This Week in Startups - Is there an AI coding bubble? Plus Meta’s new SlopTok product, Neon’s divisive app shuts down, and more | E2184

Episode Date: September 27, 2025

Today’s show:Factory, a Sequoia-backed producer of AI agents for programming, raised $50 million at a $300 million valuation. So Jason and Alex are wondering… how much capital is TOO MUCH?We’re going deep on a question that often perplexes founders: How would weigh a HUGE potential market against STIFF pre-existing competition.PLUS that Neon app that records your calls deleted itself, why AI ISN’T taking away all the radiology jobs yet, how inherent biases impact everyone’s view of the economy, ChatGPT may be getting ads very soon, and MUCH MUCH MORE.Timestamps:(0:00) The Neon app: why would anyone let an app record all of their phone calls?(04:43) Another -slop term… What is THINKSLOP?!(09:46) .TECH: Say it without saying it. Head to get.tech/twist or your favorite registrar to get a clean, sharp .tech domain today.(11:25) Why is TikTok US only worth $14 billion?!(17:10) Wait, why ISN’T AI taking jobs away from radiologists?(20:06) Public - Take your investing to the next level with Public. Build a multi-asset portfolio and earn 4.1% APY on your cash—with no fees or minimums. Start now at public.com/twist(28:38) How can xAI afford to sell AI to the US gov’t for 42 cents?(29:37) Alphasense - Get deeper insights into your business with the power of AI search and market intelligence. Start with a free trial at https://www.alpha-sense.com/twist(33:44) How inherent biases impact everyone’s view of the economy.(46:29) Why is everyone dunking on Meta’s new SlopTok app?(53:37) It certainly seems like ChatGPT is going to get ads soon…(03:49) The first entry in our Gamma Pitch Deck Competition is II off-roading marketplace Where2WheelSubscribe 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:.TECH: Say it without saying it. Head to get.tech/twist or your favorite registrar to get a clean, sharp .tech domain today.Public - Take your investing to the next level with Public. Build a multi-asset portfolio and earn 4.1% APY on your cash—with no fees or minimums. Start now at public.com/twistAlphasense - Get deeper insights into your business with the power of AI search and market intelligence. Start with a free trial at https://www.alpha-sense.com/twistGreat 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

When founders see a big tank and competition to this rabid, should that push them away from building that space or should have drawn towards building that space? You should be drawn to going into that space. If a lot of people find a great beach to surf, you probably want to surf that beach, right? Because it could get better and better and people could be underestimating the tam. So if I told you when sidecar and lift were battling it out for ride chairing, should Uber join the race? Yeah, they should. If I told you before XI had released an LLM and Chat GPT and Claude had, and Gemini had run away with it, should XAI jump into the race, you might say, no, it's too late. It's never too late. Somebody could start an LLM right now. I know that sounds crazy. But if all of a sudden Apple decided this is strategic for us, let's put $100 billion into it and let's double the offers everybody else have and let's find a way to put this on every iPhone and we'll design the entire iPhone around the idea of it being an H-100. Like, and the idea of building an H-100 competitor and competing with Nvidia.

Okay, it sounds crazy. But if there's a big market, there's always going to be room for innovation. This weekend startups is brought to you by Alpha Sense. Get deeper insights into your business with the power of AI search and market intelligence. Start with a free trial at Alpha-sense.com slash twist. Public. You take investing seriously. Public does too.

Build a multi-asset portfolio and earn an industry-leading 4.1% APY on your cash with no fees or minimums. Learn more at public.com slash twist. And dot-tech. Say it without saying it. Head to get. Dottec slash twist or your favorite registrar to get a clean, sharp dot-tech domain today. All right, everybody, welcome back to this week in startups. It's Friday. And there's a ton of news.

Alex, let's get to work here. What's our first story? First story today, Jason, is all about an application that reached the top, the pinnacle of the app store before being taken offline by the founder because there was an enormous security problem. The app in question is called neon, and its model was interesting. It paid people to upload their phone calls that it would then sell the phone data to AI companies for training. In theory, you could get up to 30 cents a minute if both people in the phone call were neon users, or up to $30 a day max if it was just a one side of a phone call. The problem is, TechWrench took a look at the application,

did some, I would say, reasonable testing on it because they have an excellent cybersecurity team over there with Zach Whitaker. And they found that, quoting here loosely, the back-end servers were capable of spitting out reams of other people's call recordings. Neon servers could produce data about the most recent calls made by the apps of users, as well as providing public links to the raw audio files,

and the servers in question could be manipulated to reveal the most recent call records. So, Tackerman told him about this and they took it down. Didn't actually say there was a huge flaw when they told their users they were improving security, but in the wake of T, the app we talked about a couple weeks back that also leaked quite a lot of user information. I thought it was a great moment to sit back and say, cybersecurity for startups. We should think about it. And I was curious what your take is and what you tell founders. I mean, if you look at Palo Alto networks and all these cybersecurity firms and how amazing they're doing, the Keshe over there is just a beast. And he's been buying

up all these, you know, disparate cybersecurity firms, and it feels like there are vectors all over the place. Now, as a consumer, you should use this really elite technique. It's called Common Sense. Oh, my. Yeah. It's a new technique. It's, you know, it's trending in some subredits. But common sense would tell you that recording your phone calls is an absolutely stupid thing to do. So, but this is the world we live in where everybody's transcribing everything. At a recent management team meeting, I just told everybody, bring a mullskin in a pen. You know, we'll use the notion has the built-in transcribing software. So one person will do that who's hosting the meeting.

But everybody else, laptops close. Let's try to have a concept where everybody writes down what they're working on and what they think is important. and then we can compare it to the AI notes. What I'm thinking is happening is we talked about work slop the other day. Yes, we did. And that is when people create work product using AI and their coworkers think they're dumb and not trustworthy. There's another one that I think is super important,

which is how you think. So I'm going to coin a term here, think slop. I think a lot of people are getting sloppy, in their thinking because they're not doing the basic steps required for first principle thinking or just, I don't know, logic. You have to process information in your brain to come to conclusions and have strategies, then you have downtime, and then you go back to the product. This is why bookmarking and taking notes and then having on your schedule a time to review

your bookmarks or notes. And so I started doing this. I've been bookmarking stuff. And I'll say, oh, there's a potential blog post. This is an interesting business idea. This is an interesting investing idea. And then, you know, a couple times a week when I don't have downtime, I will go through my bookmarks and just say, why don't I bookmark that? And then sometimes I'll unbookmark something or I'll actually take action on it.

It's the same thing with notes. When you write stuff down, you remember it. When you then transcribe it into a Slack message, you have another chance of thinking it through. So it sounds inefficient, and certainly it is, to take notes in a meeting when you have a note taker, and then it seems incredibly inefficient to read your own notes and then write them into, you know, your notion page or your Slack. But I made everybody do it, and I just felt people were more present. And so I'm putting that out there as a good strategy for people who are into getting stuff done.

Yeah. So what Jason's advocating for here, as far as I can tell you, is one of the best study techniques that you should have learned in high school and college, which is take your notes. And then if you rewrite your notes, you get an amazing refresher of what you've learned to prep for a test or just to have that information in your head. But Jason, you also said something about taking a break and a pause and getting a little distance from what you're doing.

I found that if I'm having an exceptionally busy workday, the best thing I can do to get more done is to get away from my desk for 20 minutes and take a walk. Because in that time, I actually had the ability to think, digest, and then come back with some ideas, some thoughts, some ways to move the ball. forward. So a lot to be said about that. Don't do work slot folks. But this story in particular, Jason, this neon startup, two things here I'm kind of curious about. One, do you think the economics were ever going to work out for paying people up to 30 bucks a day for phone call recordings? And two, what step should founders take to ensure that when they build something before it goes

viral, that they had the right cybersecurity bricks in place? You know, there are firms that are processing large amounts of data with experts, asking experts to answer questions, to do reinforcement learning, all that kind of stuff. And I think this is an interesting concept. If you were, I'm sure the people who are doing the note-taking for doctors or therapists, we have somebody in our accelerator who's doing it for therapists. Yeah, just having an expert who is, I don't know, doing taxes, talking to their customers and recording the calls and then training based on those calls.

That's a very interesting idea. if the person is giving great advice and somebody else looks at that training data, because garbage in garbage out. Yep. And you see this in customer support. All customer support calls are logged, right? They tell you that on the recording when you call. So that's why everybody's going after that first.

It's an expense and it's all recorded. So you have the data set already. You can go backwards and look. Here's the data set for this product. Oh, we have a second product line. Yeah, here's the data set for this set of speakers. here's the data set for this set of headphones. They probably have nothing in common in terms of customer support calls.

