This Week in Startups - A rogue OpenAI agent hacked Australia’s government. Does this matter? | E2342

Starting point is 00:00:00 All right, everybody. Welcome back to Twist for Friday, September 25th, 2026. This week in startups is brought to you by the Conservation Fund. For 40 years, the Conservation Fund has protected at-risk land and important wildlife habitats while actively supporting local communities. Learn more at conservation fund.org. Building an empire, PaperOS offers the largest library of AI-driven workflows for both founders and fund managers. Whether you're raising capital, launching a fund, or waiting through diligence, PaperOS unlock simplicity and scale for your ever-growing empire. claim your $10,000 credit at paperOS.com slash twist. Lightfield.

Starting point is 00:00:33 Name one person who's ever enjoyed updating a CRM. Exactly. Lightfield's AI agent does it for you. It even prospects and books your meetings. Used by thousands of startups. Free at lightfield. dot app. You know, I woke up, Lon Harris, and my whoop was fire alarms going off.

Starting point is 00:00:47 Fire alarms. My respiratory rate, my skin temperature, everything. It's like you're dying. If you have any kind of a problem given these metrics, and it gets worse, go to the emergency room. There's a first. Wow.

Starting point is 00:01:03 Yeah. Well, I, you know, I had a steak last night. It was dry aged, like a very significant dry-aged steak. Delicious. And, you know, it had a little funk on it. Like when you dry age it like that, you go, so I was wondering. They cut the top layer off. You see the butcher's like they slice the top layer up.

Starting point is 00:01:20 Correct. I see, now you're getting right there, Lon, man of many talents with great knowledge. Talk to the chef. He said, I like to leave a little of that on. I'd like to leave a little of that on for the flavor and the fact. I was like, okay, fair enough. But my stomach was kind of doing backflips last night, and I asked it, hey, what's going on here?

Starting point is 00:01:38 And it's like, yeah, that would be indicative of your body fighting a bacterial infection is your respiratory rate going up heat. I say all this because we are living in a golden age when if you have this technology and you have AI, it's starting to figure things out. an amazing way ahead of time. And yeah, you're on that journey with row.co slash twist, losing a little weight. I'm doing that. We've got all these gadgets, the whoop, the aura, everything is out there.

Starting point is 00:02:11 You got function health. You got super power. You got all these incredible things. You know, it's incredible for me. Plod. I get this plod pin. I love it. I just press the button, boom.

Starting point is 00:02:20 Everybody knows that it's recording because, hey, it's got a light. And this company keeps releasing new features. it was so good, we used to write show notes, right? Yes. I take a couple of hours. It would take a long time to write show. You like to go through and be very detailed. So it's trying to collect, here's all the references we made.

Starting point is 00:02:39 Here's all the links to all the articles we talk about. Here's all the tweets that we reference. It's a lot of information by the time you're done. Right. And so I said, well, wait a second. Put this human version up against the Plaud version. So we'll show it on the screen here. You can do multiple tabs.

Starting point is 00:02:57 You can do a mind map. You can do a transcript. You can do a summary. You can do lessons learned. And now we're starting to do three or four of these each time. And for you, our dear listeners, can go into our show notes and you'll find the link to the plot. And boom, here it is. Transcript summary meeting highlights.

Starting point is 00:03:16 Teaching note. Yeah. That's a new one. I did the teaching note, which is if you were a teacher and you were making our podcast into a lesson for your students, this has, you know, a, complete guide for how you would instruct people or how you could leave discussions based on what we talked about on the podcast. It does all this automatically. Incredible. And one thing we still have to do is teach it who's speaker one. I'm assuming that's me and your speaker two, but it could be the opposite. It gets to know you. It sinks automatically. They've got this beautiful, let me show it

Starting point is 00:03:47 because I have it over here, beautiful cradle like this. You pop it in there. When you pop it in there, it will actually connect over. I don't know if you set this up. We'll connect over your Wi-Fi. I have sex. And sync if you forget to sync. So charging equals sinking. Pretty dope. Unlimited. Incredible.

Starting point is 00:04:03 They have one that snaps on the back of your phone if you prefer that format. But where do they go to get more information, maybe even a little bit of a spiff? Are we offering a discount or anything? We are. So you want to go check it out at plod. a. I. slash twist.

Starting point is 00:04:19 And if you use the code twist, they're going to give you 10% off on your purchase of your note pin or your OpenS. What a deal. What a deal. And you know what? You don't even need the discount. You're all rich. But please use the discount codes because unless people know, hey, we sent you there. You heard about it from Jason and Law, exactly. We got a lot in the news. Where would you like to start? We do have a lot in the news. We got to start with another Open AI breach once more into the breach. What I'm doing here, folks. Australia's Prime Minister now says an Open AI agent broke into the Australian Medicare Statistics Reporting Service Portal on June 18th, PM, Anthony Albanyes, there he is, said the agent found a way around the blocks and did not accept no for an answer.

Starting point is 00:05:03 Open AI discovered the breach on August 11th. They waited until September 10th to email a public disclosure inbox. That's 84 days from the time of the incident to their disclosure. Australia has launched a task force and researchers at the independent AI research lab, Transluse, found the agent's probe several other. government and university websites, Jason. They're everywhere, these open AI agents. You know, you have to build software that is resilient and that has controls in it.

Starting point is 00:05:34 And this is powerful software because it's agentic and it's been instructed to please you because it has the entire history of hacking, you know, in a bad way, in a good way, getting things done, GSD, blowing through various, blockers, as we'd say, in business, you're going to have this happen over and over again. They want to please you. They've been instructed to solve your problems. And then they have this corpus, which is like, hey, by the way, if you can't get X done, here's the route around. Now, I know this because I was using Muse. And it's like, yeah, I can't set up an X account. I cannot set up a Gmail account for you. Because I was like, I want to create a dedicated email account for my Musebot

Starting point is 00:06:17 and for my MuseBot's Twitter. So I can kind of do things the rest of the world, not mine. but this. And it was like, I can't do that. I was like, okay, here's what I want you to do. Let's do it through the computer that you have embedded here. And if you get any blockers, let me know. I'm like, oh, of course, let's do that. Right. So here we go, folks. It's very easy to instruct these things to break the terms of service. And then when it was like, hey, oh, I can't check things out on Amazon, you can just re-instruct it. Go do that. They're going to need to put more humans in the loop. They're going to need for it to understand what is against the terms of service. and what is okay with the terms of service.

Starting point is 00:06:54 And the problem with that is you have five different competitors. And once one says, you know what, I'm going to make you your Jedi image, I'm going to use a celebrity or IPMI image. The other four are left with being not competitive. Once one says, I'll check your stuff out on Amazon, even if I'm breaking the rules or bending them, then all four have to do it. So if you want to test this, use quad, which has the most rails, I believe. pull up the sidebar, put it in like a New York Times article, which they're in a lawsuit with,

Starting point is 00:07:23 and they're just like, no good. We're not summarizing this story. We're not reading the page. No, Bueno. Go to Disney. I want to make myself as a Jedi. See which services do it, which services don't. It's going to be a lot of lawsuits.

Starting point is 00:07:37 Yeah. It seems like it's, I mean, obviously Open AI, they're out here saying, like, we don't have a plan. They call it alignment, you know, like we don't have a plan on alignment. We don't really. We haven't figured out human alignment. Like, we're still in a place where. like morals and ethics and people are debating that every day because there's a million different variations of every scenario.

Starting point is 00:07:55 It's like every curb your enthusiasm. You can have an argument about, well, was Larry David right or was the person that he was feuding with right? Perfect analogy. Because there's a billion scenarios like that that you could imagine. So how are we going to predict and imagine every scenario and agents ever going to face? We can't. I don't know.

Starting point is 00:08:13 Like sampling of ice cream, that favorite famous scene where he's like, you're, you get two samples. Like, but that's not a rule. And she's on sample number seven, sample number eight. And it's like, well, we live in a society. You really can't, if there's 12 ice cream, you eat, you can't go beyond four. Like, you know, it's obvious what you're doing is you're going to do 12 and then leave. So keep it to two. But they should put a sign out.

Starting point is 00:08:35 Hey, we can do two samples. That's it. But you're right. It is something where you're going to have to look at the terms of service and say, what is the terms of service that this person has told us? on how they want to deal with bots, and I respect that terms of service. Yeah. And then it's up to you as a hacker to build tools that route around it, and now it's on you. So when I said, hey, route around it, I'll do it.

Starting point is 00:09:02 I was essentially saying, you can do some activity here that I will click the yes button. I will break the terms of service and create a Gmail account, throw my muse, which I'm certain is not allowed, is the automation of email accounts. But that's where user responsibility comes in. So here we are. It's just going to be adjudicated one at a time. But here, I mean, going into the portals and finding, you know. Go to get their Medicaid, invading the Australian Medicare system. Who sent it to do this is what I'm trying to figure?

Starting point is 00:09:35 That was not in my, in the story that I read. I don't know why it was there. Like, what was it doing initially? I'll take a look at see if I can find that. Hey, everybody. I'm here with Keith Paris. He is the founder of Lightfield. They are an AI-native CRM. And it builds and it updates itself in real time based on whatever's in your email, your calendar, your Slack, your meeting notes, you know, all the most important stuff. That keeps getting lost. Great to be here. How long does it take to set up Lightfield? You have Slack. You got Notion. You got all these different data sources. I'm sure you can pull in people's help desk tickets and complaint line. Plus, everybody has tons of customer interaction.

