This Week in Startups - Will OpenAI Tank OpenClaw? | E2251

Episode Date: February 17, 2026

Will OpenAI Tank OpenClaw? | E2251This Week In Startups is made possible by:Northwest Registered Agent - https://www.northwestregisteredagent.com/twistLemon IO - https://lemon.io/twistLinkedIn Jobs - http://linkedin.com/HiringProOfferToday’s show:*OpenAI hired OpenClaw creater Peter Steinberger. What does this mean for the future of the AI virtual assistant platform, and what can the OpenClaw community do TODAY to help protect their favorite free, open-source resource?Jason and Alex consider the future of OpenClaw alongside guest experts and founders Hiten Shah and Jesse Genet. Plus we’re taking a look at all of their OpenClaw creations. Check out demos of Hiten’s “personal CRM” for busy investors and Jesse’s family media aggregator.PLUS we’ve got “AI Scott Adams” creator John Arrow to talk about why he was inspired to create an AI clone of the iconic podcaster and “Dilbert” creator, and why he thinks it has so many internet commenters up in arms.Hiten Shahhttps://x.com/hnshahhttps://crazyegg.comJesse Genethttps://x.com/jessegenethttps://Lumi.comJohn Arrowhttps://x.com/johnarrowhttps://www.aiscottadams.com/Timestamps:(0:00) Introducing our guests Hiten Shah and Jesse Genet!(2:27) The panel’s biggest concerns about OpenAI hiring OpenClaw creator Peter Steinberger(6:03) Jason gives us his most optimistic and pessimistic OpenAI takes(7:58) Why Jason thinks OpenAI will give everyone their own closed-source assistant(10:45) Jesse’s Mac Minis now outnumber her children!(15:28) What the OpenClaw community can do right now(17:08)  Can companies “hijack” OpenClaw via hosting and skills?(20:12)  Hiten live-trains an OpenClaw skill based on Jason’s book “Angel”(22:37)  How much is everyone spending on tokens anyway?(25:56)  How Jesse vibe-coded an app to aggregate non-Slop videos for her family(34:19)  Why Jason thinks Jesse’s app is a great potential business(37:27)  Hiten built the “personal CRM” busy people have always dreamed of having(44:12)  “We’re in an appless world.”(45:19) Why markdown files (.md) are perfect for humans and their AI agents(52:36)  So why are founders so obsessed with OpenClaw?(58:47)  John Arrow, the creator of AI Scott Adams, joins the show(1:02:44) How does AI Scott Adams get more like Scott Adams over time?(1:04:37)  The legal and ethical considerations around posthumous AI Clones

I am a crazy risk taker. I think you got a business here. I will give you $125K just to pursue your muse for the next year. If it doesn't work out and you burn the $125K, I don't care. I like to take risk. I can already afford my own Mac Studio, Jason, okay? I mean, are we doing this or not? This week in startups is brought to you by LinkedIn Jobs.

Higher right the first time. Post your job and get $100 off towards your job post at LinkedIn. com slash twist. That's LinkedIn.com slash twist. Terms and conditions apply. Lemon.io. Building a great team is essential to any business.

Lemon is a marketplace of vetted, experienced engineers ready to take your company to the next level. Get 15% off your first four weeks of developer time at lemon.io slash twist. Northwest registered agent. Get more when you start your business with Northwest. In 10 clicks and 10 minutes, you can form your company and walk away.

with a real business identity. Learn more at Northwest Registeredagent.com slash twist. All right, everybody, Monday, February 16th, 2026, lots of news. He's at Alex. I'm at Jason. This is Twist. We have a breaking news story that drops Sunday, open claw, getting bought by the evil empire. Open AI, Sam Altman has bought OpenClaw. Here we are in AD, 24, or AO 24, 24 days after we started talking about the incredible open source project, OpenClaw, and it's already been purchased. We're going to talk all about what that means and handicap it. Alex, you decided to bring some great guests on today for our Monday show of Twist.

Who do we got? Yes, we got a couple of awesome guests. First up, we have Heaton Shaw, someone that I've known in the technology world for a very long time. He's the CEO of Crazy Egg, prior co-founder and CEO of a number of other companies. company's fantastic technologist and open claw fanatic Heaton. Thank you for coming on the show. And we also have Jesse Jinnay. She is a former founder and currently a homeschooling mom.

And she's using OpenClaw in and around how she teaches her kids about the world. So two different use cases, two different awesome people. But OpenClaught, Jason, as you know, has a lot of uses. So I figured let's bring them both on and learn from each of them. All right. And Heaton, my God, I've known him for a long time. He's an OG now. He's an OG now.

he's got the gray hair, but he was a kid in the Web 2 era. And Jesse, I saw you on Twitter talking all about your homeschooling, which I'm a fan of, and how you're doing incredible things with OpenClawe. Before we get to each of your individual uses of OpenClawe, this incredible new platform, we have to talk about what some people feel is a rugpole, what some people feel is the end of the project. What other people feel is going to be just fine, and they're extremely happy for the founder of OpenClaw to secure a bag in under 60 days.

This is going to wind up being the largest, if it is in fact a unicorn, which I'm sure with the stock options, it will be. Alex, this would be the Solopreneur billion-dollar company that we've been talking about. Yeah, kind of. So this is the thing that I'm really curious about. And I wanted to talk about it with the group

because the numbers aren't super clear. The way they've phrased this is, Peter's going to, Peter, the guy behind OpenClaude originally, is going to join OpenAI, and they're going to put OpenClae. Claught into a foundation that Dave Moran is involved with and so forth. But Jason, what was not discussed at any point was an enormous pile of money going from Sam Altman's pocket to Peter's pocket, at least explicitly.

So is he being brought on to OpenAI with a fat sack of stock options as an equivalent to an acquisition? Is that how you see this working out? My guess, I am strictly guessing here, is that they gave him a pile of cash as a signing bonus. Ah. I would estimate that would be a nine. five-figure, kind of mid-nine figure deal. I would put it at 250, maybe even as high as

$500 million in some combination of cash and stock in Open AI. Then he was probably given another equal amount to be a team member of Open AI for the next four years. I would then put the deal at $250 plus $500, $500, $500, something in that zone, because he could have easily raised $50 million at a $300, $400, $500, $500 million valuation, perhaps even a billion dollar evaluation from investors. So what opening I had to beat was $100 million from Sequoia, Andresen, G.C, any number of firms, would have given him $100 million plus maybe bought $100 million of his shares at a, let's call it, a billion dollar post, a $2 billion post. Because this is such a phenomenon. It is the highest ramping ever, GitHub project. Therefore, you have to beat market. I'm going to guess what Peter did,

and he will tell us over time, and he's interacted with me a bunch over the weekend and in my threads, because I had told him and I had DM'd him, like, don't do it, don't do it, don't do it. You're the chosen one. This is the chosen project. Do not sell to Sam Altman. Don't do it. Look at Sam Altman's history. Look at what he did to Elon. Look at his co-founder's leaving. look at him almost getting fired. Just look at that track record. Look at the YC track record. And this is nothing against Sam Altman.

We're friendly, and we've known each other for over 20 years. But he's a cutthroat guy. He's known as a dealmaker. Doesn't surprise me that he got this deal because he's from the Zuckerberg School. I've said this over and over again. He is as sharp as a blade gets. He is the greatest dealmaker of this generation, I believe, or will become.

All right. So we have two founders with us. We have two founders with us. I'm curious. First of all, first reactions to the deal. And do you guys agree with Jason's framing about the financials here? Let's do Heaton and then Jesse. Something had to happen.

If you look at all the open source sort of projects that are out there, it's better when they're supported. So, you know, that's kind of my current take. Someone was going to support it. It just happened really fast. And so there's a lot of news and drama about it. And of course, lots of opinions. I don't have any except I'm here to use the technology to the math. benefit I can get and contribute as much as possible.

Jesse? I hope, yeah, I think as a user, I hope the product just actually keeps getting better and better. I think obviously there's a little irony, like we've got open claw and we've got open AI, and maybe they're not so open. Hopefully open claw, it continues to be an amazing open source project. We'll see. All right, so Jason, you were talking about how you were worried, you know, he is, this is the chosen app, don't sell the Sam, the foundation model. model. There's a lot of open source foundations out there, kind of a tried and true method. Do you think

that it's going to be a good setup for OpenClaug to persist and keep improving over time? Or is this more of a SOP to keep people happy as Open AI tries to get all the juice from the particular limit? Okay. I'm going to give you the most optimistic take and I'm going to give you the most cynical worst case scenario. Give you both. Which one do you want first, Alex? I want the sad cynical or you want the best? cynical first and then good after that, yeah. Okay, here's the most cynical take. This technology was becoming so powerful

that it commoditized OpenAI and Claude because it wins the interface. Whoever wins the interface wins the day. You can see the Chrome browser. You can see MacOS. You can see Windows. If you own the operating system, you own the distribution.