And you can just make little language models out of those and make that the inference data, et cetera, or in the context window, rather, and, man, super powerful. So I think it's not a terrible idea. Okay. But I think as a large-scale project, it's kind of weird because it assumes that people are intelligent. Which, going back to your common sense point, maybe a bit of a stretch of an idea? Could be a stretch. Like, what exactly are we trying to accomplish here?

Do people say things on the phone that are more valuable than the people who choose to contribute to a Quora or a social network or a Reddit thread or a hacker news thread or email? Like, I would think when people write stuff down, perhaps they're more thoughtful. But who knows? I mean, it is data and signal. but, you know, what if you are on the phone and people are just spewing off things they read that they never verified? So now you've got things people overheard or, you know, we see this in breaking news, right? People think they understand a situation in an emergency news situation,

and they just, they've heard a couple of theories and then they repeat them. Is that what you're putting in your training data? It's kind of weird. Garbage in, garbage out. We all understand the importance of a crisp, memorable easy to spell domain name. One of those names you can say over the phone and people know how to type it in without asking you the spelling. But let's get real.

The good ones are either taken or there's some poacher who's holding it and waiting for some huge payday and they don't reply to you. Even if you want to pay for a premium domain, you don't want to use up all your runway on a domain name. That's just the truth for a startup.

You want to put that valuable cash back into your startup's operations. So you should consider this, a dot tech domain. You can get a clean, crisp, super memorable name for your website and company and signal out loud to your customers and investors. We're a tech company that's instant branding for you. That's why over 500,000 founders have collectively raised over $5 billion in investment, building their companies on DotTech. So skip the hassle.

Head to www.gat.git.com slash twist. Or go to your favorite registrar and grab your dot tech domain today. Yeah, so I think we're going to see a lot more of this. People were really talking about how like, oh, this must have been vibe coded, et cetera, et cetera, et cetera. I don't think we have any evidence in the case of neon that it was in fact built with a cursor or a windsor for a similar sort of service. But as people do use those services more, as they become more popular. And as it becomes easier to go from prompt to app to publish service, I think we're going to see a lot more problems like this. So if you're listening and someone's offering you money from an app, probably don't trust it.

It's probably not exactly as built as well as you'd hope it to be to protect. your data. All right, next up, the big TikTok deal. A lot of questions about this. We've touched on it before, Jason, about the consortium, the delays in the law that was going to ban or force the divestment. Now we're here. We have outlines of what the Trump administration is proposing. And most importantly, people are confused about why TikTok, the U.S. operations of TikTok, are being valued at a mere $14 billion. I went back through all the reporting. The most reasoned and most accurate number that I could find for TikTok scale was that it did $16 billion with a revenue in 2003, so it's a rather dated figure, implying that it's much larger now. And just for fun,

if you presume 25% growth in 24 and 25, it would have about $25 billion in revenue this year. My question is, why is it value of $14 billion? I don't want to be a conspiracy there is, but I also don't understand that figure. So I wanted to get your take on the number. Okay. That number doesn't make any sense. People are right to question it. Where, what is the source of this number? Is this officially from the White House? Is it from bite dance? Is it from, you know, Oracle who's paying for it? Do we know the source and the providence of this number? J.D. Vance, the vice president. In the Oval Office about 20 hours ago, yeah. Okay. So the administration said the valuation is $14 billion. About $14, yeah. So something's wrong here. And it could be that

The $14 billion is for the 80% folks are buying and the 20% that China is keeping, or $14 billion is the amount being invested. That's what I think is going on here. Somebody misspoke. I think the existing investors like Jeff Yoss are keeping their shares. Brad Gershner, keeping his shares, I would assume, in all of this. Great trade for him, by the way, Bestie Brad. So let's say the new group is investing $14 billion. And there's a communication area here where the $14 billion isn't the valuation.

It's the $14 billion going into the company. $14 billion going in for a third of the company, if we assume the other two thirds is the Chinese shareholders are keeping 20 percent and the existing shareholder base. Let's say only a third of this is new. That would make it a $45 billion or a $42 billion valuation. $42 billion valuation would be three times revenue, two times revenue, something in that range, maybe. And so it would make more logical sense. That's what I'm going to say is going on here. I just pulled the numbers. It's expected to be a 45% stake from the new investing group that includes

Oracle and Michael Dillon and Murdoch. So that would value TikTok at about $31 billion. Okay. Makes more sense. Makes more sense, but I pulled the bunch of it's still low. It's still low. And the reason why I think that is there's two kind of comps for this company. There's the social media world, and then there's the e-commerce world. So if you look at price, sales, multiples from public companies that we have data for, meta, about 11, snap, about two and a half, Pinterest, about six. And then on the e-commerce side, Shopify is valued at 18, 19 times revenue, and Amazon's worth about three and a half X. So no matter where you look, size of company, sector, whatever comp you pick, it still feels incredibly cheap. The steel man of this is from Keith Rabeau.

investors that we've talked about on the show quite a lot. And he responded to, Shiel Monhot from the Fintech Venture Capital firm that I always forget the name of, BTC Ventures. And he said the company would have been worth zero under federal law, so this is a heroic outcome. That is true, but this does not feel like the correct valuation for the company. If you told me I could buy shares of TikTok right now for $14 billion,

I would sell every share I own of every company. and I would plow or I sell 80% of every holding I have and put it in there because one-time's revenue makes no sense or less than one-time's revenue if that's in fact true. And so I don't believe that's true. So we'll have to wait. A lot of times he's breaking news stories,

especially out of this administration, which tends to talk into microphones a lot. I've noticed. And they like having five different announcements a day about different topics, whereas the last administration, like Biden had two press conferences in four years. Like, it's such a stark difference here in terms of communication strategy. I think they're kind of like your friend at a bar who's like, yeah, totally.

We're taking TikTok private. And it's like, okay, you took TikTok private? Like, yeah, we took a private. I'm like, okay, yeah, yeah. Can I get it on the deal? Like, yeah, yeah, right as we sign the papers. I'm like, wait, wait, wait, you just said you closed the deal. It's like, yeah, deals closed.

It's like, wait, but you haven't signed the paper so the deal's not closed. He's like, yeah, yeah, no, no, we have a letter of understanding. And then every time you double click on something with this administration, it's like, okay, what's the exact reality of this? Oh, we have, we closed a trade deal with the UK in principle and it's being papered or it's a memorandum of understanding. It's always a little bit like that, right? It happened with just this week, to give another example with H-1Bs.

It's 100,000 per year, or it's 100,000. thousand for one-time fee or 100,000 for 10 years. So it's 10K a year or five years, 20-K year. All these details seem to, you know, get a little mixed up. It's, uh, and it is what it is. I like the fact that we get like more real-time information, but it does make us in situations like this come to the conclusion that we're going to have to wait and see. So I always tell people, whenever Trump tweets something, there was 72-hour rule always is in effect. All right, next up is an interesting bit of news about radiology. Of course, this is not a medical show, Jason, but one of the places where people really thought

that AI was going to have the most direct impact on employment and also on our lives was going to be in the radiology department, reading the scans of your body that we use humans for today to find cancers, to find tumors, to find anomalies. I'm saying medical words, it's not my world, but it was supposed to be a big darn deal. But as it turns out, according to a study from works in progress, not really the case. Demand for radiologists is at an all-time high. They're actually expanding the resident classes from medical schools to include more radiology positions

because we need more of them. So the question is, why didn't this really go the way people thought? Condensing dramatically here to fit it into our show, but essentially, medical imaging models often don't perform as well in the wild as they do against benchmarks. Not a huge shock, but it gets more interesting than that. If an AI model was trained on imaging data from one particular hospital, it doesn't actually translate as well to other hospitals.

hospitals that are doing similar tasks. So there's some data biases inside of how we're training these that are making them not quite as good on the ground. Another thing is that they tend to be trained for single issues versus a radiologist being able to look for all sorts of problems on a scan at the same time, et cetera, et cetera, et cetera. And this all echo is a 2024 study from, I think it was Harvard Medical School that found as well that, quote, use of AI can interfere with a radiologist's performance and interfere with the accuracy of their interpretation. So I bring this to the show, not to just say boohoo AI, but because there are a number of startups that I keep tabs on that are working on medical imaging AI models and bringing AI to the medical world. And so I just wanted to talk about this as a, have we oversold what AI can do, Jason, or are we just sitting here in kind of the trough of despair until we get these models to be a little bit smarter and a little bit more multimodal?