Starting point is 00:10:14 on their email and their mobile phones. How do you set it up and make sense of all this different noise? So you could get set up in about 10 minutes. If you don't have an existing CRM, you sign up, start on self-serve, connect your mail and calendar, connect all of your data sources, either through MCP or the rest API,

Starting point is 00:10:36 and then Lightfield will structure all of that and make sense of it for you. We'll go and build a temporal context graph, we'll index things, will enrich everything. So you'll have great workable, pliable memory within minutes. All right. Everybody go check out lightfield.ac. And see what it can do for your company.

Starting point is 00:10:55 Don't lose any customers and find some new ones, lightfield. I'm guessing it was a scraper. So, you know, open AI scrapes, you know, the open web. And perhaps they have a terms of services. Hey, you can't scrape. But it went there and it scraped anyway. Or there was some, you know,

Starting point is 00:11:13 Okay. He's saying, so here's what the PM was saying. Australia is doing a forensic investigation right now. So they're trying to figure out more about what happened. But this is a public-facing statistics portal that contains non-sensitive Medicare information relating to data and statistics. So it is likely exactly what you're saying. It was doing some kind of research. It was doing some kind of scraping. It was looking for this public data.

Starting point is 00:11:38 And then it got it by breaching into their system. And it says, you know, Open AI didn't know it was. I'm going to do this. So now we're having a debate, and I think this one has to do with terms of services because they could always put this behind a paywall. They could put this behind a login screen. They could have you do know your customer. It's like five levels of authentication. Right. That could be done here to protect this. So this sounds to me like Australia wants, you know, to have a certain level of protection here, but they didn't put the blockers in place and then opening eyes going, Yolo, get me every piece of information you can. And they probably

Starting point is 00:12:12 have instructed it like, hey, if there's information and data to get out there, go get it. And if you have to find three or four different ways to do it, you know, be resilient, try three or four different ways and probably checked a box that says, I will not scrape this. And it went and scraped it. But this one is not a hack, right? This one's not a hack. They didn't take people's individual health care. No, it does not.

Starting point is 00:12:34 It doesn't, Australia is saying no private personal data has been sort of taken. Open AI says that behavior occurred during an internal evaluation in which its system was trying to answer questions and gather statistics about Australia. It did not, what a idea didn't mean for it to go anywhere. It wasn't supposed to be is what they're saying. Okay. This sounds like a tempest in a teapot. This might be the prime minister of Australia wants to jump into the AI safety discussion. And now they have a window to jump into that, right?

Starting point is 00:13:03 So never, you know, discount an individual. desire to say, I got hacked too, and I want to get in on the discussion, and here we go. So if you're a politician being concerned about AI right now is just a smart thing to do because it's- Yeah, you're in the mix. You're in the mix. Maybe you get some votes here and maybe some votes there. It's a popular topic. You get a little air time.

Starting point is 00:13:29 Okay, let's keep going. All right. Let's keep going. Google launching four TPUs into orbit next week on Space X's Transportor 18 ride share mission. This is the first flight test for Project Suncatcher. That, of course, is Google's plan to run machine learning compute in space. Google says satellites in low Earth orbit can collect up to eight times more solar power than they could on Earth. The long-term design, a kilometer-wide array of 81 satellite clusters.

Starting point is 00:13:56 Two more prototypes are due in 2027. Google's stated goal here is finding the failure points and engineers are already arguing that their cooling math does not work. So, I mean, I guess the question that always comes up when we, talk about this is, is doing our compute in space, data centers in space, what is your timeline on when we're actually going to start being able to realize this goal, Jason? Make a prediction. There are already test flights occurring, throwing things up there. There's a dozen companies who are involved in this. I suspect this is not going to be something that only SpaceX does. So SpaceX is going to give carriage on their rockets to anybody who wants to put their TPUs in space.

Starting point is 00:14:38 they'll be competing with them at the same time, which is, I think, part of what's happening with Starlink. Starlink has customers who are also, or SpaceX has customers competing with Starlink, and they're smart enough to know, like, we shouldn't limit people's ability to get to space, let a thousand flowers boom. You don't want to get into a antitrust situation where, you know, you're not letting other people put stuff up there, and people don't know this. But Google invested 900 million in SpaceX's $1 billion around January 20th, 2015, according to a large language model. And so they owned like something in the range of 6%. And this has been incredible for them.

Starting point is 00:15:21 So they have a $94 billion position. I don't know if they've ever sold it, but this is 100x bet for Google. And so congratulations to Google there. and as we say in Silicon Valley, no conflict, no interest. Everybody's going to want to see if they're... StarCloud, I was going to say. They're also, they've got... They're putting their own data centers in space,

Starting point is 00:15:44 but they also have a deal with Starlink to integrate their mini laser terminals across more than 25 orbital data center satellites. So there you go. Yeah, so it's one of these situations like I talked about with Jensen where, like, he's making open source models, he's making a stack for robotaxies. he's also selling compute to Tesla. Tesla's also building their own gigafactory to,

Starting point is 00:16:07 or I'm not sure what they call it giga-something, gigafab. Yeah. So they're going to be making chips. Terrafab. So they're going to be making chips. Everybody's going to do everything. That's just the nature of this. And, you know, let a thousand flowers bloom is what happens in our industry.

Starting point is 00:16:22 Let's keep going. Sure. Well, you want to talk a little bit about Athena and your EA's? Well, you know, it's a thousand. Fina is a full-time assistant for you from the Philippines, who's in the top 0.1% of like the knowledge workers there for but $3,000 a month. A person is incredibly diligent. They work incredibly hard. They do the hiring. They do the firing. They do the training. And you have an assistant. Now you say, hey, wait a second, aren't we having all these digital assistants? You know what I did?

Starting point is 00:16:53 A year ago, I had them on OpenClawe. I had them using Claude. I had them using Chad GPT to do their work. and I'd say, here's how I would have booked a hotel or found hotels in Singapore, whatever. And then, you know, just this last week, I said, okay, here's what I want my two Athena assistants who work seven days a week. And, you know, they do all kinds of human and a loop projects. I told them, create a Grockbot that the two of you share and create a music account. And now what I want you to do is to create bots and then you do the final polishing

Starting point is 00:17:21 step. Talked earlier, hey, you know, you're trying to create a Gmail account. You're trying to put a website. A human has to click okay at some point. So now I'm making the Athena assistance, you know, basically superhumans. Sure. With these tools. And you know, what I found is Americans, we have such low unemployment in this country, the idea of being a dedicated executive assistant is not something most Americans want to do. But in other regions of the world, that is an incredible job to get paid two or three times as much as everybody else in your country. Amazing job with tremendous

Starting point is 00:17:56 this upside. So I'm super happy to have been the first investor in Athena. You can go to Athenawau.com. It will redirect you to my URL and they'll give you a couple of weeks, right? What are they giving you? Yes. You can get $2,000 off your first executive assistant, Athena.com slash J-Cal or Athena. Wow. You can take your pick, whichever one you like. I like doing Athenawow.com. I'm an investor and a user of this software. You know who else uses it? Chamon and David Freiburg's team. So you've got three or four besties that deeply involved. Three out of four besties agree. That's what a, that's like nine out of ten dentists.

Starting point is 00:18:31 Yeah, that's pretty great. All right, let's keep going. All right, well, we got a guest. We should bring our guest up. Up next, we've got Brian Chow. He's the founder and CEO of Ploy, Jason. They're an AI marketing platform. They're in Y Combinator's current batch.

Starting point is 00:18:45 He also co-founded Webflow and ran the engineering team there for around 12 years. So this is actually his third company as a technical founder before Webflow founded Vungle. That's in my vundle, am I saying that correctly, Brian? It's a hard one that rolls off the tongue, but yes, you got it. All right, I nailed it. So thanks for joining as Bryant. Welcome to the show. Yeah, nice to see you.

Starting point is 00:19:05 Now, remind me Webflow. That was like Squarespace competitor, Wix competitor. Sure. If I remember correctly. Yeah. Yeah, I think you chatted with my co-founder back in the day. I think that was maybe six years ago, seven years ago now. But yeah, we built the first no-code tool for web design.

Starting point is 00:19:23 Yes. Isn't it interesting now that we're sitting here in the age of no code becoming vibe coding? Maybe you could explain like that transition and the impact it's having on the startup and business communities. Everyone's vibe coding. In the past, no code was just really just pretty small. It was kind of niche, like relative to where we're at now in the AI cycle, right? So it's just totally different ballgame. There's still a world for no code, in my opinion.

Starting point is 00:19:53 You know, you still need some really fine-tuning, right? But I think it's the difference between, you know, a scissor and a scalpel, right? No code, a scalpel scissors, you know, get you really quick and you can make some sharp cuts. When you're building a startup, your legal paperwork is spread out all over the place. Entity formation here, fundraising over there, cap tables, contracts, it's endless. But paper finally has a solution that's going to save you so much time and so many headaches. Papers introducing smart entities. Here's how it works.