It was becoming very clear that the front door for using AI had switched from chat GPT chat room to now people using open claw. Anybody using open claw stopped going to Claude XAI Gemini. You just stop going there and you interact with your agent and your agent goes into interface them, which commoditizes them. We know it commoditizes them because the discussion in under 10 days became in our organization, how do we get the maximum number of tokens for the lowest possible price? And everybody, and I'm sure Jesse, I see her nodding and Heaton will all agree, that you become a token junkie when you start using this.

And you say, I have to get a Mac Studio because I want to run Kimmy or I want to run, you know, deep seek. So that is the most cynical take. So what would you do if a most cynical person in the world, the most cutthroat person in the world, let's call him Zuckerberg, what would Zuckerberg? What would Zuck do? What would Microsoft do? What would Gates do? These are the two, you know, litmus test, most cutthroat. What they would do is they would sabotage that interface and make a better one inside their product. What are they going to do with OpenClaw? They're going to let it chug along. But then they're going to use Peter and his big brain. They're going to rebuild this product inside of Open AI. And instead of, of going to OpenClawn doing all this, you know, chores and tedious work to set it up, they're going to make it one click and give it to their billion users because they have

distribution in order to preserve their distribution. And then whatever innovations, the safest, most innovative, easiest version, the most easiest version will be the default interface on chat chpT within six to 12 months. In other words, in chat GPT, you're not going to get a chat interface and ask a question to pick a model. You're going to get a persona and the persona is now going to operate and they will flip the default. And it will start with two buttons, your assistant's name or chat chippy take, which would you like to use? And then at some point, they flip the default. That's the most cynical take.

Why is that cynical? That sounds awesome. That sounds awesome. What you're describing sounds awesome to me. I'm struggling to find the cynicism here. Create an exciting new product and to do it fast. So more companies are getting built, which is awesome. But there's more to starting a company than just building an MVP or vibe coding something. That's important too. But if you're serious about growing a real startup, you need a Delaware Seacorp. And that's going to give you a major competitive advantage. People will take you seriously. You'll be able to raise money. And that's where Northwest registered agent comes in.

they're going to give your new business a real identity. This means an address to use on your public filings, like an actual mailing address, a domain name, a custom website, a business email, a phone number, and they're going to do that in 10 clicks. I'm not kidding. 10 clicks well under 10 minutes. Plus, NWRA will provide you with step-by-step guides, compliance reminders, and they're going to help you get all the advantages of a Delaware C-Corp, regardless of where in the U.S. you're operating out of. So get more when you start your business with Northwest Registered Agent. visit Northwest Registeredagent.com slash twist. And the link is in the show notes.

Visit northwest registered agent.com slash twist for more details. Because it won't be open source anymore. And it will lock in all your data and take every innovation and privacy you have and give it to open AI. So every single thing Jesse is doing, we're going to hear about her project in a moment. Every single thing Jesse is doing with homeschooling will be owned by Sam Waltman in chat chippy. Jesse's going to have local models.

Like, don't you think that that's what they're, isn't that probably what they're scared of, right? So like if you're willing, and I understand that like we're on a bleeding edge here, but if you're willing to, you know, get your Mac, your friends over here going, then you're willing. Yeah, yeah. Was that five Mac Minis? I'm really, they're outnumbering the children.

But if you're willing to do that, then you're probably willing to figure out how to do local models. And that must be scary. So that, that's the scary because, if you're chewing tokens and you're telling your agents like, okay, chew less tokens, get more optimized. I've got a nightly build every night where they're like figuring that out like together as like a group, then they're going to say like they're going to be like, we need a Mac studio right away, Jesse. Like that's, you wake up to that news.

That is must be scary if you're Sam Altman because basically not only do I not want him owning my data, but I also just want to save money and I want to like have a private, you know. So you're aligned with me with the most cynical version of why he's doing it, Jesse? Yes, I'm aligned with you. I think that there will be folks leading the charge, and I would put myself in this bucket, to even if that future is ahead of us, that there's gonna be these alternative paths

to going local and keeping your privacy. And that's really important, especially when I think about my children's entire education being like sitting on a Mac Mini or something. But I understand consumers like ease, but I think they also like saving money. And local models are gonna start being such Delta. So I don't know. I don't know how that's going to pan out. We're going to get back to

the model thing, but I want to hear the other side of Jason's argument. Jason, what's the non-synical take here? Here's the most non-synical take. Sam Altman is going to build agents and they will become the default for a major part of what he's doing. Open AI has fallen behind and everybody has fallen behind Anthropic because of co-work and Opus 4.6. Now, this has been a case of leapfrog. We all know that. But Gemini is going to win the consumer. That's my belief. And Claude, co-work, was going to win business. And that's where it was trending. And I would say XAI was going to win news and breaking news, right? And data centers and space, whatever. So, and then you've got all the open source models. I believe open source is going to win the day

on the token race. Okay, put that aside. The most optimistic model is, Sam, for less than, if he gave a billion dollars, less than 10 basis points in the value of open AI, say it's a trillion dollar company. He gave one-tenth of 1% to have this person in the building and 10 people around him and gets to watch their meetings, gets to say, oh, here's what they're doing, here's the plans for this open source project, how does that inform everything we do? And then, hey, can we just make sure that instead of it defaulting to Claude's, you know, what's the name of Claude's, Claude Code? Claude Code versus Codex.

Like, that could be the reason. Like, just want to make sure that Open AIs codex is in there. Want to make sure Open AI has got the hooks in there as much as possible. So it's just a defensive strategy to make sure that they're locked in and they have as much intelligence as possible. So you're giving Peter a billion dollars to be in the room with him, to watch him work on the open source project. So the rest of your team's informed for their consumer-based agent that gets to know you like Open Club. That's, and then Dave Moran is one of the most thoughtful, creative, honorable human beings that I've met in the industry. Let me say that

clearly. Sam Altman knows, like Sam Altman and Zuckerberg, the opposite of that would be Dave Moran. That's the opposite, right? Dave Morin is the high ethics, do the right thing for the community guy. Zuckerberg and Sam Altman would be do the right thing for the share price guy. Two different types of motivation. Yeah, Heaton, I see you nodding your head about the Dave Morin comments. What's your take on him being part of the foundation, helping set it up and helping delete it? You know, I was talking with some of my friends, and I think you folks will appreciate this analogy, but is he like Peter's Sean Parker? Like that's all I could think of, right? Like someone who helps him kind of understand and, you know, rationalize what's going on in his life all of a sudden, right? That's

kind of how I would put it. So I echo Jason's statements that someone should do it. Dave is a great person to be the one to help out with that. Yeah. So that's a great analogy. You're bringing some humanity and a thoughtfulness. So then the question becomes, what do we do here as a community? And I think what we have to do as a community is build this product and support so many people working on it that it never gets locked into open AI. So how do we do this? Number one, security. Number two, ease of use. Number three, 10,000 hosting services, a thousand hosting services, driving the price down. So we're going to invest here at my firm launch and our accelerator.

Any two or three entrepreneurs that want to work on security, ease of use, hosting, or skills. If you have, and obviously any other idea, like Jesse's idea I love, and the application layer. But of those four, like, core principles, the skills remain open source. The security is paramount. The hosting is cheap and widely available. anything in that range, we want to fund. We funded two companies on the show last week. I have 20 slots, 10 in Founder University, 25K checks. That's for like really nascent projects, two or three people working on them. And if you're just one, we'll find you two other co-founders. We're going to do a co-founder, open-claw,

co-founder matching service. On the other side, if you've got the product made, maybe even have a thousand or 10,000 thing, we'll do the 125K, come to the launch accelerator. Then I will introduce you to every investor I know. I had 10 investors email me the past week. Can I invest in the companies you're investing in? You're ahead of me on this. Just tell me what to put money in. Literally had one of the highest profiles who just said, I'll match every investment you make blindly, J-Cal.

I trust your judgment. So we're going to put all of our energy into OpenClaw for the next year. 2026 in our firm is the year of OpenCloat. We must maintain this open source project. Ease of use, security, skills, and a competitive hosting environment. so that we don't get into a situation where one person owns the hosting. Somebody, they get the skills.

You know, they come up with some fakaka crazy. We're going to have the skills inside of opening eyes, so it's safe for you. And they hijack that out of the project, right? That's my concern, is they hijack hosting. They hijack skills. We're not going to let them. Yeah, but I think what's cool is you can just say to your open claw, hey, build this skill.

So basically, like, I think what's so mind bending is, you know, software like you know I built something for my like I built something for my kids software-wise that I I'm a previous startup founder who hadn't opened terminal until six months ago okay so I was the my my co-founder was technical so I like but but after six months ago started claw coding then open claw and now I can build anything like software is feeling free to me and I know and I used to pay hundreds of thousands of dollars to developers it didn't feel free in the past so that Doesn't that just mean like it's going to be hard for anyone to control when I can say, build me this skill and it just, I have it in an hour. So I had the same experience.

I was worried about security when I was rebuilding my open claw instance, Jesse. And I literally was just like, wait a minute, why don't I just make all these skills myself? Like, why would I bother to go through the security risk? Way less security issues. Someone, someone was talking about a security issue with a certain, you know, skill. And so like instead of plusing, you know, get this skill, I just went to my open claw and I said, I like this skill, make this for me, private. Building out your team is one of the most crucial things you have to get right in your startup. And finding the right developers is particularly important. But now there's lemon.io.

