co-founder? You were much more positive about that than I expected, but I'm totally here for it. I also like the idea that people can create what they don't have on their own and just be a little bit less dependent on other people joining on early. So that's a good way to build more companies. Build a board member. AI board member would be incredible. They're out there looking for opportunities. They know what board members ask. So I think there should, that would actually be like a really interesting tool to give to founders

is an AI board member and give them like the five classical personalities of board members, the venture capitalist, finance person, the former CFO, again, another finance person, compliance person and attorney, a former founder who made it big and is now retired, a product-driven person, you know, a strategic,

somebody who worked at another company that made an investment, you know, they're working on Intel or Microsoft and their venture arm investing your company. And that board all bundled together and built out is reading your board deck and you get a preview of the questions they might ask. I think you could do that today. You could just take your board deck and say, you know, what would board members ask about these? And I wonder, I bet your founders are doing that already.

So, uh, Lon just told us producer Lon says that he had a founder university company in his pod that had a board of directors that was entirely chat GPT. So apparently we're trailing founders here by a little bit, but I think this is going to become pretty common because not every board member is so unique as to be unreplicable with a digital system. Yeah, you know, I have this domain named Begin.com that I was able to acquire, and I've been thinking about something in this zone for it, like for startups, for companies, like the Begin, you know, board or, you know, co-CEO. We always talk about Co-CEO as being a bad idea. What if, like,

again was the co-CEO looking at all the data and telling this, you know, CEO coach, right? A CEO co-pilot. Well, that's interesting idea. CEO co-pilot. I like that. I like that a lot, too, especially if you want to think about like cybersecurity and compliance checks and what are all the boring things that you forget, but you need to know as a CEO. Yeah.

Yeah. I dig that. I want to talk really quickly about one more founder thing. This is a story that I think is pretty cool. So, Jason, we've talked about Mark Zuckerberg. who is trying to hire over at meta, a kind of cream of the crop of AI talent to build out a new super intelligence team. There was a recent story from the journal that came out. And it's a bit of a retread of the overall idea that what Mark's working on and how much money he's going to spend.

But what struck me is how hands-on, Zuck is here. So he's actually in a WhatsApp chat with Ruta Singh, the meta-exec in charge of recruiting. Zuckerberg's reading all the AI papers to find out who's doing the work, reaching out to their preferred communication method personally, having them over for dinner at his house, promising unlimited compute, and he is staying in the process of recruiting them, quote, right down to planning their desk locations. And we've talked a lot about founder mode on the show in the last year. And I don't think there's a better example of that than this, because Mark Zuckerberg is one of the world's richest people. He's the CEO and founder of one of the world's most wealthy companies. And he is literally in there handholding the recruiting process to get

the talent that he needs. So I think this shows that no one ever outgrows the founder role. And also, there's no work that's beneath the CEO. It is a moment in time where there are people who have a, there's a finite number of people who have the skill. It might be an infinite game, AI, like building tools and everything like that. But there are a finite number of really good players today. And whoever has them has a greater chance of winning a big prize. If you think the big prize is worth, let's call it, $10 trillion, you know, there's probably no better use than recruiting right now. If you recruit somebody who comes up with the idea of how to leverage, you know, Facebook and WhatsApp and Instagram's data to serve a, you know, better ad or a new ad format.

It was really questionable what the ads would do on Facebook. I didn't believe that ads would work in social because you would be interrupting people's conversations, et cetera. I thought it would be really hard to do, whereas Google ads were very specific, right? You type in a keyword, you know what it is. And I thought these ads would be not as effective in the feed. It turns out I was right. They were not as effective in the feed on a cost per click basis, on a click-through basis, right? If you type in a very specific keyword on Google, your chances of clicking the ads are very high. In a social network, you know, if you scroll by something, it's very low, until they made the ads, and it still is very low, but people are in there for three hours.

Whereas on Google search, you're on there for 30 seconds and you do 10 searches a day, you're on it three minutes versus 30. So, you know, there's a big difference between 200 minutes or 300 minutes in a service and three. They figured out a way to make it work. And they also had a lot of psychographic data on the person who they're friends with, et cetera. So I think Zuck is doing the right thing, trying to get people to come work for him. And if you were a freelancer, if you, if you were talent at this point in time, it would be foolish to not test your market value. If you, like, had your deal a year or two, three, four years ago, you probably want to test your market value because the value has gone 10x in a year or two.