So maybe we can take on this type of challenge and I'll live a little bit longer. Okay. It's hard to know what's going on here because they're saying that these models are not as good in the field as they are in testing. Am I correct? Yes. That is the core of this. So why wouldn't it work in the field like it worked in the laboratory is my question. What's the theory there? Is there a theory there? A couple of things. One of which is the way people present data is a little bit different. So you won't always get the exact same data brought in. into a prompt scenario. So if you take a scan, it's not always in the same format or the same size or the same location. Okay, so standards. The data going in could be variable, whereas the data's cleaner going in during testing.

Okay, that feels like an iteration problem. And I wonder if in the field, in practice, there are radiologists who are, I don't want to say they're trying to sync the technology, but they're creating roadblocks for the technology. All right, folks, the secret is out. Smart, motivated people are using AI tools constantly to improve their productivity, to solve their problems, and just generally make their lives better. But isn't it time that all this incredible technology started making you money? That's why you need to go to public.com. They're not using AI as some bells and whistles type feature. No, it's baked into every facet of their app. We're talking automated portfolio

insights, earning call recaps, and clickable summaries on every stock's performance chart. You can even ask the app a question about any stock in plain language and get a straightforward direct answer. It's everything you need to build a multi-asset portfolio that includes things like stocks, bonds, options, crypto, ETFs, whatever you're into. Plus, public offers industry leading yields, earning you 4.1% APY on your cash with no fees. For a limited time only, public's offering a 1% match on all IRA deposits, transfers, and 401k rollovers. It's time to access the investing platform for people who take their money seriously. Go to public.com slash twist to find your account in five minutes or less. That's public.com slash twist. So that's really interesting because I was thinking

about the same thing. Wouldn't a radiologists want to say, hey, this doesn't work. Look at how good we are. They actually looked at what they call a pair system, I believe, when you have one human and one AI. Don't you think that would be a great combination? Have the human, have the AI work together. It turns out that that actually yielded worse outcomes, more callbacks of patients and no improving in cancer detection because the doctors were immediately like, well, the AI will take a look at it. And so they weren't trying as hard. So when we tried to have kind of the human in the loop or expert in the loop back to your earlier

point, it actually made for worse outcomes. So they're not trying to stand back it. They're just trying to work less. Yeah, interesting. I think we'll probably see more and more of this as AI hits the field. You know, if you go on self-driving Redits, subreddits, and you look at FSD or you look at Zooks or Waymo, you know, you will see tons and tons of real world situations where these cars get confused, et cetera. You'll have a hard time finding, you know, a crash because they're typically not on a, highways. They're doing things at low speed. They're doing things on controlled courses, essentially.

And the companies are almost universally being extremely cautious. In fact, I would say universally, they're being extremely cautious, either with a safety driver or low speeds, et cetera. So you can have these troughs of despair where, you know, you're correct. The technology was hyped, And you had this hype cycle. And then the reality is we got to figure out how to deal with cones being put on the hood of cars. We've got to figure out how to deal with, you know, somebody double parking on both sides of the street like they do in Brooklyn, you know, Fifth Avenue, third Avenue where I grew up. People are double parked. They're triple parked.

These mafioso guys would park their cars, leave them running. And you'd have like, you know, three Cadillacs outside of my dad's bar, you know, double parked. the police wouldn't give them tickets because they knew the guys were connected. If the guys did get tickets, they could hand him for the cops with a $20 bill to get ripped up. You know, it's a whole thing. And so, yeah, here we are folks. Because it sounds like you grew up in a movie. A bit of a movie.

Like, you're describing like the opening scene of like some sort of like 1960s gangster film. Like, oh, no, don't worry about it, really. 70s. Fair enough. So I think the other thing that will happen, and I had some feedback from Amazon where I was, you know, somebody from corporate comms at Amazon, you know, kind of, I was like, hey, just want to make sure we check reality and, you know, here's some information for you

so that you're more informed, which I appreciate, by the way. And they were like, even though we've put tons of robots in Amazon factories already, we've still grown our employee base. And you're like, and the reason is, because we keep adding new products or services that the robots haven't gotten to, like same day delivery, like this, like that, you know, just tons of different things they're doing, you know, packing groceries. Groceries is a big business for them now and,

and, you know, putting pears into brown bags and making sure they don't get damaged. Not exactly something you want a robot doing yet. Eventually, the robot will do it better than a human, but for now. So, yeah, it's going to take a little time. And then additionally, consumption could go up. And people have been talking about that paradox. Jevon's paradox. Jevon's paradox. If you make something cheaper, it induces more use. Induced traffic is another example of that paradox at work, which is, oh, you're like, we're going to add two more lanes to this highway, and then all of a sudden, traffic gets worse. And you're like, why did traffic get worse? It's like, oh, because people saw

us improve the highway. And so they lived further away because we went from a two lane highway, you know, out here in Hill Country. They're doing this to the 290. It was like three lanes each way and tons of stoplights. Now they're building trenches and overpasses and no stoplights. and putting in another six lanes. And so that means more people will move out to dripping springs or Wimberley because you'll be able to get there quicker, which then fills it up. So, yeah, I think radiologists aren't going to magically go away. What will probably happen is people will go for x-rays more often.

And the idea of, like, sending somebody for an x-ray, I'm sure because of the cost, insurance companies might turn it down or might say it's not necessary. you know, when you get a calm and cold, they don't send you for an x-ray, but if you get like severe upper respiratory, they do send you for one, right? But I think that costs $200. So are you going to send everybody who's got a cough to get an x-ray? Also, x-rays, I think there could be some side effects, so you don't want to go for them

too often. Yeah, that's why you have to wear the lead vest at the dentist when they're doing the scans of your face. I also want to make the bullish case here. I didn't just bring this up because I wanted to say boohoo AI. So I read the article, and I read some other studies, just to kind of, kind of, you know, to get my arms around it. Quite often, these seem to rest on historical data from 2018, from 2016. So when you read these analyses, I think it's good to keep in mind that we're

always looking backwards a little bit in time, and AI has gotten smarter and better and more capable. So I think that that's important. Also, there are several startups that are working in this space that are doing quite well. So one's called RAD AI. They built what they call powerful radiology AI. They raised 151 million total, Jason, and they ranked number 107 on the Inc. 5,000 this year with 3,229% three-year growth. So companies that are start-ups are doing well, another one, AI doc, building a holistic AI system for medical imaging, and raised over $400 million. So even though some studies are saying this, the startups are on the case, and I am optimistic that they are going to get it right. And then we can all have free and faster

and smarter medical imaging, and we don't have to pay radiologists an average of $500,000 per year. Yeah. And it'll be cheaper. That's the thing I'm looking to. forward to like just make everything so much cheaper and getting your blood work done, you know, you can use superpower function. I think Woop is adding the ability to just go with self-directed healthcare and get your blood work done. People are like Nutrisense, we're investors in. You can get one of those glucose monitors directly and start checking that out. So all of this is going to result in consumers buying healthcare like they buy travel services. You used to have to go through a travel agent to figure out, is there, how do I get to Italy? How do I get to Rome?

What are the flight options? And they would type on their computer and then explain it to you and give you some advice. And now it's like, uh, you know, just open up an app and go to Google flights or go to Expedia or whatever it happens to be. And, you know, you disintermediated that entire layer of flight or travel consultants, right? Travel agents was, they were stormy. of travel agents when I was growing up. I don't know when they went away, but sometime in the 90s, travel agencies went away. But if you were going to plan a vacation, you would walk down the street, walk into a storefront for people who are listening, and you'd sit there and do an hour or two consultation with somebody of, here are different, and they give you brochures, here's different

places you can go on vacation. Here's how you get there. Here's the hotels. Now all that information is online. So, moving on. XAI has priced its government AI product at Jason. I'm I'm very proud of report. Not 69 cents. They didn't do that. Instead, they went with 42 cents. And so we all need to give Google points because Google priced its AI product for government at 47 cents.

So X-A-I had to go lower than that. What's the only meme number under 47? I guess it's 42 if you're a Douglas Adams fan. So we now have four different offerings available to government agencies from now through 2027, Open AI for $1, Anthropic for $1,000, for $4.47, or X-Ey-Ey-I-I-I-Ey-Sandthroper-Sense, or X-AI for 42 cents. So I guess we're going to bring a lot of great AI to the government. My question is, how the hell can these companies afford that? Because they're all losing so much money. And now they're going to support the government for effectively zero. As an investor, Jason,

doesn't that make you a little, eh, scared? If you don't have the right data, your company can't make the right decisions. It's really that simple, folks. But your information, it's spread out all over the place. You got shared drives. You got legacy systems. Maybe you're taking meeting notes. using AI to do that, you need one source of truth where you can find the most reliable, accurate, and real-time information possible. The answer is Alpha Sense. They're the AI search and marketing intelligence platform trained on over 500 million business documents. Their gen search platform gives you a blazingly fast start on your research projects pulling together an analyst level deep dive on any topic industry or a company in just minutes. Plus, you can use

gen grid to keep tabs on multiple different companies or topics right now or over time. They've even got an autonomous AI interviewer, keeping you plugged in to the latest thoughts from industry experts without the need to spend an entire afternoon jumping on Zoom calls. It's time to remove uncertainty from your process and start making better, smarter decisions. We're going to start you off with a free trial. Go to alpha-dash sense.com slash twist. That's alpha-dash sense.com slash twist for a free trial. I think they're trying to jumpstart the government's interest in these technologies.