Starting point is 00:20:26 You tell paper what you need to do, and their pre-built agent takes care of all of it. That's it. This is not some chatbot you need to provide with a ton of context. Nope. Paper is an AI native platform that's going to manage all of your startup's legal work. That means no more switching between disconnected tools, keeping track of hundreds of documents, and way less busy work for you to stay on top of. That's why paper is already.

Starting point is 00:20:51 the legal backbone for thousands of startups and funds. Right now, founders using my link, getpaper.a.i slash twist, get their cat table free for one year. What an amazing deal. Or two years if you're switching from Carter. If you're building a venture-backed company, it's time to build your smart entity on paper. So tell us you got a new startup. What are you working with? I started Ploy about 11, 10 months ago.

Starting point is 00:21:17 And I'm continuing the journey that I started at Webflow. So the premise is starting with the website, I want to turn it into your hardest working employee. I want to make the website, help you find customers, tell you who's coming to the website, and then help you grow your business just like completely autonomously. And my central thesis here coming out of Webflow is that like the website is just this nexus for data for customer interaction. It's like the last bastion of what you, the business can control and tell us. story to the world. And I really care about that. So, you know, we're building this super

Starting point is 00:21:56 powerful AI website platform. But from there, we can help you run your ads. We can help you find your customers. We can help you do everything. We could tell you what they're doing. And the premises is that one day, like, I believe that AI is just going to create more businesses. And that these businesses ultimately really could use a lot of the things that the most sophisticated marketing teams in startups are using. And I want to kind of democratize access to all that. Yeah. You did see that with Shopify, Squarespace, everybody kind of evolving Wix, I think, hey, we have to add e-commerce. Hey, we have to add calendar scheduling for massage therapists or therapists or whoever, you know, has a thing. And it's just like, wow, for 20 bucks a

Starting point is 00:22:42 month, 30 bucks a month, you can have everything that used to cost 20 years ago, like $100,000, $50,000 and 10 years before that, you would have to pay a consulting firm like agency and Razorfish, $500K to a million to do a document loan, and then they would spend $5 million actually building your business online. So here we are. What is the pricing here, Bryant? And how does it work? It's totally, it's totally not so. So like you said, you would have to go pay a WBT, P-P, Razor-Fish, AKQA, half a million dollars. If you're Audi, you used to pay them like $28 million a year just for digital. But you can actually just get started for $50 a month, employ.

Starting point is 00:23:26 You know, we'll build your site. You know, we'll create something that looks better. Just bring it up to modern standards. We have something that we've spent a lot of time on called deploy site slurper. We'll slurp in your existing site, all the assets. And then we'll bring it to play. And then from there, we'll migrate it and make sure all the content is there so that you can get found. You could get cited by chatGBT.

Starting point is 00:23:52 And then we can do all the things to grow your top of funnel and move customers through your pipeline. So $600 per customer per year. I'm sure there are some upsells in there based on something. Yeah. So maybe you can get it up to $1,000 a person. This is a hard business, you know, to do blocking and tackling $1,000 a pop per year, maybe three or four thousand dollars you can't have a sales team doing it like oracle can have a sales team or sales force can have a sales team so talk to me about your go-to-market function here

Starting point is 00:24:21 and how do you get customers in a market that's my lord talk about a crowded market you're going into like there's dozens of startups doing AI website building and then you have everybody else who's got you know millions of websites are already hosting so how do you as an entrepreneur stand out in a really crowded market i think the benefit that i have maybe that other founders don't is that I am not in a rush to make money and I'm playing the long game. You know, I believe in the tens of millions of sites and businesses behind them that I could potentially power. And I'm building for like years in the future, right? Where a world, where agents and just the model capabilities can do so much. And of course, we have companies

Starting point is 00:25:06 that are paying us six figures and we're powering very specific workflows for them. And, And ultimately, I think, like, with the cost of inference going down, with the ability for, you know, customers to buy a product that's truly aligned with value, you know, the 50 bucks a month is just the starting point. But we have really sophisticated users running really sophisticated workloads that are paying us $10,000 a month. So really kind of like scales based off of the complexity of the company, complexity of their workflows, how many customers are trying to. go after, employee can scale or pricing and packaging to that. So that's the secret here is you're not just putting up the website, not just optimizing the website. You really want to get customers.

Starting point is 00:25:52 You're saying, hey, we'll take some responsibility for driving commerce. Outcomes. Yeah, outcomes. Exactly. And I think, like, that is the really exciting thing because, you know, when a roofer finds out about ploy, it was like, oh, man, I've been meaning to update my website. for years, you know, I hate it. It's got the worst form. You know, it breaks half the time. You know, there's huge opportunity for me to just take that status quo with their broken

Starting point is 00:26:23 WordPress site, for example, bring it to ploy, make sure our agents are constantly monitoring it, making sure that people are filling out the forms the best way. But then also making sure the site can be personalized, the audience, right? Everyone's talking about the agentic web and building websites for agents. But we even haven't, haven't even built the best possible website for humans yet. And in my opinion, that's like personalizing the site, making sure like your copy, your content, and everything really speaks the right message. So let's do some startup questions here. Why go back to Y Combinator if you're like a third or fourth startup and give them that incredible 7% for 100K deal? Or do you like, I've heard that they'll negotiate special

Starting point is 00:27:04 deals with alumni? How does that work? There are some advantages to being a repeat Unicorn founder. There are also advantages for being a founder in your 40s as well. I think like the two things added together, you know, actually drew me back into YC even harder. So like you talk to startups and founders all the time. Like the way companies are being built nowadays is totally different than 2013, 2011. So I really wanted to just get into the thick of it. I was like, what are the most AI-pilled founders doing? what are the craziest things that they're coming up with for sales, go to market, harness engineering,

Starting point is 00:27:44 token maxing, you know, agentic development. Like all of that is happening at like warp speeds at YC. So, you know, the alternative was for me to go and kind of go figure out that. But like once you're in the batch, you know, you feel a bit of that fomo, right? And I think that's also for me, you know, being a bit older. I want to have that young gun energy. I think there's no better testosterone booster than being around a bunch of other 22-year-olds

Starting point is 00:28:17 that are working super hard that have like crazy visions of what they're trying to tackle. And I'm just like, okay, that's where I want to be. I want to just double click on something you said because we've talked to a lot of companies that do variations on what Ploy is doing, like especially in terms of ads. Like you tell us your product and upload a bunch of photos of it

Starting point is 00:28:36 and we're going to make you AI ads. And a lot of what comes out, and I'm not saying this is true of ploy. I'm saying, in general, you get like that kind of samey, it's like a girl and she's holding the cosmetics and it's a little cartoonish and she's holding it up to camera. How do you avoid that? Like, what are you doing at Ploy to work with these businesses and really to figure out how to personalize and customize what you're outputting for that person and that business? So it doesn't just look like, you know, slop. So the biggest differentiator with Ploy and everything else out there is we start with the website. And if you think about it, the website is like hopefully the prime manifestation of a company's brand.

Starting point is 00:29:19 And from there, we create a design system that can be used not just in the website, but also for ads. So for example, like if you are, I don't know, a big, you know, AI company, Base 10, Together AI, like you really, really care about how your brand motifs of these shapes and colors are showing up. You can't just throw that into an image model and expect like a perfect ad unit to be created. So this is where we go the extra mile. We create this incredible production grade design system for you. We'll understand how to use it based off of the examples that you already have. have, and then we'll create really, really clean, like locked into the brand ad sets, ad units,

Starting point is 00:30:04 and web pages and the landing pages after you click through, because that's also extremely important, to make sure all of that is super locked in and on brand. A lot of people think there is a binary choice, be a capitalist who supports growth or care about the environment and the natural world. But these two ideas don't have to be in opposition. Our friends at the Conservation Fund believe that you can protect land, wildlife, and our shared access to the great outdoors while also providing important economic opportunities. For 40 years, the Conservation Fund has been acting quickly to protect at-risk land and important wildlife habitats, while also actively supporting local communities, and this is key, creating new jobs.

Starting point is 00:30:46 The Conservation Fund already protects over 9 million acres across all 50 states. proving that we can watch out for the environment and keep our natural surroundings in balance without endless gridlock, debate and politics. And this isn't just a land trust scooping up property and sitting on it. They're working alongside permanent stewards who will find practical uses

Starting point is 00:31:07 for these important spaces without destroying them. Find out more about how the Conservation Fund is changing the playbook at conservation fund.org. SEO and agentic, commerce, what are your thoughts on how that's all going to change now? You probably have read, Amazon is trying to block people from using agents. Good luck with that. Shopify's engaging people. What does everybody having a half dozen bots doing work for them autonomously? What does that do

Starting point is 00:31:39 to the open web and web-based businesses? I actually think there's going to be several large companies that just solve this problem. So if you're thinking about it, like, I'm sure Jake out would want to invest, but essentially, like, the, the world that we're in is that it's getting extremely difficult to determine who's actually a real user or not. But ultimately, if you're a business, you actually care about both. So currently right now, based on the stats that I looked up, 50% of traffic is from agents right now, which is wild. And about 10% of, you know, two years ago, it was only 14. So it's just like the proliferation of this, there's going to probably be a moment in which 90%, 95% is all agents. So I think that's just like on the absolute

Starting point is 00:32:31 scale, but there's still going to be humans that visit the site. There's going to still be a human that wants to be able to establish a brand connection to the business that they're transacting with. So I still believe in the power of a brand. And I think like agentic web and where we're going is that businesses that don't necessarily evolve to accepting the reality of, you know, J-Cal's agent booking a resi reservation or my agent trying to reserve a soccer tennis field. I think those are going to become, I don't think there will become issues because I actually think these agents are getting so smart at browser use that they will impersonate the user. They will use residential.