These people are super, super valuable. How cool would it be to go to Mark Zuckerberg's house in Palo Alto and have dinner? Like, I don't know. I would answer his email just for that. One funny anecdote. Someone actually just didn't think that it was actually Mark reaching out to them. And so they didn't respond to him for days and days and days because they thought it was spam. And I thought that was a kind of an interesting problem to have if you're Mark.

Like, no, it's really me. I'm here. All right. I want to do one more story before we jump to our interview today, Jason. Just a quick note on an acquisition. Are you familiar with the company called Couch Base? No. Couch Base.

No. That's the most honest. I love that. Okay. So really briefly. base when public back in 2021, and they offer what's called a source available, it's not open source, but similar to it, no SQL database. They have not done particularly well since they went public, and that's why they're getting

sold. They're selling to Haveli investments for about 2450 a share. That's 50 cents more per share than they went public at, so this was effectively a flat IPO from 2021 through today. But just some data points for founders out there who are curious, what is the clearing price today for slower growing SaaS companies. Well, the company had ARR of about 250 million at the end of Q1, Jason. And so that means it's selling for about 6.5X Q1 and a recurring revenue, but that's with the premium in the take private. So I think probably like 5x, 4 to 5x for slow growing

SaaS, I think is the current market clearing place today, even with markets at or near all-time highs. Yeah, there's a lot of private equity folks looking at underperform. assets that went public. This has always been the case. You know, this is something that happens in public markets. And they go private. They change the management, typically. They rip out a bunch of expenses, and then they look for somebody to flip it to.

So, you know, it's even if your company doesn't do well or is slow growth, these guys are growing 20% year over year, 10, 20%. That's still pretty good growth. growth. But I guess the market, for whatever reason, doesn't believe in this company. And so taking a private is a great way to go. And you're going to see that over and over again. If a company doesn't break out, you have this backstop of private equity folks looking at it, saying, yeah, I'll take that. And there have been a number of interesting companies. Dell, I think, probably would be the biggest example. I think that was the largest go private ever. I think was Dow. At some point, it went private and then went public again. And it's obviously doing phenomenal now.

$24.4 billion back in February of 2013. That makes me feel old. I thought that was much more recent. I didn't realize it was 12 years ago. And now it's worth 80 billion, right? So, you know, it's, it'd be interesting because, you know, when now if you go look at the max chart for Dell, you know, you get a stock price to 2016, which I guess is when it came back out. It'd be really interesting to look at It's market cap when it was public, then take out the private. You could kind of try to blend how much the company's been worth over all that time. But, you know, here we are. Going private lets you clean up your cap table, your product line, all of that, you know, in a very clean way.

So, yeah. All right. Now, today, Jason, we have a bit of a return, if you will. We're going to talk to Scott Hickle from Throne Science. Now, you may recall that lawn, myself, and. And you back in May looked at Throne Science, which is the company that puts a camera in your toilet. And we made one to seven to 500 different jokes about it. Yes.

And the founder was very, very kind. He said on Twitter, thanks for talking about my company. I don't think you guys fully got it. So Lon said, let's have him on the show. Right. And ask him what we missed. And generally speaking, take another crack in this. Are you saying that we isched on the company and now we're making up for it?

Yes. I'm saying that we're going to give ourselves a swirly by having Scott Hickle now. Are you saying we're going to get the straight dope from the founder? I'm hoping so. Scott, can we bring you up? And we're going to hear all about this. Scott, there you are. Can you just give us the huge dump on like what's going on with this business? I mean, you have a toilet behind you. Yeah, this is the office. I mean, the number of jokes and the fundraising, when you went and raised funding, what was the best joke that landed? Who dropped the best? By far my favorite. I could not tell you who did this. I've heard 10 million of them at this point.