And if they do get value from it long term, they'll want to buy the newer, fresher models that come out in the future. And they'll eventually want to pay for it. They'll eventually want it to have on-prem versions of it. So it's, this is the same. Number one, it's awesome that the technology is doing this as a group, all the technologists, all the technology companies in supporting the government to lower costs. And I think they will wind up benefiting from it if the government becomes dependent on this technology and implements it.

Just like, you know, you can start out with Notion. I think Notion has a free plan. Slack has a free plan. A lot of these products start with free plans. Ah, this is premium for government. Okay. I think this is, I would say this is, I'm not saying that to be cynical. No, no, no, no, no, that makes sense to me. I think that's what this is because there'll be other ways to monetize this.

Like I said, you know, if you're the CIA, if you're, you know, the Department of Defense, if you, I mean, if you're the IRS and you want to start implementing this technology, at some point you might say, you know what, can I get some on-prem? And one of these companies, XAI, I might say, sure, we're going to set up a bunch of H-100s at your location or we're going to cord an off in our data center, a bunch of these inference machines. It'll be your own world. Nobody can get into it. we can't get into it.

There's all kinds of ways for this to get monetized later. But the government's and the government health care and education are usually the three last cities to fall when a new technology goes on a, you know, on a global empire building quest. Those are the last three cities. They're the most resistant. So how do you make them not resistant? 42 cents per agency.

Or 47 cents. Or the shocking cost of 100 cents. from Open AI and Anthropic. Can't believe those money grubbers. I know. They're charging a whole dollar. How dare they? Well, so a couple things that I think about this.

One, I think Google's in a really good place here because they can afford to subsidize any amount of inference for the government. The other companies have to raise either equity or debt capital to pay for this. So we'll see how that works out for their economics. But I do see the premium option. I had a more cynical take that I wanted to bring up. Do you think that we're seeing private sector companies offered the government low-cost, high-quality services,

essentially to curry favor with the White House. Is that something that we should now keep in mind in these moments? Or am I being two tinfoil hat? Yeah. I mean, you're a lefty. I'm a moderate. Other people are, you know, far on the right or MAGA. And, you know, you are correct that people will interpret behaviors based upon the lens they're

coming from. In fact, that's how people will determine how they feel the economy is going. There's this economic paradox where during the Biden presidency, Republicans felt the economy was doing terrible. During the Trump presidency's 45th, and now this one, Democrats over-indexing it's worse. I think it's like 20 points of difference in terms of approval ratings for these things. And so that phenomenon also will happen when interpreting any business relationship between private companies and the government. And when you have the president of the United States

bring every giant leader to the White House for a dinner, that can cause challenges where you just assume causation, correlation. At least these things are happening in public. That's the one thing I'll say. And we do have a pretty good system for conflicts here. So to the extent people are, you know, doing anything nefarious,

that is going to get investigated as we've seen, both of these parties. And I'm not a member of either of them. And I'm just praying there's going to be a third party option. And that's what I'm putting my energy into is the third party option. Because I think both of these groups are going to just keep investigating each other. One of them gets out of the White House or the Congress flips. You know, they're going to just start investigating each other over and over and over again. It's never going to end.

It'll be lawfare and investigating. forever. And I think people will generally keep their books clean. Okay, good. That's my guess. Because think about the cost of getting dragged into, oh my God, you did something nefarish. You did some quid pro quo. It's just after what happened with like Hunter Biden or the meme coin stuff with the Trump meme coin, all of these things take an enormous amount of energy to then go clean. or defend, choose may not be worth to squeeze. It might just be much easier to donate to a campaign, donate to a PAC, like Jeff Yoss donates a lot of money to PACs, I understand, or I read online.

Tons and millions in the last election, yeah. Yeah, and he has a seat at the table when it comes to TikTok, right, which is his big breakout investment. He can at least get his message to the White House. Does that mean they're going to do everything he asks? Obviously not. But I think that that's the nature of our country. is you can give a large amount of money to packs.

Maybe that's the thing that has to go away. Two things have to be fixed. One, we have to get some way to prevent people from being able to drop $100 million on an election. Love to your friend, Elon, but I do think it... Oh, there's multiple people who've done that, right? Soros, Elon, Jeff Yass. I just want to point out that I'm not trying to make some sort of subtle dig.

And then also gerrymandering has to go. Because then you end up with people that are only on the extremer sides of their parties. There's no one in the middle like yourself. And then we end up with no one making any progress. But Jason, just to back up what you said, I pulled the chart. This is the one that I was thinking of. This shows how public opinion about the U.S. economy changes by administration. So no Republican thought the economy was good.

Trump could elected 100% dead. Biden got it flips. It's both parties that are doing this, except for COVID. Look at the spread between public opinion on the U.S. economy. Democrats think, like, 60% of them, wait, wait, this is very good. yes. So during the Biden presidency, look at the gap. It's so polarized. 60% of Democrats at the peak thought the Biden economy was great. And Republicans, like 10% thought it was great. It's a 50-point swing. Yeah. It's insane. And it's so dramatic too, because you can literally just right here,

this is where Biden got elected, right? Or inaugurated. And then immediately all the results. It just flips. So that shows a lot about our country. And I think we need to do. get as far away from that as we can, which is why I really think we need to pass something on gerrymandering. Yeah, that feels like another leak in our system. Our system has leaks. You know, sometimes you've got to plug the leaks. You got to iterate on them.

Things change over a couple hundred years or 250 years, you know, since the founding of this experiment. Like, things have changed. You've got to tweak them. Hey, can somebody ask producer Claude? What that phenomenon of your opinion tracking your party affiliation according to the economy is? I know there's a specific psychological term for this.

So I'll just ask producer Claude to do some deep research on that for us. And we'll break back in when we have it. All right, what else do we have in the news? So, Jason, next up, we're talking about a company called Factory. Sequoia backed $300 million evaluation. $50 million raise, N.A, J.P. Morgan, and InVIDIA, also in the deal. It's another company that wants to help you write code. Their product is called droids, what they call their AI agents.

And their idea is that you can have. a lot of them working at once, no matter where you code, be it in the web, in your IDE, at the command line, whatever you want, they follow you around. Cool. Fair enough. The company says it's going to reach, like, maybe like a 20 million run rate into this year, whatever. The thing that I'm really amazed by is how much money is flowing into the space. The FT reports via pitchbook data that $5 billion has been invested into AI software development startups in the third quarter, or $7.5 billion, if you want to count in the windsurf exit to Google. So what do you do if you're a founder?

Because on one hand, it is clear that people think the TAM here is enormous. Sean McGuire said that the market for coding assisting startups is going to be $10 trillion. But at the same time, the competition is so rabid because the TAM is so large. So how should founders kind of balance going after markets where the TAM is very attractive and also balancing out how many more competitors are going to have by going after those larger TAM environments? I was just curious what you thought about that. Yeah. This is going to be a huge base.

I've heard there's 30 million developers. I've heard there's 50 million developers globally. And I think the reason this is such a huge TAM is because of that fact, right? 50 million developers. So if the average value of a developer is $200,000, you know, 50 million times 200,000, that's a big number. I'll have you pull that up in a second. But then you can then say, well, if you may, if you can make, if you can double the number of developers with this software, so people who are bad developers who aren't in that 50 million group, if you can just but double it to 100 million, you've now doubled the TAM. You've doubled the number of developers. I think that's what's happening here. We talked about induced traffic earlier. You make more roads. You make more lanes. You get more cars because people will decide to live further and further from their job. the city center, that's what's going to happen here.

These things are going to take the number of developers. And I would say very easily in a decade, if we had to make a bet here, a polymarket of how many developers there were from now until 10 years from now, which is people who, as their primary job function, primary job function, not like some of the time, but primary job function, push code and develop software, just develop software. because that's obviously going to change, like what that means to develop software. I would say we'll go from 50 to 250 in a decade, 5x in a decade. That's an absurdly fast growth rate, absurdly fast.