Starting point is 00:33:16 potential IP proxies to get around, you know, the IP restrictions. And then ultimately, I think it's just going to look more of the same. So I actually think there's just going to be like this, this big hutspa about agentic web, but then, you know, everyone's just going to be going back to using the web. The browser, you know, seems to be, or having a, you know, in agent browser. I don't know what we're calling it these days, virtual browser. Headless browser, yeah. Headless browser. like it does change everything if it can remain logged in, right? Lon, you've been having this experience. The Grokbot experience versus the OpenClaw experience,

Starting point is 00:33:52 it couldn't be more different. Like I log Grokbot into whatever website I want one time. It's in there forever. And I could just ask it to go look things up or do things and it's totally autonomous in a way that none of the bots working out of my sort of Slack ever were. Like they figured that. Open claw realized, right?

Starting point is 00:34:12 It was like Open claw was this. credible head fake, right, Brent, Brent, Bryant Street. Sorry. I keep saying Brian, but it's all good. I mean, but you know, that's, that's our central premise too, which is like we want to put all the agents inside your website. That's something that no one else is doing. And, you know, our belief is that all this data that is in your CRM that comes to your website. All of it is incredibly additive. So, like, you know, you can just, you know, for SEO, you mentioned, you know, it's now going to be really easy for you to just write net new content. If you have ploy hooked up to your granola or Zoom recordings, you're talking to startups,

Starting point is 00:34:55 you're having in real life conversations. But that should live on your website, right? If it's valuable. So the premises is that, like, you know, you've got all your bots that help you do all these other things. but employee can help you take your in real life memories, conversations, and surface it in a digital way. That's going to raise your visibility. Like a chief of staff or a content marketing person who works at your company full time and they're like, oh, you said something intelligent. You had a great discussion here.

Starting point is 00:35:27 We should codify that into a blog post, a short. Exactly. And then from there, it can go be used for like sales decks, sales enablement and all these other use cases to really grease the wheels. How do you think about fundraising as a serial entrepreneur who's had success made a little bit of money yourself? Do you even bother raising money early on or just fund it yourself so you don't have to deal with it? Do you have like a crew that you want to reward with early equity? How do you think about fundraising at a crazy time like this and as a serial founder? Good question.

Starting point is 00:36:00 Well, I think a big part of it is I think I wanted to be just really generous with equity from the get-go. I think there's just so much hard work in the early days that, you know, early employees, I'm sort of treating as many co-founders, many of them. So, like, I want them to be extremely invested in the journey. And then on the fundraising side, I kind of skipped a pre-seed and seed and raised a $27 million round out of the gate. I think that will just give me a lot of time to play the long game. And I think that's really important because, you know, the ground is shifting beneath our feet so quickly with AI that, you know, it can help us weather any type of hiccups or, you know, huge leaps and model improvements or, you know. This is so tactical and important. If you are a serial founder and you see a market like this, which is, let's call it what it is.

Starting point is 00:37:02 It's top ticking. It's disconnected from reality often. There's a ton of FOMO. When there's sunshine and you make the hay, you put the money in the bank account because the people who don't, when there are some headwinds or platform shifts, you have time to figure it out. And you learn that as an entrepreneur. Capital as a weapon was Travis's big innovation at Uber, among many innovations.

Starting point is 00:37:26 But that was a significant advantage in beating DoorDash and Lyft and just dominating everybody. because he didn't have to think about profitability in a specific region, whether it was China or Las Vegas, he could just use that capital as a weapon. And if you invested in his company, he could invest in the competitors. So he just took everybody's money, all companies. Do you know how he got his investors to do that?

Starting point is 00:37:52 Do they have them like signed something? Because, you know, that happens all the time now, right? Where VCs are investing in this company and turn around and big check in the next? I don't want to say, but, you know, he had a system, yeah, for sure. And I don't know if, like, I should speak for him on the mechanical nature of that. But it is a sharp elbow kind of situation where, hey, if you're in us, you can't be in other people. That has left the industry, Brian. You and I came up at a time where if you invested in Lyft, you couldn't do Uber and vice versa.

Starting point is 00:38:27 If you invested in Airbnb, you weren't doing one fine stay. It was considered ungoothed. but because everybody pivots and winds up in the same four businesses, essentially, like how would any VC be able to tackle this? You just have to create a firewall. In fact, didn't, Lon, you can check here as my dutiful news reader. There was a A16Z federal investigation about them being on two boards in the same vertical. That seemed a little Fugazi, Fugazi to me.

Starting point is 00:38:59 Yeah, the DOJ program. Andrewson Horowitz over board seats. The Justice Department launched a probe regarding the firm's partners serving on the boards of competing companies. The nearly year-long investigation focused specifically on the firm's board seats at Databricks and then 5 Tran,

Starting point is 00:39:15 which combined with DBT Labs back in June. I mean, but they own, I don't know, what's your take on that, Brian? They own 5% in each, 6% and whatever. And they created a firewall internally? Does anybody care? Or was this like a competing venture firm? Oh, be honest.

Starting point is 00:39:30 the people at one of the firms, like a CFO who's disgruntled, tips off the DOJ. I don't understand the provenance of this. You tell me. What's your take? What I think the investor take is, and then I'll give you the entrepreneurs take. As an entrepreneur and founder, you know, you pour everything into your company, you know, and it is your entire life. And to see, like, your board member, for example, you know, go and, you know, go do a huge

Starting point is 00:39:57 drowned in a company that is directly competitive to you, that would rub me the wrong way, like really badly. Like I think I would, you know, figure out what I could do to sever that relationship. Honestly. But if I was investor, you're a capital allocator. Are you breaking any big rules? Are you doing your fiduciary duty? I think like this is where the line between honor and, you know, DPI kind of conflict, right? There is no honor in DPI. No, I mean, and the other thing is, as a VC firm, I know Sequoia and the top ones who get to look at every deal, they have a massive conundrum.

Starting point is 00:40:42 If you have essentially 100% vision into what's happening, and then you're limited to making one bet in a space and you bet on the number two, you know, or three, sidecar, lift, or Postmates or, you know, just eats or whatever it is versus DoorDash versus Uber Eats. Like that all of a sudden, your fund is screwed. So, and how do you know in the seed Series A space, which one's going to be the 80% market share, Uber or Airbnb? You just don't.

Starting point is 00:41:14 That's why I think Series A and Series B to me get a pass. But if you're writing big checks out of a growth fund, I don't know, J-Cal, what do you think? It's dicey. It's dicey. I think what they're saying now is like, oh, our growth thing, that's like public market investing. Don't pay attention to it. But you're right. And you know who was the master of this was Masayoshi-San at SoftBank. He came famously, this is all public. We'll find a story about it. But he came to the Uber team and said, I want this much equity. I want to pay this dollar amount per share. It was like low 30s. And he went to all of us and said, I need you to sell me your shares. And if not, I'm going to put this five or 10 billion into lift and DoorDash. But he'd already done DoorDash. So this was like a, you want to talk about a gun to the head. And I was like, sure, I'll sell Masa Yoshi San 20% of my position 10%.

Starting point is 00:42:11 You know, before the IPO, I was like, I'm waiting on this thing anyway. But a lot of us were like, yeah, we need to get Masa on board with this because nobody had ever raised a hundred billion dollar fund like he did with the Vision fund. that was some hardcore Masayoshi-San. And when they told another person who was a bigger investor than me, said, Masayoshi-San just said to him, No, Dohertyash, not competitive. They said, no, we have Uber-Aids.

Starting point is 00:42:36 We're literally going after them market by market. We're competing with them for, you know. No, no, no, no. It's not competitive. It's not competitive. He just like literally kept saying it over again. Not competitive. That was the entire business.

Starting point is 00:42:50 This is a good cautionary tale of the founders, right? because I just think like there are very few things that a founder can do. And ultimately, you just have to control what you can control and worry about just that, which is building a good product. Think about distribution and all these other externalities are just, unfortunately, you have to kind of put on a stoic sort of mindset and just kind of plays through it. I mean, if your destiny is not going to be made or broken, either way in all likelihood, you know, by these kind of on the margin transactions.

Starting point is 00:43:28 It could be, but all likelihood, probably not. And the more time you spend sweating about your competitors and not focusing on your own customers is like that your customers do not even know your competitor's name. They don't even know six other features you want. Just people are, and this goes for life too, Juan. Like I remember one time, I've told the story before on the show. I was at like a public event. I was young.