By my favorite is poop jokes aren't my favorite, but they're a solid number two. Poop jokes aren't my favorite, but they're a solid number two. That's strong because you get, you got layers of joke there. It's layers of joke. But in all seriousness, how did you birth this one? So in truth, it started as a joke. So my now co-founder, Tim and I met. He moved to Austin back in 2021 as part of the COVID exodus from San Francisco. My best friend from college knew him and was like, hey, please give Tim a soft landing when he arrives to Austin. I brought him into a poker game. And we were sitting around the poker table talking about startup ideas.

You'd love to start, but would not want your name associated with. And so everyone's pitching sex, drugs, rock and roll, vice industry stuff. And Tim was like, you're all fools. Clearly, the money isn't smart toilets. This is inevitable. At the time, his idea was a toilet seat with a scale on it. They would just weigh you before and after as kind of a vanity metric. And I was like, you know, that's hilarious. Clearly, you would put a leaderboard on it, sell it into frat houses and name that company thrown. So that was the original idea for what is now, I think, a deadly serious invention. But, you know, Tim and I joked about it for two years. We ended up working together. And it wasn't until 2023. when we were looking for a startup idea in earnest to go start something, I called my mother, who's a geriatrician. I was like, hey, mom, is there any medical utility to looking at people's waste? And her response was, honey, in the field of geriatrics, there's an old joke that is

old people care about three things and three things only. Their kids, their meds, and their poop. As you age, your gut motility starts to slow down. You are very aware of changes to your digestive system and you talk about it all the time. In fact, she says she gave, she stopped giving her phone number to her patients years ago because they would send her so many unsolicited pictures of poop. So how does the product work? Like if and how did you find product market fit for this specific product? And I'm asking this in all serious now. We'll make more jokes after this. We had jokes before it. But what is the reason, the killer reason, to invest in this product and to buy it? What is the great outcome here? Yeah. Great question, two answers. The first is the health and

wellness answer. So there are roughly 60 to 70 million Americans with chronic digestive diseases who do not have anything to track those conditions on a daily basis, the same way that we have 50 devices that track sleep and exercise and respiratory rate and cardio, right? Like I'm wearing a whoop. I've had an aura. There's nothing looking at gut health, urinary function, or hydration in those same daily ways. And then ultimately, kind of our North Star, the mission for us is we want to build a smoke detector for colon cancer. So colon cancer is the second deadliest cancer in the country behind lung cancer. And the kind of pernicious thing about colon cancer is if you catch it early, it is one of the easiest cancers to cure it, right? It's a 45-minute outpatient procedure. You get a

colonoscopy and you can make it to your kid's little league game that night. If you catch it late, by the time it's gone to distal metastasation, you're looking at like a 15% chance of survival five years from now. And the crazier thing about it is that you have basically a seven to 10 year gestation window for a polyp to develop into a malignant tumor. So if we can detect it in that gestation period by looking for trace amounts of blood that are invisible to the naked eye, that that will not only improve health in the, you know, helping people monitor gut health, but ultimately save lives. Scott, though, you were talking about. You were talking on Twitter about your fundraising and how you guys were trying to pitch it as essentially a smoke

detector for colon cancer and also, I believe the phrase was whoop for your poop. And then you had to flip that. Can you just walk me through how VC's response to this and then what you had to change to make the pitch work? Yeah. So exactly right. So I think some of the most well-meaning, but ultimately unhelpful advice I got going into the fundraise was, you know, we have this long-term vision of building the smoke detector for colon cancer, but short term we're building whoop for your poop, bright like the daily gut health and hydration tracker. And so people would say start with the big vision first. And when you start with the big vision first and say, here is this $100 billion opportunity that can save tens of thousands of lives around the planet every single year,

you lose people's attention when you start going into it. And here's how we're getting there by building this health and wellness device first. Whereas when you say, we're building this health and wellness device, that's a multi-billion dollar opportunity because there's whoop and ore that have proven the market. And, you know, Garmin and Fitbit have educated. Now one in six Americans is already using a wearable every single day, when you start there and then say, and then by the way, kind of the call option here is if we can build the device that also detects fecal occult blood and that can save lives, just ending on that note is far more powerful than starting with, here's the high note, and then by the way, it kind of tapers off.

All right. On that note, continue to success. And keep us informed. Let us know when the new version drops. Alex, Jason, thanks for having me. Thanks, Scott. Appreciate it, man.