It also begins to become an actual large chunk of the human population. If you could extrapolate out how many people are going to have at that point in time, 250 million, no matter how you cut it up, is going to be a single-digit percentage of the total amount of humans. Jason, you wanted 50 million, sorry, 50 million developers times $200,000 a year. That turns out it's 10 trillion. I wonder if that's where Shaw McGuire got the number he was talking about, kind of a very simple bit of calculation. But that's, okay.

That's a big tam. We all agree, big tam. But when founders see a big tam and competition, this rabid, should that push them away from building that space or should have drawn them towards building to that space? You should be drawn to going into that space. If a lot of people find a great beach to surf, you probably want to surf that beach, right? Because it could get better and better, and people could be underestimating the Tam.

So if I told you when sidecar and Lyft were battling it out for ride-charing, should Uber join the race? Yeah, they should. If I told you before XI had released an LLM and Chat GPT and Claude had and Gemini had run away with it, should XAI jump into the race, you might say, no, it's too late. it's never too late. Somebody could start an LLM right now. I know that sounds crazy, but if all of a sudden Apple decided this is strategic for us, let's put $100 billion into it and let's double the offers everybody else have and let's find a way to put this on every iPhone and we'll design the entire iPhone around the idea of it being an H-100. And the idea of building an H-100 competitor and competing with Nvidia, okay, it sounds crazy. But if there's a big market, there's always going to be room for

innovation. And even if you lose in a big market, the bronze or the honorable mention could be huge. Could be huge. I mean, Lyft investors were super bummed that they got their asses handed to them, you know, by Uber. And they got the silver, while the bronze, depending on the market. Silver or bronze still pretty good. It's like a $15 billion company. If you invested under five billion, you at least tripled your money or something to that effect. So still a good win. I think this is a space where there's also going to be things that emerge that you didn't expect. So for Uber or for Airbnb, I don't think in their initial business plans, premium dedicated locations for staying somewhere, like entire apartments or houses or complexes being built

to be dedicated Airbnb inventory, I don't think was ever in the business plan. And I can tell you, Like, as amazing as Travis was of pointing out where this would go, I don't think ride sharing was in the original business plan. It was always going to be Lincoln Town cars. Ride sharing came after that, like, as a concept. And so here we are, folks. Yeah. Partisan bias, by the way.

If we asked, we asked producer Claude, partisan perceptual biased or partisan motivated reasoning is the cognitive biases. You know, you have these cognitive biases that everybody likes to talk about when you're an investor. You like to study cognitive systems, how people think. So people like to think about thinking. That phenomenon is partisan perceptual bias. So your partisan and your perception is biased. Makes total sense.

I have stock market bias. My view of the economy is where the stock market is because then I either have more money or less. So right now, economy is great. And if you are not in the market, like 50% of the country is not, then their view of the market is their hourly wage, which is why we have a bifurcated, you know, when you look at the moderates, that's where you can kind of start to get a really good idea of what the country actually thinks, because you take out the partisan bias of the moderates, with moderates. And wages, not moving is how people perceive the state of the economy.

like, wow, the stock market's out an all-time high. Bitcoin's at an all-time high, housing, all-time-time-high. How can people feel bad about this economy? It's like, they're still making, you know, $12 an hour, $14 an hour, and it's just not a living wage, and they're the primary breadwinner in their family, or they're, you know, a duo of primary, you have two primary breadwinners both making under $20 an hour. It's kind of tough to live a sub-20 an hour. You have to make $30, $40 an hour in this country to sustain yourself. So really, if major cities aside, New York, not everybody is entitled to live in Manhattan, Beverly Hills in San Francisco. You take those three cities out and you start looking at every other city in the country. Yeah, you can live off 50K, 60K. If you have two people

making that, you can live off of that. Going to public school and, you know, it's not easy. I'm not saying that. But man, under 20 bucks an hour, you really need to figure out how to get above 20. And, you you know, do it fast because you're not raising a family on 20 bucks an hour, 40 bucks a year. So, Meta's AI team, the much vaunted, TBAH, Alexander Wang from Scale AI, the smartest folks in the world, Jason, nine figure salaries. They've announced a new product, and it is called vibes. And I'm going to play it for everybody. Sounds promising. What is it? Here it is. Coming to you live.

I know what it is. I know. Here we go, folks. All right, this is the promo video, and this is part of meta-AI. So there's a meta-AI app, by the way, folks, that lets you use your meta-glasses. And so here we're seeing a bunch of AI videos that you can create with tools, and you can make your own AI slop. And then it's a feed of really increasingly better-looking AI slop. So this is TikTok Slop. This is Tick-Slop.

Slop talk It's slop talk Tick Slop I think you won I think Slop talk It's Slop talk Yeah But it's not even good

So we're watching a guy Kayack and there's whales Flying overhead Or an astronaut on a bike Or a bear fighting Another animal Or it's just To me this is just

The absolute nadir of imagination It's like what if X was Y And I don't know why They built this They got a lot of criticism For Jason I'm gonna go ahead and cut off the,

um, whatever the hell that is. Uh, everyone dunked on them immediately. Uh, Nikita Beer, who now runs, uh,

a good chunk of X, made fun of it. Alexia Bonazos, formerly AICF tech runch, mocked it. Uh, even Sean, the CTO of Palantir said,

read the wall of potteats that categorically reject this AI slop. We are so back. So everyone just despises it, right? Thinks it's terrible. I'm going to steal man this for everybody.

Okay. Because I, I think, I think, I think Alexander Wang is getting unfair hate. This is not for us. Everyone who's on X does not want this, does not need this. But the core Facebook audience, right, which is probably where they want to eventually bring this,

is full of older users who famously and historically love AI slop. You heard about the Shrimp Jesus phenomena over on the Big Blue app? Go ahead and explain that. For some reason, for a period of time, there was a ton of people of the boomer generation sharing images and memes of essentially, us that looked partially like a shrimp. It was all AI, and I never really understood it, but it was a huge phenomena. So I think because older folks love AI slop, what meta has created is a slop trough that cost them little to put together and might dramatically increase their

advertising inventory with a critical and lucrative demographic of users. So it's all... So they got rid of paying content providers for content like Disney has to. And they got consumers to make content. And some platforms like YouTube share that revenue, X.com sharing revenue. And now they've decided consumers creating the content, we're just going to have AI created and not have to pay anyone.

No, this is users actually making the slop. I'm with you. I think this is going to become, I think if people use it creatively, they're going to come up with interesting ideas, like they could make little mini stories or memes. So I saw somebody raise $3 million a startup for meme lord.com. Did you see that?

They kept at mentioning me. I did not see the someone meme lord.com. If you go on to X, they made a video of me. And so the keyboard is the new battlefield, meme lord.com. I'll pull this up for everybody. It's got like an 80, it's got like a 90s, Windows interface. And you can sign up. And then, yeah, you go make memes.

You make memes. This new phenomena, by the way, of having the website look like a fake computer. Do you see the new post-hog website? They did the same thing. I have not. Oh, there it is, yeah. Yeah, making a desktop. Yeah, this is a new trend.

Jason, this is just a meme maker? Yeah, because there's like other meme makers on the web, but I guess the idea here is this is to make movies. And yeah. Oh, man, I don't know if I love this new hype train. Actually, this is a good topic for founders. I've seen a lot of people talking about launch videos, building viral hype, and really trying to leverage and hack social media in a way to draw outside attention to your startup.

And, of course, attention is good, Jason. Yes, attention's everything. Well, I mean, famously according to the Google paper. But do you think that becoming a meme advanced user, I'm not going to say master, is it actually a skill that most founders should have? Or is this a younger slice of the founder demographic that should get really good at this stuff? The memes are a way to make people feel an emotion in a way that a logo used to. to do it. So I look at them as something between a joke and a logo or a quip or a clever title for a magazine. They kind of fill that space. It used to be like magazine titles. We sit in a room and we

tried to come up with a really good one. And when they were at their best, you know, and people always did the same one. I am pitching a story about Microsoft and their new product. And it's called The Empire Strikes Back because isn't that clever. And I'm like, this is literally, every third pitch session, the Empire strikes back as a, you know, concept for a meme comes up. And then we're going to like make a montage or, what do they call it? Yeah, like a montage of like, you know, Bill Gates as Darth Vader or something. And Luke Skywalker is, you know, Steve Jobs or something. It's just all ridiculous.

So that's where I think this sits is where memes sit. And I do think there are opportunities here to build a decent size business. and I think there'll be a new art form that comes out of things like Meme Lord or this vibes. I think there'll be a new art form that might not be as sloppy as it appears today. Okay. The slop might turn into art. People said rap was like slop. They said electronic music was slop when it first came out.