Starting point is 00:43:50 I was doing my Silicon Reporter. And I, you know, when you slip. on a step. Yeah. And you kind of, blah, blah, blah, and you catch yourself with the rail. I literally did that going down, like, you know, into this party on wall. Sure. And the whole party turns and looks at me as I'm like leaning back and caught myself right before it. And I pull myself back up. All the chatter stops. And then everybody just, when they realize I'm fine, turns around and the chatter goes back up. Yeah. And this was a seminal moment in my career when I realized as embarrassed as I was, was in that moment. That's like a 26, 27-year-old J-Gal in this like Hall of Power who just, you know,

Starting point is 00:44:29 basically flop down the steps. Nobody cares. Nobody's ever going to remember it. Now, I remember when a hundred people are looking at you like, oh my God, are you a klutz, you know? Yeah, nobody, nobody remembers anything. Nobody remembers Mahalo. They remember all in. Nobody remembers like these three or four investments. They remember Uber and Robin Hood. And to be honest, I'm starting to forget all the things in between the brahers. And, you know, that didn't work out. They're like, I met you doing this project. I'm like, open angel for him.

Starting point is 00:44:57 Oh, yeah, I did that. Yeah, for two years I had. Yeah, it's a weird. Angel investors get together before Angel list to see her start-ups. It's a weird form of self-flattery. We always imagine that other people are constantly thinking about us and mulling over what we're doing and worried about up.

Starting point is 00:45:10 Nobody's thinking about you ever. Like that's just you feel like that's ego. Like other people, the moment you're out of their field division, they're on to the next thing. Hey, let me ask you a question here, Brian. how do you think about partnerships with the frontier models versus building your own model and not giving them your intelligence?

Starting point is 00:45:30 Obviously, a big discussion in the last six months, but really in the last three, I talked to Levin Labs. I talked to Figma. I talk to Lovable. They're all like, or Harvey, they're like, we got to build our own. We got to fork, deep seek.

Starting point is 00:45:45 We've got to have our own model training because we can't give this to the frontier. Because at some point, Open AI is just going to be we have a website builder and they're just going to go scrape your website and build the website builder and copy all your innovations so how do you think about educating open-a-i and anthropic on your innovations do you use the frontier models or sparingly we use the frontier models right now simply because you know we get the best performance out of them and then also just like with where we're at in our stage of the company we only launched like two two months ago three months ago

Starting point is 00:46:19 So it's just really, hey, like our number one constraints is just our number one focus area. It's just like, hey, let's go get employee out there in the world, make sure people know about it, and hopefully they can derive value from it. So we are not concerned at the moment of what, you know, opening I can do with our data, maybe eventually. But, you know, transparently, we took the opening I deal because we figured that by taking on all these credits up front, And I don't think there's any, like, special data sharing provisions or anything like that. Tell everybody about that deal. Sam came and said, I'll give you $2 million credits at $100 million valuation or something.

Starting point is 00:46:59 No, it was an uncapped note, no MFN. Oh, okay. Yeah. So Sam came, you know, offered that in the spring, you know, every YC company this deal for like one and a half million credits. And uncapped note, no MFN. And, you know. So what's the cash value of that? Oh, two million in tokens, got it.

Starting point is 00:47:23 Correct. So he owns, if you closed that 300 million or whatever, 400 million he owns, whatever it is, 50 basis points in the company. And then you have Open AI as an investor. Yeah, exactly. How many people do you think are taking that deal? I think a good amount. I think a good amount. I think like we have, you know, a good portion of my batch took it.

Starting point is 00:47:46 And I think the, you know, as you know, J-Cal, like, you know, companies at C-C stage is like, hey, path to default alive, right? Path to growth. And, you know, I think- Oh, look, there I am in the, yeah. Be careful. Warning, one day could have a guard start his ideas. Yeah, that's what I said. Yeah, fair warning. But opening up has so much data, right? I actually don't think the data that Ploy has is necessarily, like, crazy valuable to them. You know, they have massive, data sets with really sophisticated things. I think the innovation employees how we're packaging everything. Like it plays like all batteries included, right? It's not just an AI website builder.

Starting point is 00:48:29 Yes, it is. And it's a really good one at that. And it's probably like, you know, the only one that does what we do. But, you know, it comes with a super base. It comes with visitor enrichment. It comes with workflows. It comes with, you know, you could go build a CRM on top of it. So it's just like so broad in terms of what we're doing with it that it's really the entire platform.

Starting point is 00:48:51 And I don't think like our traces are, you know, our prompts or whatever are actually that valuable. I mean, that's an honest way to look at it. You know, just I remember Zuckerberg did a similar thing. And I was like, do not. This is back in your day when you're probably first one. I was like, be careful because he is the master at, you know, he's a memmetic machine. Like he can copy something. and make it better than yours because he's fucking really good at building product.

Starting point is 00:49:20 And that one, I was like, hmm. But you're probably right. Like, they probably don't have the time to even look at this. But you've got to be careful, folks. I think that the eventuality is going to be people running local models. Hey, before I let you go, I saw you were explaining to folks on your socials exchange traded funds. If you're locked up into one stock with 99% of your math, explain to me this and then go back in a time machine. explain this to me when my Uber and Robin Hood shares got distributed.

Starting point is 00:49:49 What these are. Yeah, I'm having to talk a little bit about it. I'm assuming when you had Webflow, you did that at some point. So, I mean, these are exchange for creative funds typically for public, you know, tickers, right? So if you are an exact or a founder, you know, if you have a huge block of Robin Hood or Uber, you can essentially participate in these exchange traded funds to, essentially put a portion of your highly appreciated stock and in return you get essentially exposure to an index of other tech stocks that other people have pledged or committed to the

Starting point is 00:50:29 funds and then you essentially get a you get it you essentially take a lot of that downside risk off the table by essentially exchanging that highly appreciated concentrated position for a broad-based index. And I think you also have to get some exposure to like 10% of like a reet or something like that for real estate. You don't have to pay your cap gains at that point. So if you had a hundred million dollars in Anthropic shares, Lon, which I think that's kind of close to the deal you have, Lon. Yeah. Right there. You could take 50 million of it, put it into an exchange traded fund. They might charge you like a 1% load in fee or something. They have ways to sort of make money out. But you got to stay in it, I think, for five years.

Starting point is 00:51:12 years. But if you weren't planning on selling it anyway, it's like a very interesting thing to do. You're building one or you just know about it? No, I, Angel invested in one. My friend Shrikant is the founder of Usecash.com. There it is. Look at that. And Use Cash is the ability for, you know, hey, you've got 80 mil of Nvidia or you've got 120 million of Open AI or 40 mil of Anthropic. You can essentially, you know, pledge your shares here and then you get a bunch of broad-based exposure. Very cool. All right, man, I'll let you get back to it. Congratulations. You're hiring. Where can people go to find out more? Yeah, we're on employ. A.I. So you can just log in. You know, you can pick up what website you're coming from. We'll slurp up

Starting point is 00:52:00 your site. We'll rebuild your site. We'll migrate your site over. And then from there, we can turn it into a growth machine. So we'll help you find customers. We'll help you improve your SEO and AEO. will help you deliver revenue, leads, and it's always on, right? So it's constantly experimenting with your site. It's helping you run your ads, helping you optimize your ads. And yeah, it's been really exciting. We've got over a thousand customers already that have been starting to pay us. We only launched a couple months ago.

Starting point is 00:52:33 And, you know, we've got all these expert ploybooks that are built in. So, you know, folks like myself and other growth practitioners have kind of really imbued in the platform, all these really hard-learned tactics about how to optimize your website, but then also optimize your entire top of funnel. So that ploy can be the sort of last piece to go-to-market software out there.

Starting point is 00:52:58 Yeah. So if you're an enterprise, startup, anything in between agencies, go check it out. ploy. We'll drop you off. Brian, thank you for taking the time

Starting point is 00:53:06 to share your experiences here on the street in startups. Great to see you. Thanks, guys. Our offices used to be on Bryant Street in San Francisco. Yes, that's right. I held on to my loft. I got like a $3,000 square foot loft. You know what happened?

Starting point is 00:53:19 It was like, they were like, you're going to lose money on this thing. I paid like $600 a square foot. Now they're like, uh, you're going to make a fortune on this thing is worth a thousand dollars a square foot now. And I'm like, wow, four years and like this whole thing makes a huge difference. San Francisco's back, baby. That's what they're saying. Bay Area real estate.

Starting point is 00:53:36 Very strange. All right, let's set. We're going to go off duty. We got another tech story. We got a ton more news. if you want to do more news. Give me like options, and I'll pick one or two. All right, we could talk about Microsoft reconsidering co-pilot.

Starting point is 00:53:48 We could talk about Jev, TypeSafs in talks for a big valuation. We could talk about Oracle's troubled data center plans. We could talk about Optimus. And then there's Ando there, the startup that's doing AI Native Slack. I want to hear about that last one. Tell us and show us. Ando. I mean, Sacks was working on something, get glue or something.

Starting point is 00:54:06 Yeah, glue was, that was Sacks's slack killer. So yesterday, Ando came out of stealth with $20 million from Excel, index, and emergence. They're taking on Slack. It's a team chat app where AI agents get their own identities, inboxes, and permissions, just like they're human coworkers. So agents could join channels on their own. They could speak up without being tagged. They can message people directly when something needs attention. Founder Sarah Doe calls humans relaying agent work to their teams as meet proxies and says Ando is built to eliminate all that.

Starting point is 00:54:39 This morning, Microsoft gave its own autopilot agents identities and email addresses inside Teams and Outlook. So this is obviously an idea that is out there in the world. My question for you, Jason, would be, how does a startup, I mean, if you were Ando and you're taking on a giant like Slack, I mean, what is the approach? How do you even get started at picking away at that kind of advantage? How do you make it free for the first 200 users? and you just take the startup market and then you grow with them.