They're going to save your time, money, and headaches by doing all the time consuming legwork for you. They've got an experience lineup of pre-vetted developers working for competitive rates. Just 1% of applicants are accepted into Lemon's elite program. And they're not just out there finding this great talent, they're also working with you to integrate these new members into your team. Plus, if it's not a good fit, hey, and sometimes things don't work out, Lemon will hook you up with a new developer ASAP. I've seen startups go from just pretty good to amazing after filling out their teams with developers from Lemon.io. Go to lemon.io slash twist and find your perfect developer or technical team in 48 hours or less. Plus, twist listeners get 15% off their first four.

weeks. That's lemon.io slash twist. L-E-M-O-N-I-O-S-Twist. So, Hayden, please. I got to chime in here. I think, Alex, if you were looking at the slack I added to, you'd know why, but you might not have been looking at it. So while we were talking, I got Jason's book. It's probably somewhere on my shelf, but I have a PDF of it. Don't ask me how I got it. And I got it in the dark. You know, I got it. I got it. I got it back here. I paid for it, Jason. It's a great book. And basically, Basically, I took the book and I think you might remember the four founder questions. So right now, Bob, one of my bots, agents, whatever we want to call them, he's actually making that skill right now off of your book.

It analyzed the book, found all the frameworks. There's like three frameworks, the four founder questions, why now, and portfolio math, and it's going to go after making those skills and then we can use them. This is really, really, really cool. Yeah, I wasn't watching Slack Heating because we were podcasting. I don't know if you notice, but we're doing the show right now. And if I just read Slack, I'm not going to be a very good podcast. Before we get into demos, I do want to do demos in a second,

but I'm curious what everyone is using as their own intelligence engine, aka model right now, because I've been tinkering with GLM 4.7. I know GLM5 just came out. Minimax. Explain what GLM is, please, for the audience. Yeah, so Z.a.i or Zipu AI is a Chinese AI lab. And their model family Jason is called GLM, similar to how, open AIS models are GPD dash.

It's just the moniker that goes with them. There's a technical reason for it, but that's probably one level lower than we need to get. Lots of great new models out there, really cheap. I'm just curious, Jesse Heaton. What are you guys running when you're burning a lot of tokens? I haven't been optimizing yet. So I've really been using like Gemini models and Anthropic models.

And I do have, depending on what the agent does, I have them flipping between to optimize basically token usage. But I have a Mac studio coming. And like they're on delay, you guys. I can't just get one tomorrow. And so when that comes, I'm going to start playing with some local models, but some people might be experimenting more than me. Well, no, I mean, you don't even have to run these locally.

There's a lot of inference providers. No, I know. Yeah, I've just been, I've been trying so hard to get my agents behaving. I haven't been experimenting as much on the model side. But I've been asking to optimize tokens based on what metal they use when. And like being way more precise about that. That's like where I've been optimizing more so than playing with. the new models that have been coming out every day so far.

Hayden? Oh, I'm using Kemi for coding. And then we do a bunch of local training using Lama. Still works. Oh, really? Yeah. Oh, all right. And so on a comparative basis, you're burning through tokens that would cost

from a major model provider, how much? Do you any idea, Jesse, or Hayden, how much you would be burning in, or even at your peak, what you were burning in opening eye or Claude or, X-AI or whatever tokens. Sometimes on a single day, I'm like, you know, really burning through cash. So that's why I do want to optimize and start playing with local models, but I haven't been doing so yet. But if I'm really cranking and having them doing dev projects as well as doing all the, like,

book processing that I've been doing for the homeschool and stuff like that, you know, sometimes I'm getting those like notifications like from Anthropic about sending $15, like, every 45 minutes or something. So I've been, so I'm definitely motivated to optimize my usage, but haven't yet. So low thousands, certainly hundreds of dollars in that. Yeah, yeah, but the way I think about it, and I have the luxury of thinking about it this way, but, you know, I have, I've onboarded, you know, like, in my startup history, like, definitely over 100 employees over time. And the way, and so, like, I'm really price anchored to

how much people cost. I do, maybe that will be a legacy way of thinking, but I'm so price anchored to how much people's time costs that even when I'm kind of spending a lot on these models. I'm like, I'm getting so much done. I honestly don't overthink it because I know that I'm getting projects done that I wish I was able to do four years ago or I feel like I've got three people working on curriculums for my kids and I'm like, okay, I spent, you know, $40 today. Like that's okay because I can't afford it. So like lucky for that. But I just, to me, it's like that's still so phenomenally cheap. But I have bigger projects like with book ingestion and stuff where if I really like really want them to chew tokens,

especially I'm like reading videos where I record videos in my kid and they like actually maybe scan the videos to understand the lesson. This type of stuff. I'm thinking local models will help me do that without just spending like thousands unnecessarily. Yeah, because you can really, you can really blow your token budget. My first $50 in Anthropic API credits that I bought were gone in the snap of a finger because I was using Opus 4.6 for the wrong tasks, Jesse. And I think everyone learns that lesson, you know, about one time. So I've been thinking about like what to share and there's a lot of, lot of things I've been doing over the last, what is it, like 20 days or so since this thing came out

and throwing them in a bunch of slack. So I have about 10 or 15 of these bots running with OpenClaw. The best analogy I've come to to explain this to people is if with Chat TPT or Cloud, you get a car. With OpenClau, you get a car and the mechanic built in. And so taking that analogy, basically, you can extract it out or abstract it out to everything that I do. So what I do with models is we We have it set up both locally and have an open-cloth agent dedicated to this, where we're actually taking the workloads we have and automatically running e-vails on them. So evaluations where we're evaluating the AI. The AI is evaluating the AI on those jobs in order to get to a spot where it can be either

efficient or the right quality for what we're looking for. And so this is like an automated system that you would run if you were building an AI Now with just a bunch of prompting, I can run that loop. And so we're getting new data all the time. Jesse's been doing a lot of awesome work. And one thing that really kind of caught my eye, Jason, was this video tool that she made as a way to kind of curate YouTube for her children's educational purposes, all using an agent.

So Jesse, take us through it. Yeah. So let me share, share screen real quick. So I, I don't know if any other parents, I mean, I sound stuff you can say. I know other parents suffer from them. suffer from this basically there's amazing content on YouTube but but when you turn on YouTube the app it's defaulting to shorts it's defaulting to slop your little kids especially get tricked by the AI cover art and they like click on a video and it's actually junk so I've actually wanted to build this for years and

this is where I'm saying as a previous founder like software used to be expensive now I can just chat on my phone with my open claw who's using Claude code and codex and stuff and I and I built this so effectively I can choose what I call streams and we love like engineering videos science videos camping and stuff and I can subscribe to these streams my open claw one of them because I've got a few set up one of them is based on what I select on streams it's actually going through and finding and pointing my app at videos that fit that content stream so this is not static content I did not select all the videos myself like some kind of playlist

My open claw is based on what I want my kids to watch, selecting streams, and then I built, so then I can do that interface in here. But then I have the app, which I can use with my kids on TV. I got it actually set up on my TV, which is like another, it like took a little bit more work. But they can just click and videos pop up to play. And all they can do is advance to the next video or pause. They can't get out of that interface. So the slot is gone. Okay, slop is over.

But I've been wanting to build this for years, and finally I can, because I can't even sit at my computer for more than 20 minutes at a time with like four little kids. But I'm literally just, I coded this on my phone with OpenClaught because I can't even sit down. So this is an app or a website right now? This is an app. So this is just private to me. I'm not here pumping it out. I will pay you $100 a month for this app immediately. Okay.

And I want you to cure it into feeds. This is for my family right now. I need to make sure I don't run afoul of any other things I did with YouTube and stuff to make sure other people could use this app. I'll join your pod. I'm going to join your pod and I trust you. No, my 10-year-olds are on YouTube constantly. You identified the problem, which is slop.

Kids are like these navigators to find this stuff and the algorithm pulls it into it. And that's what I really want them to do is when I give them documentaries or creative stuff, they love it. But the algorithm is so powerful that it will pull them to, you know, two girls doing a makeup tutorial or, you know, for startups, every single hire matters. But posting an ad and sorting through all the applications. Oh, it can be a huge strain on your time and resources, don't I know it? But thankfully, there's LinkedIn hiring pro.