And then people were like, rock and roll. It's just noise, right? And so over time, people have learned to appreciate or people have elevated things that started as slop. I'll just leave it at that. So we should all essentially invest in shrimp Jesus and put down our fine art and invest in boomer memes. Okay. You're already here first, folks.

One more, Jason? Do you have time for one more? Yeah. What do you got? All right. Let's talk about ads coming to Open AI. Now, we have a clip here from an older twist episode that I'm going to play because we have talked about ads coming to Open AI.

before. Let's take a listen to these fine folks named Jason Calicanus and Alex Wilhelm. Oh, those guys are smart. Yeah, here they go. It could be cost per click. It could be cost per acquisition. It could be cost per lead, cost per click, or it could be cost for transaction. All of those are on the table. Let's do a long bet. Advertising and chat GPT. Within 18 months, they launch an advertising test. In 18 months, you take the over, the under. Oh, the under. By, by, by, by 12 months. Over under. Under. Okay, 12 months it is. I say, I take, I take the over. Okay. You take the under. A hundred dollar bet one year from now. So, I was taking notes on my physical notepad. When, what date was that? And how far are we out from the bet here? That's the

other piece of information we need to know because I'm going to have to ship you a Hyundai. Oh, Jason, that was, that was a month ago. Okay. So you got 11 months for ads to show up. Now, it's looking good. Your bet's looking like you're in the money because of these factors, these clues, these breadcrumbs. So a couple of things. First of all, if you go back in time to April of this year, OpenAI did some reporting on revised OpenAI revenue growth. And if you take a look at this chart right here, Jason, which I pulled from their reporting, check out theinformation.com. Love those guys. They expanded their revenue projections here on the right, and they added this little bit

called new products, including free user monetization. Essentially, they were telling their say ads. Free user monetization, Jason, does not necessarily mean ads. It could be... Selling their data? Selling their data, stealing their blood. Who knows?

But they're expecting quite a lot of revenue to come out of this. So that was clue number one. Then most recent... They expect $25 billion, it looks like. Is that right? In 2029, they expect to have $25 billion in ad revenue.

Yes. Actually, while I talk about OpenAI, producer Claude, can you get me the latest on Amazon's advertising business from, let's say, of this year. I just want that number for reference in a couple seconds. The other thing to know is that according to basically everyone online, the moment the Open

AI announced a new product called Pulse, which is currently only available to people on their pro tier, the $200 a month tier, it's essentially like a morning app for you. It takes and all stuff you've talked about, tries to find stuff for you to give you a morning shot of information, I guess, something like that. And people are like, wait, it wants to know about you and show you things, ad platform. And then finally, Alex Heath on his Sources newsletter, published Alex. I like Alex. I respect

Alex because, you know, all these writers are just not good at writing. He's a good writer. I like to read his stuff. And now he's gone solo, dolo. He's now independent. Let's go ahead. I'm going to drop a screenshot here of his article just to give him some love.

Sources.com. Sources.com. I pay for this. I pay for Sources. Not news. I'm going to start. Actually, I admit that he gone off on his own. I knew he had a it was only a week ago, so, or two weeks ago. You didn't miss it by much. I didn't. I feel better then.

Anyways, the headline here from sources is OpenAI is looking for an ad's chief. Fiji Simo, of course, previously the CEO of Instacart through the IPO is now CEO of applications over at OpenAI. She's going to hire someone underneath her to sell advertisements and also work on monetization. So if you take the plans to grow revenue from ads, the new product that's good for ads,

And the fact that they're hiring an ads person, Jason, it may be that ads are coming to Chad GPT. Ads are coming to Chad GPT. And, you know, the reason I took the over on 12 months is not that I didn't think they were going to do it. Just how quickly would they do it? If it's growing really fast, you know, Sam Altman is lived through Web 2.0 participated. And he watched people just grow user bases and get them addicted to products before turning on monetization. And so, and, you know, the Zuckerberg's, you know, Zuckerberg and Google would be two really good examples of that. Gmail has marketing in it today. I understand I don't really use Gmail anymore. I use superhuman. But I understand when you go to your promotions tab, they insert promotions from people you didn't subscribe to. Have you seen that? I wound up on my Gmail account. And in the promotions tab, I saw, you know, like if it was putting J. Crew or Tom Favre, Ford in there, it would go put Amber Cromby and Fitch in between them as a promotional email being served to me that I didn't sign up for in my promotions tab. And they didn't do that until whatever, year 10 or something, maybe. But that to me is literally how big companies fail,

because they had to find some place to sick a new ad and they didn't think you'd care. And you know what, the first seven times they sweep the knee. I don't fall down. But eventually I'm like, you know what, F off. You and I are weirdos at the tip of the spear. That's not how anybody else thinks. Nobody cares about advertising. It's only like 10% of us care. That's depressing. But we do have the information from our dear friends of the production team. Amazon's advertising services revenue was wow. Okay, I was off by a couple billion. About 15.7 billion in the second quarter of this year, up about 23%. So, thinking about open. AIs user base, Jason.

700 million weekly active on chat GPD, I think is the last number we saw. How much money do you think they could pull out of that per quarter? If we presume that, let's say, 90% of them are free, so call it 630 million. Let's just say they're all, they all have ads, just for, so 700 a week. And so I think you would just look at how much you could make from them yearly, which you could then look at how much does Facebook or Google make from them. it's a big difference when you have boots on the ground in India, Japan, Australia, and you have ad teams there getting ads online.

So the international ad network's going to take a long time to build. You know, Google took decades to build it. Facebook took decades to build it. But if we just said U.S. users, I could see them easily monetizing with light advertising at $20 an American. $20, if there's 100 million Americans using it a year, you know, that would be $2 billion right there. And that would be very low monetization.

You're talking about making $1.50 a month per user, which means they clicked on but two ads a month, that's $0.75 a click or, you know, 1.5 ads, if it's a dollar a click, let's say you get people to click five times a month. Now you're going to make, at $1.50 a click, $750,000. Let's say these are really high-quality clicks. So we're just making it a dollar, make it easy.

So let's just say the average user, Some users might click on 20 ads, some might click on zero. But let's say the average user clicks on four ads per month, $4 a month, $12 months a year, $48 a year in ads per user. Now you compare that to the people paying $15 or $20 a month. Free users are making $50 a year, and then the paid users are getting to $250 a year. However, over time, they could make very aggressive ads. I think the way they should do this is an interstitial on the free product. You type in what you want to do.

While it's thinking, it's showing you the ads. Oh, so it doesn't slow your work down, but it monetizes your wait time. That would encourage slower AI models and slower inference, which actually would save them on compute because they'd have more time. That's a perverse inverse incentive. Yeah. Let me ask my team to make me a version of that. Show me a version of that.

Somebody designed Lucas or Marcus, like make a little mock up of that. Lucas is going to make a mock-ups of it's doing deep research. And while it's doing deep research, I say, hey, what's the best? I ask it a question, like, what's the best vacation I could go on with a family of five? And I want it to be warm and it's in December. And while it's doing that, I just see the ads for that, right? While it's doing its research. And it shows me three different options for cruises, whatever.

Man, that could be so powerful. And what if it was an ad? What if it was a video ad? in the interstitial page. So it's deep thinking about your answer, and it just plays you the ad while it's waiting. And it counts down in 10 seconds, you'll see your answer.

I think anybody's, nobody's gonna complain about 10 seconds or five second interstitial. Yeah, you'll pay. You're a part of the 10%. I know that I'm a freak, but I'm still mad that they put ads in my paid prime videos. Anyways, I'm trying to get to say that

the question about monetization for Open AI in the realm of advertising is actually gonna be very variable based on where their users are. So Jason, if you look at this, this is from SNAP's latest earnings report, and it shows their global average revenue per user, basically flat year-over-year, about $2.87. However, in North America, $8.33, and then in the rest of the world, 96 cents. Now, why do we care about this? Because I think I saw recently that OpenAI launched a $5 a month service in India, probably a large market for AI, given a lot of people that are there online.