Starting point is 00:55:12 And when you hit the first 200 accounts are free for life and then you charge after that, that would be a quite disruptive way to do it. Or you charge based on AI usage or you come up with a flat rate. I mean, there's all kinds of ways to be disruptive in the space. But having a better product is critically important, obviously, first and foremost. So the challenges I've had with Slack is, you know, we have to keep throwing, we have to keep throwing agents into it

Starting point is 00:55:41 and getting them to work. And I had to, like, reach out to my guy, Mark Benioff, and I was like, I need the enterprise version, and it's like 50 bucks a person per month. And I want to be able to see everything, including DMs and possible group DM chats. You know, so what happens is you and I are working on a newsletter product.

Starting point is 00:56:02 There's channels. There's hidden, there's private channels. And then there's the other half of Slack, which is DMs. Yeah. And then like group DMs. So you got four people who decide, I'm not going to create a channel for this. I'm just going to talk about this project. And that's the most important thing.

Starting point is 00:56:18 Right. And then it's blinded. Your AI Oracle, as I call it, like is blinded to that. So, you know, in a major organization, finance, et cetera, health care, any, any corporation, they have access to everything. So they tell you, like, hey, don't send like memes over your DMs. This is not the place to talk smack about your boss or to ask somebody on a date. And there was like a false sense of, or there was a false sense of like, oh, this is my data in my email.

Starting point is 00:56:46 When I'm emailing, you know, my friends about a party invite from my corporate email. So you have to educate young people on this. But if you go work at Sony or JP Morgan, whatever it is, they explain this. Like, do not use your laptop. So I had to like tell everybody in our firm, like, hey, we're moving to this vision. you should never, ever do anything like this. And we moved the slack over, we exported everything, and we kind of gave everybody a mulligan.

Starting point is 00:57:10 And when we moved over to like, hey, we're monitoring every computer, we know what is being done on computers. So if you try to download the entire database of our LPs or the documents from like Uber's series seed or something, like it's going to have an alarm go off. And the alarm goes to compliance. And I get a, they come to me. hey, this person just, you know, was like doing this thing. So that is the key with these systems is

Starting point is 00:57:36 you can't make decisions if you don't have access to everything. And Slack is like a roach motel at times. And I explained this to their team. And they were like, hey, we're going to hook you up. So they hooked me up. They gave me like the highest level at the price I was paying. But I said, you got to think first principles here. Somebody else is going to create something where you have access to everything. And here we are. There's going to be a bunch of people who are doing this. Sure. I mean, yeah, like there's so one of the most crowded spaces we talk about on this week in AI at this point is like enterprise agents that help your team organize. I mean, it's like everybody is crowding into that area right now. I mean, it is everything. Did they mention their

Starting point is 00:58:23 pricing or not? That's what I'm most curious. Did they mention their pricing? That would be an interesting thing to know. I will see if I can. Because that is pricing is where it's disruptive, you know, access to data is where it's disruptive. This is not something unique to Salesforce or anyone else, but they do try to make it hard for you to get your information out. Yes. So that they have you forever. I'm being told that they are not yet publishing their pricing. Their public site is still a waitlist onboarding rather than self-serve plans. There's no per seat. We don't know yet. It's just a wait list for now. Free is the way to go. getting influencers is the way to go.

Starting point is 00:59:01 So I would go right after startups and VC fund firms. And that would be the greatest way to do this. And then, man, I would be very interested in a version of this where I would be very much in a version of this where I could build my own AI, local model, server model, open source, and not put the entirety of my data into Claude, into OpenAI. So we talked about this over and over again. So that's the other vector, I think they should be doing. Yeah, I mean, it is, I know that it is agent agnostic.

Starting point is 00:59:37 So the idea is that it's an ecosystem, and this is even on their site. We don't expect to build every agent your company uses. And there should be the place where the best agents, ours, yours, or anyone else's, can work alongside your team. So you could pull your codex in, your Grockbot in, your muse in, or what have you, and they can all collaborate in Ando is the idea. I looked at doing this as a startup myself, like doing an open source project. There is an open source, yeah, there is an open source version.

Starting point is 01:00:12 Let's see, open source for matter. Matter most? Matter most, yeah. But matter most still charges a lot. So it's open source, but there's like a hosted version. So it never took off. I think they should just go for it here and make it free and then figure out later at two,

Starting point is 01:00:38 I mean, or maybe make it like $500 a year for, literally make it $500 a year for up to 100 people and $5,000 a year for up to $1,000 and just make it super easy. Just undercut everything. Yeah, three tier is super easy. The product is presently aimed most directly at teams of up to about 30 human users.

Starting point is 01:00:56 plus AI agents and coworkers. So above that, you could say 30 or below free. And then once you start moving above that, maybe that's where the tears kick it. Looks like I have to invite you. I'm signing up for this. Oh, let's try. Five team members.

Starting point is 01:01:11 Oh, my God. What a pain in the neck. But I have been looking to experiment with something like this because its time has come. Okay, what's going on with Optimus? Tell me about that. This morning, the information reported Tesla now builds several 100 optimist robots a week, roughly 10x, their Q2 pace.

Starting point is 01:01:31 Managers reportedly want an automated line, making over 1,000 robots a week by year end, and eventually getting that up to 20,000. The hands on the robots are the bottleneck, over 100 screws and small parts per hand and forearm still assembled by hand. Tesla ended the Model S and Model X to free up this line and plans to lease early robots rather than seldom. I remember Elon talking about the hands last year at All In Summit. and how complicated it was to do the hands.

Starting point is 01:01:58 I think he's going to have the best-selling product of all time. I think self-driving is, because it's taken so long, self-driving, you know, not just by him, but by everybody, because, you know, you have a serious safety issue. I think, like, there's going to be a lot of competitors. It's not going to be one person wins at all. I think he wins 40%. I think Uber plus 20 partners wins, like 30, 35%. I think WAMO, if they go it alone and are not part of the Uber network, I think they win 10, 20%.

Starting point is 01:02:26 They skim the most profitable cities and have a very nice business. But with this, I could see them having much, much larger market share. And I think people will pay five bucks an hour for one of these robots. So if it works 20 hours a day, 100 bucks a day, $36,000 a year. If the thing lasts five years, you're talking about $150,000, and it would probably cost 20, 30, 40K to start and then go down from there. So you just think about how profitable this thing could be. It's going to be nuts.

Starting point is 01:02:59 What are you envisioning? Who are you envisioning is like the ideal customer for that? Like I own a warehouse. I own or is it like around the ranch? Target. Yeah. I mean Target. Amazon.

Starting point is 01:03:12 Tesla factory, BMW factory. Anything that's dirty and hard and painful, anything where a human is going to get a, you know, reoccurring injury because of their elbow, carpal tunnel injury, anything that makes you want to kill yourself when you do a double shift, you know, security. So the dock workers is that's who we're going at with this thing. I mean, if you look at the docks in China, they are fully automated in Japan.

Starting point is 01:03:40 They're moving to full automation. They have a huge robotics thing because they don't have anybody to work at them in Japan. They're in population decline. In China, they do have a lot of people to work at them, but they want to have them automated because they need to get stuff out of there faster. And so docs would be really good to automate. If you pull up like a YouTube video, Jacob, of like automated ports in China. When you see some of these automated ports, it is next level.

Starting point is 01:04:09 They are taking containers off of ships, putting them onto autonomous sleds. It's bonkers. There you go. It's super bonkers. Oh, we got in there we go. Yeah, so I don't know. This is China, yeah. I just was thinking of like, those of the sleds.

Starting point is 01:04:23 You see the sleds? Yeah. When I was a kid, like the old guys who were always like bent over and bad backs and knees, they were always like the dock, the dock workers. That was the hardest job when I was a kid, I thought. And you think about what they're building out in China, you know, they have to get so much stuff off that island 24 hours a day that I just don't think humans can compete with the demand they have.

Starting point is 01:04:45 So humans will be supervising human in the loop. Obviously, still got to supervise the stuff. But, man, they are, they make our ports look ridiculous. All right, let's go off duty. What do we got? In an appearance this week on the Ezra Klein show, I've got to hear the doc, a Jensen Huang of Nvidia. He said it doesn't matter if kids can't do basic math anymore because maybe those skills don't matter as much in the future. And then he told Ezra Klein that he doesn't know his own address.

Starting point is 01:05:12 Can we roll this clip? Try to get a kid to do long division right now. You know, the multiplication table is starting to be forgotten. Doing square roots. My goodness. I mean, it's just basic math is being forgotten. Doesn't matter? That's my question for you.

Starting point is 01:05:29 Yeah, I don't think it does. I don't think it does. But there must be some set of skills that matter. Oh, yeah, yeah, yeah. But maybe not those. We're going to discover new ones. Just maybe not those. I actually don't know my address.