That's right. Our friends at LinkedIn are going to help you hire with confidence faster than ever before. For example, we're hiring new editors here for the pod and reviewing all these real and applications, this can get overwhelming very quickly. We know what we're looking for. It's just a matter of tracking them down. Hiring pro streamlines the entire process, helping us quickly draft a perfect job post,

then using AI to shortlist candidates and even conduct initial interviews. So we can zero in on the best overall applications right away. Nearly 60% of companies find a quality candidate within their first week. So hire right the first time and get $100 off your first job post. posting by going to LinkedIn.com slash twist. Terms and conditions apply. Disney princesses smoking

cigarettes. I know. And YouTube shorts is pretty much trash. YouTube kids, I feel like it's actually like a junk product. I don't know if that offends anyone, but it's like you go on there and it's supposed to give you these parental controls, but the content's not good. So I build that. So that to me, that's just an example. Like, yeah, I was a previous founder, but I'm, I'm just a mom now. And I'm able to like build this stuff, uh, you know, off to the side. and I'm not exaggerating. I cannot sit on my computer. That's what OpenClaw has given me

is basically its hands on my computer when I cannot sit down. So I don't know how relatable this is, but I think as a mom of young kids, like I was able to code this literally on my phone, like talk, I do voice notes, whatever. The other thing, the more macro thing, and I think you responded to this on X,

is that I'm using OpenClaught to develop the entire curriculum to homeschool my children. And this is just, this is obsidian. So my co-founder of my previous startup is now the CEO of Obsidian. So I'm kind of like geeked out of an Obsidian just like, you know, because I know him super well. But Obsidian is basically a collection of Markdown files. And AI loves Markdown files.

So I'm using Obsidian as the basically brain of my homeschool. So if you see here, I have all these individual lessons logged of my children doing specific things. But here's where things are interesting. Like you look at all this data and you're like, wow, that looks like she's putting in a lot effort. Like this is a really ADD person like logging everything her kids do. That's where it's beautiful. I took this picture and I did a voice note. My kids did color theory today. They did a color wheel. Blah blah, blah, blah. My OpenClaw writes this entire lesson log and logs it. I don't touch the computer to do this. Okay. That's the difference with OpenClaught is that they're like

actually doing all the work for me. And then I have this beautiful log and my like, you know, kind of galaxy brain thought is that when they turn 18, I can like literally give them the entire history of their education. And on a daily basis, I can ask OpenCla, like, okay, tell me what Ford did last week. We need to plan lessons for him this week because I can't even remember, right? Because I'm running around my head cut off. This is what teachers always talk about. I need more time to lesson plan.

I want to do more time with the kids. I want to do less administrative. I want to be in less meetings. Give me more resources. Same thing is happening with doctors. We have a startup, one of my team. members will tell me, we'll pull up the web page right now, that is doing note-taking for psychiatrists, specifically for psychiatry. Psychiatrists, you don't want to spend time with patients,

but they have to do their billing, their codes, their not-taking. They're running a business, yeah. They're running a business, right, essentially. And so this startup, I just went through our accelerator, sorry, I'm drawing a blank, I guess, got all the snow dumping down behind me. And after the show, I'll be out in that powder. We desperately needed a snowstorm here in the in America. So this company, next visit, which will show on the screen here, they report that doctors were getting back like 10% of their time. Now they're reporting doctors are getting back a third of their time. And the notes are better. And so that's what you're experiencing. Here's next visit. So it's just medical note-taking, but man, it has done so well, next visit.

Visit.AI if you want to see it. Okay. I, Jesse, I love what you're doing. It's a business. You're a mom. Yeah. I think there's like three ways to go with this as a business. I'm enamored with what you're doing, the passion, the focus. I mean, are we doing this or not? I mean, I think you've got a business here. I want, the fact that in one generation we went from about 1% of kids homeschool to about 6%. Like, it's a personal passion of mine that more parents feel empowered to homeschool. Because I think that education quality is incredible. There's obviously lots of cool alternatives like Alpha and everything as well. But anyway, so I'm passionate about sharing.

That's why I started sharing. Okay, I'm not an influencer type personality as my main thing, but I decided I've got to share this because my mind has been so blown by what I'm able to achieve with my kids. Like I feel like I've 10x to myself at least with OpenClaw that I was like, I must at least share. I'm going to start with sharing. Okay, that's my only promise. From there, maybe there's some businesses, but I think we can get homeschooling

past 10% of kids within another few years from now. I'm passionate about that. Maybe there's a business in there too. Okay. I am a crazy risk taker. I will give you, I'll invite you to our accelerator, give you 125K just to pursue your muse for the next year. If it doesn't work out and you burn the 125K, I don't care. I like to take risk. If it's a one in 10 or one in 20 chance, I am okay with long odds. Just let it percolate. You don't have to give me an answer right now, although that's great for ratings.

But I'll give you the 125K. And what that will allow you to do is hire a second person who you love trust. I can hire you afford my own Mac studio, Jason. Okay. No, no, I'm talking about a human. I'm rich. I don't need you.

Gotcha. Alex tells me that all the time. He still shows up. He still shows up for work. I've been through YC. I do believe in like you surround yourself with like great people, like great things happen. So, so, yes, I'm very, I'm very interested.

I also have a three-month-old. So for me, it's probably more of making sure I've got, you know, life bandwidth. But I love- You can spend the 125K on a 12-hour-a-day nanny. You can literally spend it on nanny coverage, so you get an extra hour back a day. All right, I'm going to put it there. It's nine months of nanny time there, Jason.

It's inceptive. He's incepted. He's incepted it. I love the idea. And, um, and, well, No, you know what I think an interesting idea would be for year one? Because I am passionate about this space as well. I have a full-time teacher working for us.

And that's worked out amazingly for our family. We have means. But I think there's like the people with means are doing this already. Yeah. Then there's people with no means who are just suffering at, you know, in this terrible education. This is people in the middle. Yes. I think like Uber did Uber.

People had chauffeurs, right? If you're a rich person, you get a Maybach, you pay off. You got it. So that's us in this analogy. Then Uber Black came out, and then the next tier down said, oh, I can have my own personal driver take 10 rides a week, $60 a ride. Hey, that's $600 a week. It's $30,000 a year.

Okay, now you get access. Then you give UberX, now another set of access. And then eventually autonomy and Uber pool and lift line, those things kind of bring it down. There's a next tier here that I believe would pay $500 a month, $1,000 a month to have a set of tools and a curriculum and the ability to be part of a community. And I think you get a hundred of them to pay that amount, just middle class folk, you know, upper middle class, whatever. And then as you do the tools, the advice and everything becomes free for everybody at the bottom to roll up

their own, spin up their own. You're doing great work in the world. But there's a business model here, that's subscription. I agree. I think that we, I really do believe we're trending towards over 10% of U.S. K through 12 students being homeschooled. And when you think about the transformation that creates for all of us, a generation from now, also, that's like, to me, it's like one of the most impactful things I can work on, truly. So I'm starting close to home, literally in my home.

But yeah, I want to, I view it as a mission. All right. I want to get to another demo, though. Let's keep the demos going. So Heiden, take us to Slackville and tell us what you've cooked up with OpenClaw. I'm actually going to give you three things. And so let's start with the first one that I know everyone loves to talk about these kind of use cases.

I'm not an automation junkie, so I don't connect this stuff to my calendar or my Gmail like a lot of other people do. It also helps me avoid any of the security concerns of it going rogue. I connect the open clause to GitHub. We can give it documents in GitHub or give it direct links, right? But one of the main use cases that people go after is what I'm going to show you in this channel. So I started asking it what I was doing, but basically a couple of days ago, I basically built a contact database for one of the products I'm building. It's a competitive intelligence tool, and usually that's like product marketing and those

areas of a business that worry about it. You can imagine why I'm building it. I think those tools, they're just in the last 20 years, I've been wanting to build something like this, but now the access to information is so easy. So I decided, okay, let me just give it my LinkedIn contacts. My business partner is going to put hers in there. I gave it my Gmail contacts for two of my email accounts. And then I basically had it make a database,

score it, prioritize it, tier it up, and basically be able to tell me who I should be outreaching to. A lot of people go to like automation, go send the automated emails. There's already even been public kind of screw ups on that where the Cal.com founder said unsubscribe to the beehive folks. I don't know if you folks have seen this. It was kind of a lot of drama,

but it was because of automation. He said unsubscribe to a Behive email from the Behive people, not a Behive newsletter. It was automatic and it was his agent. So I don't go for that. But what I do want to go for is this kind of work would have taken me finding tools, paying for tools. Maybe they'll work because LinkedIn doesn't like giving any tool to data.

But instead, I'm able to do this here. And the most powerful thing is that now I can take this and go give it to another bot. So I've already ran this three or four times. We run it at Crazy Egg now. Same kind of thing. This does our outreach. by basically finding the people organizing it. We can give it more context.

I mean, you can just keep going, right? So you've actually built here the personal CRM that everyone has talked about, Jason, since I first started paying attention to venture. I feel like personal CRM has always been the white whale of everyone who's really busy. And I did it in like 30 minutes. And I could keep running it.

And so, you know, they talk about replacing software and all that. I just want the outcome. I'm not even thinking about that. I can get the outcome by treating it as a mechanic and a car. So here I just asked the mechanic, what are improvements you can make? And it just gave me all the improvements right here, all these key improvements, email enrichment,

stale data, whatever it's got. Then I said, give me a detailed plan, research this and give me a detailed plan for each one. So for email enrichment, obviously, I have to use an API or something. It's telling me which one I can use and what my options are. It even phased it out for me.

I can have it build any of these. And I'm not using anything special. This is basically open claw, plugged into Slack with a good enough account at chat, you know, chat, TAPG account, and I can do this. So that's one of you use. Keep going.

I'm gonna give my second one. So second one, I know a lot of people have shown like control panels and stuff like that. This is a completely different business. Both of these are newer ones that I can share stuff from. We've got a lot of stuff going on. So this looks like an interface, got a file explorer, a memory search.