A little harder to monetize than the U.S. market because of different income levels. So they're probably going to maybe lean more towards pushing certain markets to pay subscription. Maybe other markets are a bit more monetizable on the ad side. I'm just really curious where the 700 million weekly active of Tad GTPT are, because that will determine quite a lot about how the calculations work out for the ad business. Yeah. You'll make much less per user in certain markets, but it is what it is. You know, just getting to billions of users is, you know, means you're going to be.

print money eventually, and that eventually is coming soon. And you wouldn't have to print money if you go and buy Claude Pro at my URL, which is Claude.a.I. slash twist to get started over there, you get the first three months free. Twist listeners get 50% off their first three months of Claude Pro. Claude Pro is what I use. Alex uses the whole team, Claude.a.I. slash twist to get started. All right, next up on the program, we have a special treat for you. This is our twist pitch deck competition. It's brought to you by our friends at Gamma. And in this little competition we're doing, we're trying to teach people how to make great decks, how to pitch really well. So we got the founders from 10 great startups, all of which have made their beautiful, elegant,

effective pitch decks using gamma.coma. And after we hear all the pitches, one of these companies is going to receive $25,000 large, that's right, $25,000 from launch. And what is Gamma? This is your AI design partner, helping you generate presentations, websites, social media posts, all that great stuff. They do it in a lightning fast way, all without any coding. So check it out for yourself at gamma.com. All right, first up on our competition is Brittany from where to wheel. This is a marketplace for off-roaders and landowners. They went through our founder university and our accelerator class. Brittany, are you ready to go? Are you ready to do your pitch? I am ready. Hi, I'm Brittany, founder of Where to Wheel, a marketplace that connects off-roaders

with private landowners. Meet Caitlin. She's an off-road enthusiast, and she wants to take her Jeep off-roading. But, like most off-roaders, she has nowhere nearby to do so, as current locations are extremely secluded and unwelcoming. And Caitlin is part of the global community of over 160 million off-roaders with over 70 million of them here in the U.S. While searching for places to go off-roading, Caitlin comes across where to wheel. On the where-to-will map, she is surprised to find new off-road spots in her area called W-2W parks, which is our marketplace inventory. On here, she can view descriptions, buy tickets, sign waivers, and go wheeling.

So the off-roaders buy tickets to our marketplace inventory, and we take 50% of those bookings. And the market is much larger than most would imagine, with off-roading being the second largest sector in outdoor recreation right behind camping with a tam of almost 18 billion. And in 2023, 50 million Americans went off-roading, but our product is so convenient that we're already more than doubling this behavior. We found early traction across 100KGB this year, as well as strong product market fit backed by these metrics. It's also worth mentioning that we have no competing disruptors. For Go-to-Market, we have a really, really large brand and megaphone in this space. So the minute we post about a new location of off-roading, it spreads like wildfire.

And on the landowner's side, we've lined up strategic partnerships to give us access to over 30,000 landowners in the U.S., as well as a real estate collaboration that guarantees us penetration into our beachhead markets. And we're going to use those beachhead markets on a tipping point growth strategy to reach mass market adoption. And we're targeting the Northeast corridor here because this is the most populated area of off-roaders with the most severe lack of off-road access. The team at Where to Will were all former founders, and we have over 35 years of off-road experience. And the ultimate vision for Where-to-Wel actually has three silos, a medium-marking silo, a marketplace of locations, as well as insurance, as no auto-carrier in the U.S. offers on-road and off-road insurance. And the time is now for this market, there's booming demand. It's more than dumbled in under eight years.

Supplies stagnant and outdated, and we can also leverage agritourism for quick expansion. Thanks for listening. Again, I'm Brittany, founder of where to wheel, a marketplace that connects off-roaders with private landowners. Wow. As close to a perfect pitch as you could do, incredibly well done, Brittany. You know, what makes a great pitch like this is being super efficient and understanding what's important. And your audience, the audience of people who are off-roaders, they know all these problems, right? You wouldn't have to explain to them that it's hard to find a place.

But if you were raising money from venture capitalists or you're pitching people to come work at your company. You know, they may not understand how big this market is. So what you did really well is you integrated education into the deck and into your presentation. I also love on the bottom you have those nice hash marks for each slide. So I knew where I was going. And this is where pacing matters.

And up front, you did a great job of showing a real person with a real problem in the world and a real solution. Also, you explained the business model very well towards the end. We said, you know, hey, there's a marketplace, and then there's the events, and then there's insurance. What that did was it built credibility that as an entrepreneur, you're very forward-looking, you have a big vision here. And you come away from this kind of intrigued. It's almost like you're bringing me into a world I've never experienced.

I see people off-roading on social media once in a while. But I don't know how it works exactly, and I didn't know how big the community is. I don't know who these people are, but you did a great job of explaining that. Then what was amazing was you talked about your go-to market strategy, and that also built massive credibility with me. So for people who are wondering, you know, anybody can come up with an idea. This idea, I'm sure 10,000 people have had it. Why isn't there a marketplace?

Why isn't there an Airbnb for this? There's got to be a better way. But a small number of people actually go build it. And so there might have been a dozen people who attempted to build this and just nobody succeeded or they didn't have a good go-to-market strategy. You really studied the space and you said, hey, the northeast corridor has the most density. It's got the largest amount of people. And it's the lowest serviced. So I would have guessed, maybe they'll go with Texas or, you know, I know in California there are beaches people go to.

I know in Florida people go to beaches. But, you know, I was just thinking Appalachia, I don't know, where's the best place to do this? Where there already are locations, there aren't locations. And I think Airbnb went through something similar. They were in California trying to get this sleep on your couch, rent a place. But in California, there's plenty of places to say. It's a huge giant state with many options. Then you look at New York and New York was expensive.

There weren't a lot of options and they typically were overbooked. Whereas in California, you can always get a hotel room. It's just a matter of if it's the one you like and how far it is from whatever event you're going to. It's almost never the case. Maybe during Coachella and the hour drive around it that things become so booked up that you can't find a place. So at some point, Paul Graham did the really smart thing and told them, hey, why aren't you guys in New York doing this in New York? So I love that as well, just across the board. Check, check, check, check.

And it feels super credible. Now, what could you add to it? That's a question I have for you. When you've pitched it, this deck specifically, what are the questions you typically get? Because you don't want to in a short three-minute presentation or less like this. You don't want to answer everything. But what are the first questions you've gotten from investors after, you know,

or a quick presentation like this. That's a great question. So usually the top two questions we get are, you know, can you, like, how are you defining the market size, right? So we go into that really quickly with 160 million international, 70 in the U.S. So then I go into, when they ask that question, we have statistics and data points, different research that I can share with them on how they're actually qualifying that. But we used to just say, you know, 160 million off-roaders.

So then we kind of broke it into the different bucket. so they could actually see, you know, more of a U.S. market versus an international market. And then obviously, go-to-market is always one, like, how do you reach them? Where are you hitting first? So we actually added the B-10 market slide in there. Instead of me having to answer that all the time, the questions, we built that out and added that in. So I was, it's nice to see that that helped. Well, you know, every time you pitch somebody, it's an opportunity to learn something about that investor, where they're coming from.

and, you know, 95% of investors are not interested in any particular startup. And the high art of pitching is finding the five who are into it. I know this because I introduced Uber to 21 people and three of us invested. So in that case, you know, it was literally 95% we're not interested. Right. Which is crazy when you think about it. And I love the fact that you got granular on the market size in the Northeast, even down to the state level. If I were going to say, you know, something to consider adding when you do the longer one,

I'm very curious about who these people are. And that falls under a persona. And you did a persona, a woman at the start. She has a Jeep Wrangler, I noticed. But I didn't know too much more about her. This is where I wonder, if you were to bucket, say, if I said, give me the three top personas of people. Tell me about those three top personas. So in our market, it's a range.

You got men and women, right? And so if you're going to go three top personas, you have women off-roaders where our customers are more geared towards women that are newer to the sport, right? They're looking for a safe space to go. They don't know a lot of information. They don't want to get lost out in the middle of the woods on their own. So we have that bucket. You know, they're usually around the range mid-20s, you know, into like,

their 50s, 60s. The off-road sport goes all the way up. I have people coming out and they're 76 years old. I mean, he just did amazing, but I was like, good for you. So usually it's about 25 to like 60 is usually the range. So females into outdoor sports looking for a community, something outside of technology, you know, to get into. So that's one bucket. Okay, so you'd call those touch nature. And there are a wide age group, and that's women specifically as one group. And you've identified that women are adopting the sport in some major way. Is that correct? Yeah, we thought it would be mainly men. But unpack it. What's driving that? Is it a specific group of women? Is it single women?

Is it women with kids who want to bring their kids out to nature? Are they looking to meet people or they bringing four of their girlfriends? Unpack it a little bit if you can. Yeah. So this one actually really surprised us, right? Because we were under the assumption. This is a more male-dominated field or sport, which it still is. But the off-roading industry is just absolutely exploding, right? You see every vehicle manufacturer is bringing back or adding an off-road market. They're targeting women now as a sport. You know, you can learn how to do this.

It doesn't take a lot. You know, you have a Bronco. They want to drive with their, you know, the top down, the hair blowing in the wind. There's a Jeep hair don't care. That's like one of the things they say. Interesting. That's wonderful that you've identified that persona. Let's go to the male persona, the classic persona that makes up the majority.