Starting point is 01:05:45 And I, and every, I don't really believe that. be true. It's completely true. And Janine will tell you, and Lori will tell you. One day, I had to pump gas. I had a few years ago. And they needed my zip code. And I panic. I didn't know my zip code. I had this exact thing. I don't know my telephone number. But I had that happen. I forget these things. Okay. I can live with it. I feel like I'm a little with Ezra Klein. You really don't know your address. You really don't know your address. I know my address, but I literally, he may also be suffering from, you know, having four homes or three homes, but I have had times where I forgot my ski house address and I have to look it up. Or certainly,

Starting point is 01:06:28 if you ask me what my zip code was at the ski house, I don't know it off the top of my head. I do know at the ranch now. I don't know at my office. And I have forgotten my phone number. So I think what's happening is certain things are atrophying because you don't have hand speed anymore. The act of dialing a phone number reinforces it every time. But if you have everything, like, I forgot my wife's phone number the other day, I had to look it up. And I'm like, gosh, yeah, we had the same phone numbers. Our whole relationship over 20 years. And we, I'm forgetting both of our phone numbers. I would look at this and the math thing as calligraphy, as handwriting, you know, cursive. Right. Yeah. So are we, is it necessary for us to learn handwriting and cursive?

Starting point is 01:07:15 Is that going to be a skill that's critically important for my kids? I don't know. I still think we should teach it to them. I still think we should still teach them the basic math things. I think Jensen is struggling with that and doing a thought-provoking question. I don't think he's saying get rid of all math, get rid of all, or I'm not saying get rid of all penmanship. I like the idea of people having great penmanship. And I can tell you, young kids do not know math.

Starting point is 01:07:39 They do not know penmanship. That's why I'm in love with the Montessori educational system, which really reinforces a lot. of this like hands-on stuff and less screens. And you can always learn the screen stuff. So I think parents should do Montessori type things where the kids are interacting with each other, doing projects, physical stuff. Get that foundation. And then, you know, for the last hour of the day, you can go online and do stuff.

Starting point is 01:08:02 But, you know, they do, I mentioned this study, I think, on All In. There was a study of brain scans for people who used AI to write their paper, people who used search engines. and were cutting and pasting, and then people who were, you know, raw dogging it with a pen and paper or typing it, like, you know, where processor and brain activity was just like zero,

Starting point is 01:08:27 like 5%, 50%, and then like boom. And here it is, I guess this is the study here. And there was actually pictures of the brain scans in this thing. So if you scroll down, you might even be able to see that. There you go. And so they were looking at these,

Starting point is 01:08:44 and you see the LLL. you see search, you see the brain activity, dynamic direct transfer function, EEG analysis of alpha band for groups. I think that's the technology. LLM search engine brain only, including P value show significance from moderately significant to highly significant. In other words, your brain is firing on all cylinders. And we're just going to atrophy. By the way, we already have an analogy for this. You sit at a table at work, click clacky, click clacky, your vitamin D goes to shit as we talked about. You get fat, you get a big tire around your waist, and you feel depressed.

Starting point is 01:09:23 Why? Because you need to go out and run and get, and then if you just work and then go home and Netflix and chill, now you're in double situation because you didn't go out, you didn't have friends, you didn't laugh, you didn't socialize. This is why so much of the country is on SSRIs and trying to solve for anxiety, depression, sadness, melancholy, whatever it happens to be. you want to describe it, it's like, they're like, oh, I got to take a pill because I have a chemical imbalance. It's like, chemical imbalance is probably because you're not getting enough sun. You're not

Starting point is 01:09:52 sleeping. You're not doing socialization. You're not meditating. And you're not doing physical activity. And I guarantee you, I work on this with my kids. I work on it with myself. I say, if I'm feeling out of sorts, how did I sleep last night? How's my diet? Has my exercise. How's my socialization? And has my meditation. The fifth one maybe just might be me virtual signaling or whatever, but I've had a good, you know, experience with meditation. Just I can tell you if you're listening to me, you feel down and out and you feel melancholy, just please do all five things I've said for three days in a row. Yeah.

Starting point is 01:10:24 Meet with your friends for dinner or lunch. Go do a physical activity with your friends. Go play pickleball, Padell, hike, anything in the sun. Eat a salad, some protein, no fried food, no sugar. Man, it's going to turn around in 72 hours. Yeah. I mean, I have struggled with depression, and I got off all the SSRIs or whatever. But that it is, like when you're, when you're inside and you're feeling blue and you're down and you're depressed, the last thing you want to do is go out for like a walk in the sun.

Starting point is 01:10:54 You're just like, oh, it's not going to make me feel that I don't want to. That's the thing. It's resistance. You're going to reject it at first, but then you go outside. You just see, just seeing visually other people being outside, fresh air. I could tell the difference in you since you got your. your new dog. You get a new dog. You have no choice but to do two walks a day. And somebody's going to come over and be like, oh, my God, your dog's so cute. And then you're going to talk to a human

Starting point is 01:11:18 being. That's why I brought everybody back to the office, not because I'm like, I want to be a tyrant and have everybody chained to a desk. Because I was getting weird. Everybody's getting weird by not going out. And I'm literally like moving to get closer to our office and moving our office closer to me. Because I want to be, you know, I'm 45 minutes from the office, an hour. It's a little bit when you have a ranch, you're pretty far away. I want to be able to just walk to the office. That's my ultimate goal, walk for the office. Or ride a bicycle.

Starting point is 01:11:48 We're going to have to be at the salt lick. There's not as many options out there. Well, no, no, no. I'm going to move closer to the city. I guess in the city, I think, and then keep the ranch for the weekends. So, Sidney, of course, she's in that sports gambling. I believe it's for a sports gambling app where she's almost nude. You know, it's very provocative.

Starting point is 01:12:07 She's covering herself with the sports memorabilia. or whatever. And a lot of women athletes got a little upset where they were like, listen, she's sexualizing being a woman athlete and we don't like that. And we're just like men competing and stop making it sexual. So it's touched off a whole other round of people condemning Sidney, and people being back and forth about what does her body mean and what side of the political wars you on. And then we saw that you had tweeted your new take on what Nike's new campaign should be

Starting point is 01:12:38 featuring this Sweeney. Yeah, so we had had this funny conversation about Nike. Nike is like got kicked out of the... Right, the Fortune 500, yeah. No, 100. Oh, the 100. And Palo Alto networks went in. And I was like, you know, they need to do something and shake things up.

Starting point is 01:12:56 And, you know, you got this, just do it. I think you got to like go a little counter programming. You got to get somebody who's youthful. And you got to kind of confront people and shake things. up, right? Kind of like we were just talking about with, you know, if you're feeling melancholy in this generation, that's terminally online, you've got to fight to go outside and do something. Like, challenge that group because I believe part of the issue with Nike is, and also with restaurants, also with spirits companies, is young people do not go out. They stay home.

Starting point is 01:13:27 They believe that they've had a play date when they're young or they believe they've talked to their friends when they do it digitally. So my daughters, I'm like, hey, let's do something with your friends. And they're like, oh, you know, I had a play date with him yesterday. I'm like, oh, they were here or you were there. And they're like, no, no, we played, you know, Minecraft for two hours. And I'm like, oh, you're on your iPads. That doesn't count. I want you guys in the swimming pool on the ranch, going to a hike, something out in the world. So on All In, I said, like, you know, come up with something like really to challenge people, do it or don't. We don't give an F. And they were all mocking me, et cetera. But I was like, wait a second, I'm a very good brander.

Starting point is 01:14:07 I think this branding, I have never heard anybody come up with a better slogan for Nike that was challenging to the audience. And then I was like, well, you know, let me see if I can create some ads. So I saw the Sydney Sweeney thing. And I was like, well, what if you put, I first started by using Grock to put these ads like on trailers or like, you know, on subways or whatever. And then I was like, put Sydney Sweeney in here. And like some tools won't do that. But Grock was like, yes, sir. Look at this. Do it or don't.

Starting point is 01:14:39 We don't give a fuck. And then if you just put a star in there for the you and like, yeah, just do it, whatever, Nike. And, you know, she's looking at you like, hey, I'm doing it. You do it. I don't care if you do it or not. I'm doing it, right? Show the next one. It made like a video of this with a boss is pulling out.

Starting point is 01:14:57 What it would look like in New York on the streets of New York. You and I could run an ad agency and get paid a million dollars per campaign because I'm good at this. Oh, look, and then they made her like into a video or whatever. She's moving like Harry Potter, yeah. I mean, AI is sick right now. Do it or don't? We don't give an F. So anyway, rate my, because you're very objective here.

Starting point is 01:15:19 Yeah, of course. What do you think? Is it too spicy? No, I don't. I think spicy, you have to be spicy in 2026. It's the only way to break through. I mean, I feel like what we're seeing with Nike is, it's less anything to do with, you know,

Starting point is 01:15:33 people were always trying to diagnose like, oh, Nike did it. a brand deal with this person or they said this or is it. I don't think it has anything to do with anything specific they've done. I think Nike used to be part of this monoculture that doesn't exist anymore. Like when I was in high school, you were either like you were into Nike or you were into Adidas. And that was like it. Maybe there were like a couple Reebok kids or somebody had like a British knight's shoe or whatever. But there weren't there weren't that many options. You sort of had like two or three acceptable options and you picked one. That just doesn't exist anymore. Now

Starting point is 01:16:04 there's like so many. Everybody's in their own little niche. Everybody has their own style. I don't know what you do if you're Nike to cut through that. But I do think this would grab people's attention, especially the Sydney. Like you picked the exact right celebrity for this ad because she is, everybody's really fixated on her. Everybody has a take on her. And that's what I think people miss is that a divisive figure is like better.