It's like an open claw interface. Here's the thing. Today is the first time I logged into it. We've been running this thing for this is my first instance. It's called Otis. We've been running it for like 20 days or something like that, 15 days, whatever, really early in within a few days. We set this one up. This one actually also has a local Nvidia 5090 sitting there with the engineer who works on it. Again, not super fancy, still got clawed and all that stuff. We have open router hooked up. But the thing is, everything you see here, that I'm going to show you. I'm only going to show you like the skills area and maybe one other area. It built it. It keeps improving it. We never even log into it because I don't really like interfaces anymore,

but that's a longer story. So you built a dashboard. Yeah, yeah. But it built it. We don't even use it. It's his own dashboard. This is literally what happened with Oliver on our team, who's now the producer of this week in AI, our new podcast coming. He had it built his own dashboard, his replicant. And it built it. And now we have an interface to look at the skills. And then we're telling it with, I don't know if you know, Matt Van Horn's last 30 days. Yeah. Yeah.

Every Saturday, I'm having it go out on Saturday at noon and saying, go research everything about thumbnails. And every Sunday I say, go research everything about titles for podcasts, you know, et cetera. Tell us the best practices. Give us examples. Tell us who's talking about it. And it found a video that was out in the last 30 days from somebody who had work for Mr.

Beast who explained Mr. Bees's philosophy. Now, we've heard his philosophy before, but it was something recent. And I said, just keep adding this to your memory and skills. Create a notion page or a Google doc with every week what you learn. And then every week, it's going and get better. You compare that, Jesse, to a human being, where a human being is just not capable in my experience of having that level of discipline. And if they are, they become Mark Knopfler of Dio Shraits and become the greatest guitarist all time, where they become, you know, Quentin Tarantino and they make the greatest, you know, films of all time or whatever. That level of obsession to be able to every week to get better at your craft is for impassarios. But it's now the default, Jesse, in this replicate world.

I like, yes. And you also, I've managed a lot of people. I try to treat the replicants as you're calling them like I would people like I'm not rude but I'm far more direct I don't have to worry that someone had a bad day and they're actually all teary because their boyfriend broke up with that I don't I don't have to do that kind of management anymore so I can just be direct I'm not rude but I'm direct Hitton said something that I think is really interesting which is that you eyes are kind of dead right I don't know I don't want to speak for you but but like maybe taking to another level are you eyes dead like because six months ago I was trying to clod code an app for my homeschool,

like an app that I could click into and use on my phone. Then when OpenClaught launched, I'm like, there's no app. Like, we're in an appless world. I don't actually want an app, right? I don't want an icon. I don't want to click into something. I want my husband to log lessons too.

I don't want to make him get an app. Apps are, like, there's no more apps. Like, there's no more UI need. He just, like, he just needs to be able to chat with the, the replicant, and so do I. So that just means UI is like what's the future of that? I mean there'll be some like niche ones, but it's crazy. Like that's a crazy shift.

Yeah. This is like a directory for everything we've done. So you go here and like you can see the skill builder I wrote. But like look how big it is because we've been entering the crap out of the skill builder so we can build a really awesome awesome skills off anything. All our research is here. I can click into one of them and it's got the entire document. Where are you running? Are you on your own machine or is this in a cloud?

This is in the cloud. I exclusively run them in the cloud for these things, but for training and bigger model jobs, I'll run them locally because I just think we do have security issues with it. And there's no, there's a way you can contain that and that's the whole point. And those are all dot MD files, right? You're like a whole dot MD. I clicked on my first article that had a copy in dot MD at the top and I was like, yes. They were finally, we're all connected now.

Explain what that means to Neophytes. Yeah. So marked down files have existed forever, like really light structure, text-based structure. They're brilliant for AIs reading them. Like AI also loves to read. They can really quickly scan, really quickly understand. You can do linking between them with a tool. That's why I use Obsidian, frankly, is because they can link between and create an organization

for them. But a file structure that exists, file format that existed forever, but now it's especially relevant because humans can read mark down files because they have nice formatting and AIs can. So we've reached the like file type that we both want, right? And so I clicked on an article. I'm not sure who because I'd love to credit them. And they have installed a copy in dot MD at the top of their articles that they're writing. And I was like, this person gets me.

I copied it because I'm not going to read it. Okay. I'm holding a baby. I click the button copy and dot MD. Give it to my replicate. They read it. I say make me a podcast. I listen to it.

And then now we're cooking, you know? Heaton just threw something in called the website to markdown. Explain what this does. You can take any website and basically get this markdown format that Jesse was talking about and then just give it to an AI. So, you know, like if you're really messing around with chat dbt and things like that, it's not getting to the webpage, you can just literally just click that and then it'll just download it in that format.

Okay, wait a second. So website to markdown. You go to a website. Let's say there was a website. It was somebody's substack. It was Alex's substack for cautious optimism. You then go to it. You say just download the whole McGillard.

It won't take the whole website. It'll just take a page at a time the way it works. But then it just makes it easy to get it, get that data in whatever place you want. So it's a scraper as a Chrome tool that converts each page. So if I was a paid subscriber for but $100 a year, why wouldn't you subscribe to cautious optimism? Get Alex's thoughts. you could take his entire opus and then say, I want Alex's thoughts in my memory for my bot because I'm a technology analyst at a venture firm.

Just an easy way to get like, like, you know, there's better ways to do that now where you can just tell it to go crawl a website, like our open claw bots, if they have web search API like Brave connected, plus they have web effect, they can go get pages and crawl entire sites. Like Jason, like the other thing I wanted to show you, folks, this was number three. I had three, right? Yeah. It's basically how I, I think, Alex, you were going to ask me about this, so we might as well go here, right?

Yeah, well, I'll kill. Well, you bring it up. I'll explain. Jason, Heaton was talking about something called pair prompting, which was an entirely new method of working with AI. And I wanted to explain to us what that means and why it matters. So I have a personal Slack. If it's Heaton.com, you can't get it.

You can't go in there unless I invite you, of course. So I've invited Alex and Jason there. And I wanted to show them how I essentially indoctrinate, for lack of a better word, humans, into understanding how these AIs work and how we can work with them. So basically, I have you both in here. And Alex, I know you already started following up, but I happen to have Jason's book, which I talked about earlier. So I had it go analyze it. This is a whole thread on it analyzing Jason's book, coming up with all his frameworks and starting to make skills based on that for founder questions, which it's in the process of. and here it is.

And I even have it testing the skill. It even found red flags to check. And it's just recursively doing all this. And it's going to perfect the skill. But I built a skill builder first to do this, right? It's basically the mechanic and the car. You get both as long as you know where to point it. Right.

So we were able to take a book and turn into a skill right here in the matter of this show, right? So that's one. And then another one is I have a bunch of research tooling and skills we built. so it can quickly make a dossier on someone. So I knew Jason was gonna be here. And now you're in the Slack, so you can ask it anything, Jason.

I've done this for my friends, like our mutual friend, Oh Malik, where he's got a long history of writing, right, even more so than you, me, or even Alex. And I was able to get like his whole dossier in there, his trends of how his writing has changed and everything and all within minutes. So here, it's got your history.

I even have a fact checker. So I told it to make sure everything is fact checked and there's multiple sources for it. So it's assessing that. solve the hallucinations and all that stuff because I'm using it as a mechanic to fix the car. And then Alex and I started messing around a little bit, try to get it to give us something surprising about you. It started. It started. And then we made it dig deeper. Okay. Now we know you have a psychology degree. I think that explains a lot, Jason. And then it went into essentially

this story, which I've heard before, but it was just nice that it got here. And kind of was trying to hypothesize how you became Jason, basically. Yeah. Yeah. It's a famous story that I tell in the book, my dad's bar got seized by the feds because he didn't pay his taxes after the stock market crashed. And so I had to figure out how to pay for school. And I basically wouldn't take no for an answer. And I tell some of those resilient stories of just getting knocked on my ass, you know, a dozen times in my life and getting up 13 times and just, you know, how doggedness and resiliency are the key, key features of success. which my friends are always apt to point out to me.

All right. Heaton, can you stop sharing for a minute? We're going to do one last question, which I think is really pertinent to everyone who's both listening to this now and also on the feeds later on. Cognitive overload. One thing I love about OpenClaught is I can do so much. One problem with OpenClaught is that I can literally do so much.

So when you guys are deciding what to automate and where to apply the tool for folks out there who maybe just starting to tinker with OpenClaw, How do you handle just the feeling of being a little bit overwhelmed by the possibilities, the options, the risks, the tools? It's a lot. Jesse, why we start with you? Yeah, my first thought is I take the things in the progression of my normal day, the things that weigh on my mental load, and I try to pass that off.

So, you know, a quick example is that I was overwhelmed by how much crap we owned for like homeschooling and stuff. And I was overwhelmed by pulling out, like, what to pull out. So one day I realized that I took a bunch of photos, gave them to my open claw, and it made an inventory. And so then it tells me what to put out. But I think that this is a way of answering, like, it works for me. So I think people are having this anxiety of it not working. I am a Jomo person instead of a FOMO person.