So that's like an emerging group, I think. Who's the majority? Who's the majority? Is it, you know, I was thinking like it's a redneck, you know, somebody from down south. But then I was also thinking, you know, my friends who do this, who I know are rich venture capitalists and CEOs who are spending, like they buy some Jeep for 30, 40, 50K used or slightly used. and then they spend another 30, 40, 50K on it. So I wonder if you could unpack, let's just say, the elite part of the audience, and then we'll get to the big major middle.

But who's that, like, you know, dream customer with a lot of money? Or is there one? Yeah, I'm curious. On our side, you know, it's one of those where if you're willing to pay to do your sport, you know, there's kind of a certain connotation that goes there, right? Like you're not the redneck rodeo out in the middle. of nowhere, you know, with a rust bucket. You know, you're going to go on your friend's farm for free. Like, you're not our customer and we're not really interested in that. So we kind of price it

based on the certain region and what that clientele looks like. So like our Northern Virginia markets, D.C., Loudoun County, the richest region in the United States has a different price and kind of, you know, offerings to cater to those people. So on the more elite side, I mean, it can range from any kind of occupation, lawyers, dentists, doctors, vets, you know, up to venture capitalists. We have a lot of CEOs and executives that come out and they're usually driving, like you're saying, you know, those 100K vehicles, you know, the G392 or we had a brand new Bronco Raptor.

It was like 120K from the dealership with no plates came out. And it's like, oh my gosh. So they're just, you know, they're looking for a way they can bring their kids with them. A lot of dads bring their kids. You know, the moms like get out of the house and this is what they go do. They can come do it for a couple quick hours and they get back, you know, get out to nature. I'd love to do that with my kids.

So let's talk about that group. I'm wondering if there is a premium service that you're charging typically 50 bucks, a hundred bucks to go run around on somebody's land. Is that the average ticket price? Yeah, the average is 55. It can range up to 150. are more expensive markets, you're low-end you're going to have to get 65 or more per vehicle to get in the door.

So me as like an elite customer of this service or somebody with the $100,000, $200,000 car, et cetera, they're a dentist. They're going to look at that and say, that's too cheap. I want to have a place with great bathrooms. I want to meet other people with this similar type of car. I want an instructor there. I want waiter service or a beverage counter or a latte. Is there an equivalent of the Uber Black, if you will, or the One Fine Stay, if you know, that company, One Fine Stay does, like, elite Airbnbs, I believe. And then Airbnb has Select, but, you know, is there a One Fine Stay Uber Black version of this?

Currently, no. So what we're doing to try to combat that is we actually partnered with Constellis, which is, it used to be Blackwater. It's a massive facility out here. So they have 5,000 acres. They have cafeterias, hotels, bathrooms, you know. medical on site, you know, all these things that create more of a premium experience. And that ticket is more premium ticket. And so, right now the one down here, it's, if you buy early bird, it's 150 up to a certain amount of people. And then it kicks up to 200. And that's about three to four hours of off-roading. So you get a little quick snippet. And that's worked really well. And so we've looked at partnering because there's other facilities like this across the U.S. and at making those more of like, if you want a, you know, black level experience, that's kind of where you go. I would like to see you push the envelope on this. And so I think taking a shot at a $1,000

experience that included some merch that you get, included waiter service, premium bathrooms, and an instructor. And it was targeted towards people with certain vehicle types and maybe some instruction. I think there's just so much opportunity here to capture some of the high end and learn about that. People greatly underestimate what people are willing to spend on experiences. I know this because over time, as I was able to move my station in life up, I would go to Disneyland on a Tuesday because I didn't want to hit the lines. Then I was able to afford a fast pass. Then I was able to afford like the VIP tour, which, by the way, I think is like, I don't know, $600 an hour and we do it for 10 hours or eight hours. And then I have a friend who took his thing up to another level with

club 33, their private club membership. And that's like a quarter million dollars or a half million dollars to do it. And then maybe $30,000 a year, something not insignificant. And there always is another level, another level. So, I think sometimes exploring those. Now, the one thing we didn't talk about will end on this is the manufacturers and all the gear people. You don't have a sponsorship, partnership anywhere in the deck. But I think understanding in the persona of the top two personas that you're targeting, I think understanding their salaries on average in who they are, what car they drive,

the amount of money in it. I think understanding that would propel this into more of a golf or a skiing type persona. And if you ever notice, the car companies market to golfers and skiers, right? Especially in your category, actually. So they're always, I get more defender and I have like the suburban, fancy one. I don't know what's Grand Wagonier or something. It's got some fancy name with the bigger engine and, you know, whatever, better, leather. So I think that group, to me, could be absolutely magical and change this from feeling

like it's, you know, poor to middle class folks to, hey, maybe people need to understand from the presentation, this is high-end folks. So anyway, I just think so much to noodle on there. I think education, you can really hit a home run where you just sell people, you know, a $2,000 weekend where they learn. And there must be people out there who have those businesses. already, I think either including them or acquiring one or making one could be a total unlock because now you, you know, all those people who don't do this yet but who have the car, my Lord, like what percentage of people who own a Land Rover defender or a suburban have taken it on one of these courses, we'd just say?

We actually just had a defender last weekend. So we just launched classes to help with this. So because we get a lot of beginners. Yeah. Back to my question. Of the, if there's a hundred thousand Land Rover defender, like the new fancy one that have been sold, let's say. Yeah, the pretty one. Yeah, too pretty.

That's why I didn't buy it. Of those, there's like a Grenada or something. That's like the old school one. I don't know if you've seen that. It's a Grenada, I think. That's the one I want to buy. So of that group of 100,000, let's say, Land Rover Defender's fancy new one, how many of those have been off-road, do you think, and done a course? With us or just across the board?

Just ever. Probably 20%. You think 20,000 have one in five. I thought it would be like one in 20. But okay. So that means those other 80% might like this experience. So back to Tam, if you can show, I think one of the things people will want to see

is you're growing 10 to 20% a month, month over a month. That makes us a fundable startup. So that's one. At a certain point, people are going to say, okay, you're going to reach the market saturation. What does that look like, et cetera? And so you can answer that question with, and this is growing, and we are helping onboard people as part of this revolution. I think you will cut the Series B investors off at the pass, so to speak. So great job. What's the growth like month over month? What have you been hitting 10, 20%, 5%.

Yeah, somewhere anywhere between about 15 and 40, depends on like what we've onboarded that month. Yeah. So this is what I would say. The most of the most of the most. important thing is you show you can grow your marketplace in terms of, you know, the actual GMV 10 to 20% a month consistently. Once you've done that for 10 months, for 20 months, all of a sudden, the funding discussions, all the questions go away. So there's two ways for founders sometimes to figure this out. One is they figure it out by, you know, not getting funding because they're not growing. And then the other way they figure it out is they put a couple of months of growth together.

And then people are like they throw money at them. So the chart, sometimes a chart will get an investor to take a meeting. Sometimes a chart will get an investor to write a check. Okay. So just keep that in mind. If you have the chart, how many months have you had double digit in a row? About five. Perfect.

So you're right on the cusp of greatness. When you show, and you know, you don't have to go for 50 a month, 100 a month, just to show you understand how to maintain the quality experience and do it at 20% a month, 15% a month, you know, something in that average. Man, the investors will start taking meetings at two to three times the rate. So just keep that in mind. And that's something that's very important with you as the leader of the company to just absolutely create as a North Star for your company. Marketplaces, if they prove they can't grow, they got a problem. Because now the investor community is like, oh, thank you for going down that rabbit hole and eliminating this as a category for a great next marketplace. So amazing job, great job.

And I guess if anybody wants to go off-roading, where the number two, wheel.com. Where2wheel.com and you're Brittany at where to wheel.com, B-R-I-T-T-T-T. And why? If you're an investor and you want to meet her or a partner, great job. Thank you. Really excited for your company. Everybody, if you're feeling inspired and you want to come on the program and pitch me, go to gamma.coma.com and start working on your own presentation. Brittany's got a pretty great one here. You can just take hers and deconstruct it. Thanks again to our partners at Gamma cannot wait to give somebody a cranberry. 25 dime skis coming at you.

All right, everybody. See you next time. All right, everybody. Another amazing episode of This Weekend Startups. Have a great weekend. We'll see you on Monday. Make sure you go to YouTube.com and go to YouTube.com and search for this week in startups. Hit subscribe, hit the bell.

And you'll get an alert when we can go live. And we'll take questions from the live audience like we always do. All right, Alex, have a great weekend. We'll see you Monday. New Jersey. Bye. Bye.