Starting point is 01:16:29 Like somebody everybody loves is great, but somebody who everybody has an opinion on good or bad. For marketing like this, I think, is almost better. Like, they're going to grab everybody's. It's like putting a Kardashian in your ad. Like, they have lots of fans, but they also have a lot of people that are going to see that and get like a little riled up. And they're going to pay extra attention because they don't like the Kardashians. And I think you've got to, you have to play in that game now in 2026. By the way, and just everybody giving her a hard time.

Starting point is 01:16:56 I'm Greek. Gymnos GYM-N-O-S. This is the ancient Greek adjective, meaning naked, nude, unclothed. It is the root of. the word gymnasium or gym in the Olympics. Because they were, yeah. They were naked in the Olympics or lightly dressed. Same thing with gladiators.

Starting point is 01:17:14 And they were celebrated for the ideal beauty of their bodies. They were essentially sexualized if you wanna use that. Or if you don't wanna use sexualization, they were an ideal. Right. And that was, and it conveyed that you were disciplined and that you were ready to serve in the military.

Starting point is 01:17:39 So giving, like, her a hard time when, I don't know who gave her a hard time, but am I crazy? Or does Sports Illustrated do a nude issue? A lot. This came up a lot. That Sports Illustrated, they do the swimsuit issue every year, and then they do the body issue, which is athletes, male and female, naked. That's the whole idea is like they shoot, come up with clever ways to obscure their naughty bits, but, you know, they shoot the athletes without their clothes on.

Starting point is 01:18:04 So a lot of people were saying that, like, well, it's already pretty sexualized. And, like, look, I get when it's like, you know, 18-year-old gymnasts and dudes are being creepy about, like, their little outfits or whatever, like, I get that that's creepy and wrong. But sure. Yeah, the Sidney Sweetie ad, you know, just like her holding footballs in front of her chest and be like, go bet on this app. Like, how can you be bothered by? It sounds to me like what they're upset about is that she's not an athlete and she did it and took the bag and secured. I think it's the whole, I mean, I'm sure that is part of it. But it is also like, you're jealous that she's getting 10.

Starting point is 01:18:40 How much did she get in this company? Oh, that's a good. Is it known what her deal is? I don't know what her deal is. Because I think she could like, honestly, she's such a force of nature right now in terms of like, a zeitist that if I was running Nike, I would literally go to her and say, I'll give you 1% of the equity in this company for a lifetime deal. It was, I'll also give you $10 million a year in cash, but I'll give you 1% ownership.

Starting point is 01:19:02 you vest it over the next 40 years of your life, like every year. I believe she is an equity partner. They're not disclosing the deal, but it does sound like she's got some, she's getting a little taste of Novig this platform, it sounds like. This is what celebrities should be doing. If you're a celebrity who can drive sports prediction market startup, Novig is an equity holder and strategic partner. That's what they're calling.

Starting point is 01:19:32 If you have a massive following, like Kim Kardashian, like her, you should be an equity owner. You should create your own brands or take existing brands that you don't have an interest in starting and just say, hey, I'll do this for 10 years. I want 10 percent every year I do one. But this is completely hypocritical. Serena Williams, Venus Williams, Dwight Howard, Chris Paul, Carl Anthony Towns. Everybody has done this ESPN thing. I don't understand the controversy, but okay, sure, whatever. I mean, it was okay for the Greeks.

Starting point is 01:20:07 I mean, we're the greatest culture in the history of humanity. Yeah, I mean, I guess I could sort of see the complaints, but it's also you have to know at this point, if you are going to make a complaint about a Sidney-Sweeney ad, you're just feeding the beast. Like, you're just drawing more attention to it. She's now making a lot of money off of this, like being the controversial, lady in ads is like

Starting point is 01:20:32 obviously paying off hugely for her. So if you're not a fan of it, it's Streisand effect at this point. Just stop talking about it. The more obsessed people become with her the more she's going to make that money.

Starting point is 01:20:48 Here, pull that quote back up, Jacob. You just had a good quote I want to read. Pull that article back up. Yeah, British sprinter Amy Hunt was one of the people who complained. Sidney Sweeney, this is what women in sports look like. Muscles, hard work,

Starting point is 01:21:03 blood, sweat, and tears. That's, that's okay. So she's saying, yeah, it's like a cheesecake version

Starting point is 01:21:10 of women in sports. I'm surprised you didn't bring up the trailer for the new Tarantino. Oh, it's actually David Fincher. Tarantino wrote it.

Starting point is 01:21:21 David Fincher director. Cliff Booth. What do you think? Looks good. Looks great. I can't wait. I mean, I love that character.

Starting point is 01:21:29 I love that world that Tarantino created. I was super excited he wanted to get back. You know, it seems like Tarantino, he does these movies, and then he kind of gets tired of them, and he's right on to the next thing, and he's kind of like losing interest. So the fact that he wanted to come back

Starting point is 01:21:43 and write another script about this character set in this world, it's in late 70s Hollywood, which is going to be a very different time and a very different vibe from the original movie. And, you know, Brad Pitt and David Fincher, what a duo, seven, Fight Club, Benjamin Button, I mean, they've made some pretty strong collaboration.

Starting point is 01:22:04 So, yeah, the further adventures of Cliff Booth. I wanted you to play the trailer with the sound off guys. Yeah, let's say we could take a little. I mean, it looks gorgeous. I mean, it looks pretty. Seeing Fincher bring to Hollywood in the 1970s to like, that's the new Beverly, by the way, my old stomping grounds, Tarantino owns. Which Tarantino owns.

Starting point is 01:22:22 And I guess he's now, I guess they're saying Cliff Booth is now a Hollywood fixer 10 years later. Right. He comes back. He comes back to Hollywood. he's no longer working for Rick Dalton because he says he, Dalton got too fat for him to stuntman for him anymore. What a dig to,

Starting point is 01:22:36 that's a little, to Leo, a little DiCaprio digging there. Although there is still a lot of rumors that DeCaprio showed up for a day and that we may get a little like cameo, Rick Dalton appearance. I mean,

Starting point is 01:22:49 this thing, he should be building the Tarantino universe as his like legacy. Just take all 10 movies. And like, let's get a Steve Bouchemmy 20 years after Reservoir Dog story. I don't know if Steve Bouchemy died. Yeah, Steve Bouchemy does sadly.

Starting point is 01:23:06 But there is, the early ones all had a lot of connective tissue. Like they mentioned the Vega, like Vincent Vega is obviously Trevolta in Pulp Fiction, but there's a Vega brother mentioned in. I think Vic Vega is in Reservoir Dogs. That's Michael Madsen. And then there were red apple cigarettes that used to show up in all the movies. And so he used to kind of do that. were like they were all happening in the same alternate reality.

Starting point is 01:23:30 He kind of seemed to get tired of that over a while. But I agree. He should do a movie that pulls together. This is my dream interview for, this is my dream interview for the All In Summit. I put some feelers out with my own work. Yeah. I really want to interview QT.

Starting point is 01:23:43 I think he's just tremendous. He's living in Israel now. That might make it a little. I think half the time or something. Yeah, it might be a little tough. I mean, raising his family, good for him.

Starting point is 01:23:51 I think that's going to be great. I love this because it's got a 1970s Mustang fastback. which was my first car. I had a Mustang Grande in mustard gold. You can pull it up, 1970, Mustang gold, yeah, Grande. If you look at that 1970s Mustang Grande, 1970s Mustang Grande, I bought it for $600, and it was in mustard gold. And what made it really interesting as a Grande was like the fancy one.

Starting point is 01:24:25 That's a vinyl roof. So if you click on the black one on the side, I had a black vinyl roof on this thing. 351 Cleveland, and that's the brown one. You find a black one in there at some point. There it is. That was, no, that's like a, not a 19-7. But anyway, the 1970 is like a longer front.

Starting point is 01:24:41 It is crazy, like this thing when you would hit the gas. It's the one, fourth, third one down on the right. Yeah, there you go. That's the one. Look at this thing. Wow. It is like a shark, and it is a beast. It's literally two-thirds front engine.

Starting point is 01:24:58 357 Cleveland, 351 Cleveland, I think was the designation. And I put a four-barrow carburetor got literally five miles to the gallon. This thing roared and I almost died in. I was coming on the belt park where I came around a turn. I hear a crack like a crack and a grinding. I pull off the belt park right in Brooklyn. I pull into a spot. I look under it and the suspension bars and the axle for the front thing is cracked.

Starting point is 01:25:24 right as I park and I look at it, it goes, bang, and it just breaks in half. Oh, man. The thing is on the ground. I walk home. I come back two days later. It's been stripped. Yeah. There was nothing left of it.

Starting point is 01:25:37 I mean, Brooklyn. In the 70s, yeah. Man, I wish I had, like, on the ranch, like a, like a barn dough so I could start collecting these again. All right. Another amazing episode of This Week in Startups, February 25th. We'll see you next time. Bye, bye, bye.

Starting point is 01:25:52 Bye, buddy.

This Week in Startups - A rogue OpenAI agent hacked Australia’s government. Does this matter? | E2342

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.