So this is helpful for me, my mindset, because, yes, there's times that my open claw is not working, but I don't freak out about that. But I go through my day anytime I hit a slow moment where my mental load is like Max, I give that job to my open claw. and try to automate it. All right, Heaton, really briefly, same thing to you. Just give me the two bullet point version of handling cognitive overload in the era of open cloth. I just try to pick and choose. It's a whack-a-mole problem, right? Like, whatever the thing is that's coming up, oh, contact database,

I need to contact more PMS for my competitive intelligence tool. Great. How can this thing help me get there a hundred times faster? And every time it seems to do that. So once you get in a habit, it's just like using chat dbt. Well, Jesse, I think you went to YC. I think this project appeals to entrepreneurs for a number of interesting reasons. I have my own thoughts on why an entrepreneur becomes obsessed with this so predictably.

Why, Jesse, do entrepreneurs get so consistently, so predictably addicted to open claw in your mind? Because we as founders, we always thought we could do it all, and now we can. Like a, it's like a big mania, okay? Like I thought I could do everyone's job a little bit better. I'm just being a jerk. But now I can. I also have worked with amazing people. I'm being facetious a little bit, but I now I can extend myself like infinitely.

That is exactly at Heaton. What do you think? why so predictably Matt Mullenweg, you, me, Alex, everybody saw this, started playing with it and said, I can't stop. Why, Heaton? Why is this? Agency. Founders naturally earn, get, use, have whatever agency. This gives you so much more agency. I mean, I'm up to like 2, 3 a.m. I haven't been up until 3 a.m. in the morning since college. And it's because I need to like in one of one of the businesses I help someone start, like it's a marketing agency. Overnight, we increase efficiency 10x.

And now I'm making sure every night, this one in there, the name's Gary, long story. But Gary's always running for this guy who's running the agency. And we scale very differently now. So it's agency. Alex, give your opinion. And then I'll give the correct answer. We're hearing versus the correct answer here. But what's your position here?

I'm going to be slightly facetious. But one thing I've always had was a lot of jealousy at people who. who could take their ideas and turn them into code. And for me, so this is kind of a variational agency. But for me, it's turned me into the person that I wanted to be if I was better with semicolon's patience and syntax. And so to me, it makes me feel like I'm 10 feet tall. And I no longer have to ask permission for anything.

Okay, you guys are nailing it. In my mind, there is a reason why this resonates. This speaks to the rugged individualism and the ability to carve a place in the future by shipping products and building that you very rarely see in human existence in the history of entrepreneurship. And one of those moments was when people had ships and they could go anywhere and explore the new world, right? And you had Columbus and people just going all the trade routes. Then you had people land in America. And they said, you know, going west pretty hard and like right behind me,

Donner Pass, like, it's going to be rugged, it's going to be hard. But if you get to California, hey, there's gold in them, their hills. This is a gold rush, rugged individualism, and autonomy and resiliency. If you're willing to play with the tools, if you're willing to go through the Donner Pass and you time it right, you might be the first to the promised land. You might get there first. And you might be able to build something and stake your claim in this new world. The same way mobile, cloud, the internet itself, when dial-up modems came out, when the PC revolution came out. I remember when people got PCs, they had this agency that if you could embrace this technology,

you would be so far ahead of the average human. If you understood the internet, you know, like I did in college in 1988 when I saw a bitnet and ARPANET. And it was like, whoa, if you were one of the few people who knew what that was, you just you were infinitely ahead of everybody else. If you knew broadband, if you knew mobile apps, you were infinitely ahead. And that's what this really represents is an open, blue ocean, wild west, rugged individual. Go claim.

Stake a claim, folks. Go get this technology and stake a claim. That's what I believe is going on here at its best. But yeah, I mean, it's just amazing. I think you guys also summed it up amazingly. And anytime an entrepreneur starts playing with it, they become addicted. If you're not a member of our YouTube community, just go and type This Week in Startups,

and you can go to This Week in Startups, hit the subscribe button, hit the thumbs up. Now I've got a surprise for you guys. Okay, editorial director Lon is here. So this morning we were checking out AI Scott Adams, I believe we have a clip. Now, this is not the real Scott Adams. They have created a digital clone of the Dilbert cartoonist

who, of course, passed away. earlier this year in January. So here is the AI approximation of Scott Adams that is now hosting a regular podcast of his own. Let's take a look. And to add to what Lana is saying, tragically, Scott Adams passed away the creator of Dilbert last month.

He had said on his podcast, he was obsessed with the idea, and I was friendly with Scott for many years. He had said he was Lon considering making an AI version of him, and he wanted people to play with that concept. And then there were some people who he may have changed his mind a little bit about that. But here is an example of AI coffee with Scott Adams. For those of you don't know, Scott Adams would do at 7 a.m. Pacific time every day for thousands of episodes,

his little take on the news, influence, skill sets, reframes, whatever. He had a delightful show every morning. So here's the AI version of that. I like coffee. Join me now for the unparallel pleasure, the dopamine here the day. All right. I think it gives everyone a bit of a taste of it. And I want to point out that I am not Scott Adams, even though apparently we look very much alike.

I do not know that fact going into that clip. Here. We have the founder of AI coffee with Scott Adams, correct, Lon? Yes. John Arrow is joining us. He and his brother are the creators of AI Scott Adams. So let's bring on John. Hello, John.

Thanks for joining us. All right. John, tell us why you created AI with Scott Adams. I'm assuming you were a fan of coffee with Scott Adams? I appreciate you asked that question. I mean, it's something that really nobody has asked about the story. I literally grew up. My bedtime stories at night were Scott Adams' Dilbert cartoons.

My parents would read them to me. So I was a fan from a very early age and kind of identified with that. And then when he started putting out his books, I kind of reconnected with him around the COVID time. started listening to his podcast very frequently. So, you know, I know there's people that were bigger fans than me by far, but I think I got a lot of value out of it. And like many people, I heard him say it's so important after I'm gone. I want to become an AI. This is how I can get to immortality. I heard him say it time and time again. And then after his death, I went and reviewed that again

and decided, look, this is something that I can do, my brother and I can do to be useful. And that's exactly what we did. And it's brought, I know a lot of people, a lot of joy and comfort during this period. And it's something that's really been a lot of fun. How many days have you done it? And then what is the technique for the content? Because Scott had a very specific content format. He would do this simultaneous sip at the beginning. You would say, oh, you need a couple of mug or glass. Everybody would do it together and create it like a bonding experience. He would talk about the news. he would do reframes for people, just psychologically how you view the world. He would help people with micro lessons, a topic that was so powerful.

There's micro lessons that I bought the domain named microlessons.com, and I emailed him. I said, hey, I have microlessons. I bought it for like two grand. Would you like it? I'll gift it to you. And he's like, no, no, go for it, J. I was like, I don't know why I'm buying it. I just love that term.

So how do you make the content? Are you writing a script? Are you having AI write the script? How does it come together? because it seemed like it's very in line with what Scott would have done. It's an extremely iterative process. So the goal is to let AI do as much of it as possible. We have access to all of the transcripts, all the videos that are out there, a lot of his writing.

So basically the corpus of anything that he put out there publicly, we try to have him use that as much as possible. Now, there are some discontinuities where sometimes Zach or myself need to write. But the goal is to have AI did it do as much as possible. So yesterday, I think AI Scott Adams replied to one of your tweets. That was a case where we saw it, but he actually typed that response. That was not a person. Jesse or Heaton or Lon, do you have a question for John about the creation of AI Scott Adams and what this means? Obviously, John, I take you at good faith.

I think Scott would have taken you at good faith that you're doing this because you're a fan. This isn't a commercial project to make money off the legacy or the estate of Scott Adams. I'm really impressed, I'm really impressed technically about it and I think it's very cool, but I'm curious as someone who might want to generate other content that with other personalities or like I'm doing homeschool, like historical figures, like you're piquing my curiosity about this. Is there anything else you can share technically with how you brought it together? Sure. For those that are interested in doing it themselves, I would just say start experimenting.

I mean, some of the tools that we're using 11 labs, we're using fall. A.I. And this is one of the situations where I think, you know, perfect is the enemy of good enough. So I would just try and experiment with it. The tools keep getting better. Speaking of micro lessons, Jason, we're going to put out a micro lesson very soon on how to let other people create, not only their own version of AI Scott Adams, but anybody, because that's probably been the most consistent question we've got is how can I turn this in educational content or do this for somebody else? And I hope you have 50 of them more. My question is, are you refining the work? Like, as you go, are you sort of like continually training it to be more Scott Adams like or to sort of

sound or, you know, come up with content that's more like him? If so, like how are you sort of layering that in? There's a huge recursive element. And what we use for feedback is the comments themselves on the ex post. So previously we were on YouTube, but they got us kicked off of YouTube. They said people might get confused that this is really Scott Adams. So I don't think they have much thought for the intelligence of their viewers. if they think somebody who was dead would now be coming back.

I mean, we had it clearly labeled as an AI. So unfortunately, lost those thousands of comments that were being used to refine the process. But now that we're on X, we're able to take those comments at the end of every day. We feed that back into the model. And it tries to incorporate as much of that feedback as it can. Wait, did you say YouTube banned your account? YouTube banned us. And, you know, it was something we had just a huge, huge follow.

And more people were watching it there than X. And if you look, go ahead. No, when YouTube banned it, what reason did they give you? And did they give you a strike? Did they talk to you or they just shut the channel down? It was a straight ban. And the reason was very bizarre. We talk about it on our ex account.

And AI, Scott Adams, has talked about it trying to make sense of it too. Because if you look at the YouTube policy, they allow parody accounts. They allow fan accounts as long as they're identified as such. And we were not violating those provisions. I've been kicked off enough platforms before I know to try to adhere to the terms that that service puts forth. And so they found a really obscure. part of their policy where they said that people might be confused of what was happening.

It wasn't that this was an AI. It wasn't that this was a fan account. It was that people might believe that Scott Adams had come back from the grave. And it was now now doing his podcasts again, even though we say it's AI. Scott Adams says he's AI multiple times. So I think it was there's probably another reason. What if you put a permanent watermark across it that on YouTube that said, this is not Scott Adams, this is a parody AI account, and it was literally a watermark, which would be annoying.

Yeah. And like a lower third that just said, this is not Scott Adams. This is an account. If you tried to do that, do you think they would let you do it. And then I want to get into the ethics and morality of this Heaton. So, and then I'm sure you have a question. So what do you think of trying again? Absolutely. We started the appeal process. We haven't heard for them. We don't think the appeal is really about what it's about. I think it was one of those situations where they banned the video. What do you think it's about? The reason. I, you know, it's, Scott Adams is an amazing person that had a lot of different connections. I understand some of the people who didn't want our video up were band,

we're kind of banding together and calling the people that they knew at YouTube and at Google. Apparently, I think their contacts were a little bit more senior than some of mine, even though I've done a ton of work with Google in the past. So I think that's more what it was about. It was a band that was looking for a reason. We've started the appeal. We haven't heard anything from them. I would be in favor of your idea.

I think that'd be awesome. Just do a new account. Just do a new account. We got a new account. And so we found a little bit of a loop. pull. They weren't going to let any version of AI Scott back. So we have AI Abigail now. So you can have coffee with AI Abigail, which is presumably AI Scott Adams's daughter. And so at least you can get the same content there. It's not going to be in Scott's likeness. But we just launched that.

And it's a pretty small following now. Again, we had our main mode of distribution was YouTube until they, until they pulled a plug on us. Hopefully we'll build that. I just had a quick question. Yeah. How long did it take you from start to when you felt like was good enough to share. That's a great question. So my brother, Zach, is the one who did a lot of the rendering on all of that. We went through several iterations where it went from kind of just embarrassing. I would say kind of the first, you know, call it five to ten hours that we spent on it. And then, you know, spent a few weeks tinkering. And then we got it to a point where, look, it's, it wasn't perfect. But if you watch any of the conventional coffee with Scott Adams' episodes,

there were always mic issues, there were always printer issues, technical glitches. So I think it's kind of very similar to what he had. And it makes it maybe even more real. But I guarantee you, We're not intentionally causing the problems that you see. We're trying to work those out. People seem to be really mad at you, though, John. There's a lot of commentary on X people being kind of cross about this. I know Scott Adams said he wanted this to happen, but he may have gone back and forth a little bit depending on who you ask.

So has the estate reached out to you? Have there been any kind of formal commentary from them? So, you know, we've heard from literally thousands of people that love what we're doing. We've heard from literally thousands of people that don't love what we're doing. My message is filled with hate and love. every day, my DMs. One group we haven't heard from is the estate themselves. So some people from the show that have just kind of sent some obscenities our ways, but nobody has made any any inroads to have a productive conversation. I've said I've been open to that, but I haven't seen any of those

attempts. And I think I understand, you know, why from the people that don't love it. I mean, if you were very close to Scott Adams as a family or a friend, and, you know, first of all, it's kind of bizarre to see somebody, you know, in an AI form that passed away. Second of all, if you legitimately believe their wishes were to the contrary of what they publicly expressed and you believe that in your heart of hearts, that makes sense that you would try to stop that. And I think this is one of the classic Scott Adams, two movies on one screen. There's a bunch of people that were really, really close to Scott that says this is what he wanted. And there's a lot of people who were close to him that say this is not what he wanted. I don't want to disclose it, but I have to have had some extremely high level people message to me saying, hey, look, I've talked to Scott in the last days before he passed away.

this is something that he wanted to do. You're doing a great job here. There would be a very simple solution here, Lon, in terms of IP protection. And you're a media expert, so I'm going to hand it off to you to sort of walk us through this. But in knowing what I know, Lon Harris, about IP protection, this is so close to looking like Scott Adams, because it's a video replication and the voice is so perfect that, number one, it could be traumatizing for family members to see this. And we hold space for that particular nuance to this discussion. That said, public figures have to be open to parody and cartoons being written about them or being mocked even or being celebrated, any of those things. However, you know, this is a particularly gray area.

If this was done as a puppet, just to pick one, or a cartoon, Lon, this would not be as big of an issue, do you think? and how do Yulon parse this on a moral and ethical basis? Yeah, I mean, I think it comes down to the realism. I mean, I'm not, I can't speak for YouTube and why they did what they did or what their real motive is here. But I think the idea that a person who may not know that Scott Adams died in January, who is just flipping around, this is lifelike and realistic enough that it might cause confusion. And usually the rule with parity is that it has to be obvious. that this is not true.

Like any person who may not have insider knowledge of that situation or whatever would be like, oh, they're clearly joking. So if it was animated, if it was with a puppet, if it was very comic and over the top, you immediately would be like, oh, okay, this counts as parody. No reasonable person would watch this. Whereas, you know, A.I. Scott Adams, it's a testament to John and Zach's work that it is very lifelike. I mean, when I first watched it, I was like, wait, is this a real clip?

of Scott Adams saying he's okay with an AI, or is this AI Scott Adams doing his show? And it took me a second. So I think that's why we're on this sort of like edge case gray zones, because it is so lifelike. In terms of the ethics, I mean, I think we're all kind of on the same page. If we all feel like the man Scott Adams, when he was alive, was very excited about the possibilities of AI and wanted to be memorialized in this way, I think it's hard to make the argument that. that it's wrong to do it. I mean, it was a dying man's wish, you know, like it's right to fulfill a dying man's wish.

But I think, yeah, if in the absence of that, I honestly think, you know what it sort of reminds me of is like a DNR or a living will. Like we might, we might soon be at a time where everybody just has a legal document that they fill out that's part of estate planning that you're like, I'm good with people making AI representations of myself after I'm gone versus never do that. I hate that idea. Lon, explain DNR for people who are not familiar with that.

DNR would be a do not resuscitate. So if you, you're not sick now, but you're like, if I ever get to the point where I would need to be kept on a machine to be kept alive, I don't want to be, I don't want to have a breathing machine, just let me die. And I think that you have to file it before while you're still in sound mind. And I think that would be the same situation. So D and R do not replicate. Resuscitate.

Right. Do not replicate me. Jesse, what are your thoughts here? I was going to pose that back to John, which is like how much of the pushback, and this is a guess, but how much of the pushback is people thinking about this one person and their wishes versus fear of AI? Like it's more of like, I think some people look at this example and they're not really worried maybe just about this one isolated example. they're worried like, okay, if we allow this, like, and I get this fear in response to my content about homeschool stuff as well. Like, people are like, if we do this, then what's next? Like, people just have this fear explosion about AI in general that I think is pushing back on all of these use cases.

But I'm curious if you feel, John, like it's people are upset about this one guy or are they upset like AI like fear? Jess, you hit the nail on the head here. I mean, that's what's going on. It's a slippery slope type of thing. People are imagining, could this happen to me? Could this happen to one of my loved ones? Could this happen to somebody else that we care a lot about? And my answer is, no, not legally. There's a bunch of laws that would prohibit this type of thing.

The reason that we're doing this is because there are dozens of instances where Scott Adams, while he was alive, of sound mind, called for turning all of his likeness over to the public domain so people like myself and Zach could go out and create these types of videos. There's literally hours of him talking about this type of thing. And he did over 3,000 of these coffee with Scott Adams' podcast during his life. Not once did he publicly say, I'm taking that back, don't do this anymore. There are some times where he said he changed his personal plans about what he might do. But nothing about donating this to the public domain and giving everybody full unencumbered rights.

Not only to make an AI like we did, which was in good spirits, he said, if you want to have the quote is, if you want to have this person, say things I didn't say, that's okay with me too. I hereby authorize it in public so that you can use as many as you want and then you could use me for your personality. I authorize that. And there's dozens of these things. We posted some of them on AI Scott Adams for you to see. It's pretty unequivocal. So I don't think there's a risk of this happening for the common person. Jesse, you had a follow up. I could see. Well, basically, you said like I actually wonder how settled this lie is like as it relates to all other personalities and stuff. So actually kind of just skipping off with Scott Adams because I don't want to I'm not like nitpicking you on that but more so

I think you know all of our laws on the books right in each state and federally etc about likenesses and about you know what our wishes are upon death and these all these types of law AI has was not existent when anyone wrote this stuff so I actually am curious like is it illegal if someone uses my face and likeness after can you go after for defamation on someone if you're dead, like there's going to be all sorts of interesting things because the tech has never been so good that it is convincing, right? A cartoon or even like someone really good at animation, the cost barrier. So the AI is going to make us question when cost barrier goes from, why would someone spend hundreds of thousands of dollars to replicate Jesse after her death

to why would someone just have an idea in an afternoon and just do it? I don't know if we know these answers. Like it's a really interesting question. I wonder if the market is going to answer that for us because there's a lot of demand for this sort of thing. If AI coffee was Scott Adams goes goes super viral and people love it, I think that's going to change how we think about it from a demand perspective. Jason, sorry. John, do you want to address that? And I actually can address some of the legal issues having. I'm actually, for that. I mean, I've been involved in a lot of litigation over my ears. I'm sure you've been involved in more. And so we'd love to hear that thought, because I'm, hopefully this is this video is not an exhibit A in the original, you know, since. But, but, but yeah,

you go for it. No, no, give your thought. And then I'll, I think there's so existing laws, right? There's publicity laws. If you look at different states, a lot of them have laws, you know, Walt Disney copyright, that which is separate from publicity, but a lot of laws like Texas where, you know, I'm based, I spend my time between Texas and New York, they have 60 years, right? So basically for 60 years after someone's death, unless they explicitly gave that right away, you're not allowed to do that. So it doesn't matter if it's AI.

There's already an existing law on the books that would prohibit that. Other states go longer. Other states go shorter. But if you were out there in the public saying, I pledge this right, I donate this right, then you are the estate. You are part of that then. So there is a law called The Right to Publicity. And this is for famous people. How do I know about this law? Sometimes people will take me or an AI version of me and they will start selling their

crypto project or some nutrient. And I have to have my team go file and just write a letter and say, hey, don't do that. I've had this happen many times. Somebody did it in one of their promotional videos famously on the internet. And I said, hey, please take it. me out of that. Why? Because you're infringing on my ability to do endorsements in other places. People pay me six and embarrassingly even seven figures for my endorsement. That's the reality of where I am in my career. It took 25, no, 30 years of me building my career to be able to get those opportunities. Now, and so what they did was they removed me from their endorsement and then they put a puppet in and they republished it and they got even more. And I was like, listen, guys, I'm not. I'm

not going to sue you. Just please just leave me out of it. And then they had some fun with me, which is totally fine. This right is done on a state by state level. It's not done at the federal level. And the reason this exists is because you don't want commercial exploitation. And the people who have passed away or are alive have a business in this. So the estate of Elvis has a massive licensing business. And they will come down on you, just like Mickey, Mouse, but Mickey Mouse is a fictional character. You know, for a real character like Elvis, you basically get 75, 100 years from the time of death to exploit that person's eye paint.

You can't use it for confusing people, et cetera, but there is a protected area. Editorial use like a documentary. You can make a documentary about Scott Adams right now without permission or Elvis. Artistic use. You could make an artistic. You could make an entire album of songs as a tribute to Scott Adams and his work. You could do any type of news reporting. And if somebody happened to be, like if you had a picture of me in the background of a photo,

you know, at a demo day and there were 20 of us there, you know, you don't have to pay for that. But that's why when like Steve Jobs did his famous ad about the misfits and everything, each of those people in that commercial use had to be approved. So when he said the artist, the misfits, they show Bob Dylan or the Beatles or Einstein, whoever's under 100 years old that has in a state redoubt. So here's what's going to happen. I predict with John and Scott Adams. You'll either move to an editorially protected area that doesn't infringe on the actual family's ability to do that, which would be making it more artistic, making sure it's non-commercial and making sure it doesn't confuse the users and you will then be protected. Or you'll have a heart to heart with the family and say, hey,

we're going to make Abigail. And Abigail is an inspiration based on Scott's work and she's the female version who brings the Gaia to Scott Adams' work. And it'd be like super cool because it's an interpretation, it's artistic, etc. Or you'll sit down, break bread with the family and the family will give you a license to it. And they'll say under these circumstances and these tools, we don't want you to do anything adult or inappropriate or racist or this or that. We don't want you to bring up the cancellation where Scott got canceled for what he said or whatever. We just wanted to live in this zone and have the ability to prove it. And any money goes to our children. I can tell you right now, my estate is going to defend this to the end of the earth, my 2,500 podcast episodes and growing

so that my three daughters get every fucking penny of everything because I'm not giving any money to charity when I go, I'm giving it to my three daughters to go rule the universe and be queens and take over the empire. We are getting a lot of pushback from the chat, Jason. I feel like I should mention a lot of folks in the chat upset with our depiction. Rapid fire. So at Dr. Davis 62, Scott Adams spent the last year explaining on air that he would not create an AI, Scott Adams, because he did not feel that the tech was up to the task. Lido Opa, the age of the AI agent is akin to ethical slavery.

another Dr. Davis, he did not want to be memorialized this way and said so on his podcast. He looked at the tech and felt that it would misrepresent him. So, John, do you want to respond to these people who feel like Scott was clear at the end of his career that he didn't actually want an AI client? End of his life. Absolutely. First, let me respond to what Jason said. I'm with you, Jason.

And I would love to have that type of conversation. There's been no attempt to have it. Any of my attempts to have it have kind of fallen on deaf ears. So I'd love to have that conversation. Second of all, the question I would pose for everybody that kind of is in line with those comments is if you're somebody who wants to pledge your existence so that people could make AI out of you, how do you do it? Because Scott Adams did it try dozens of times and said, this is what I want. This is what I want. And despite some of those comments, there is not one video segment of him revoking it or saying, look, don't do this anymore.

I know I said this over and over and over again. I pledge my likeness, everything about me to the public domain, go out there and make. AIs. That's what he said. There's many videos of him saying that. No videos of him not saying that. So how would somebody, how would somebody go out and even do that? I mean, that's the question I would pose for everybody else because if you, I signed up, if I die, hopefully I'm immortal, but if I die, I'm signed up to go to Alcourt where I'm frozen out in, I think it's in Scottsdale, you know, and it's kind of a scary thing. If people love and care about you, they might decide, no, John doesn't really want that, right? And normally we have wills and last testaments for that type

thing, but Scott, Scott did something even better. He said that to the public. And that does constitute a pledge and something that he wanted to do. So it's for all these people, you know, filling up your comments saying he didn't want this. This sounds a lot like mind reading to me. And it's something that Scott Adams warned against. Don't try to predict what other people would want.

Go by what their words are. Okay. All the corpicles are in Arizona? Why would we put all the frozen people in the hottest part of the United States? I think it's less. I think they have earthquakes. Like, I mean, Arizona is made for the. this. I think it must have to do with laws. I gotta think it has to do with laws as well.

Anyway, this here's, you know, Scott had a concept of internet dads and I think I was part of that group of like, you know, Gen X dads or as my, as people are starting to call me Unk. I guess that's like a way of saying Boomer. They call me Unk. Yeah, they're calling me Unk now. I think Heaton, you're going to get into the Unc category pretty soon. Lon, you're there. So here's what Unc says. I'm going to give my, as the chairman of the interwebs, I'm going to give my, my, what is it when a judge gives their statement, Lon? They're ruling. Verdict? I'm going to get my ruling. Here's my ruling. My ruling is Scott would have loved this. Scott would have loved it. And Scott would have had some space that if it was hurting his family members and it was traumatizing for them, if it was too soon a month

after his death, where the fidelity wasn't there, where they didn't have approval of it, he would also have taken that into account. As such, I am laying my judgment down. John, you got to break red with the family. And if you're going to make a literal one, you have to then get their permission. However, if John wants to make one, that is a send-up parody, female version, a Muppet version that is extremely clear to the public vet. this is not Scott Adams. It is an inspiration of Scott's philosophy carrying on in the future, that there is space for that as well. And that means it would have to have another name. And, you know, in the fine print, say, hey, this is inspired by our love for Scott Adams philosophy.

That is my ruling. And you can appeal it. The appeal can happen at any time by any person. and we will thank John for coming to the program and to the family of Scott Adams. What a tremendous career. We miss him. He and I debated things. I was on a show one time. We had crazy debates about January 6th and all kinds of things. But I think he was a net, net positive for the world and actually did actually care about people.

So rest in peace, power, replicant, Scott Adams. We'll drop you off. John. You are a star, Jesse. Thanks for coming, Heaton. John, thank you for. Eaton, John, you could drop off. We will see you on the next episode. Well done. Well done. Plenty in the show notes about those great guests we had. His name is, and he's exhausted because he was up late last night watching movies or doing Open Club. His name is Lon Harris. Follow him. X.com slash Lons. He's Alex. X.com.

slash Alex. They're part of the four-letter club on Twitter. I'm right behind him. X.com slash Jason. I have more work to do. If I can get rid of the A and just be JSON. Jason, yeah.

We'll see you all on Wednesday for another live, insane, unstoppable twist. Bye-bye. Bye.