5 New AI Models That Are Smarter (and Cheaper) Than GPT-5

Hey, welcome to the Next Wave podcast. I am Matt Wolf and I am here once again with Joe Fear. And we're gonna do what we did last week. We're gonna break down the stories that we think are the most interesting, that we think you should know about, talk about why you should know about them and be thinking about them. And we're also gonna break down a really cool demo of how to use some of these AI tools in your own business. And how do you put your products into AI and sell your stuff better with AI? We've got some cool demos and tips that we're going to share throughout this episode around that as well. So without further ado, let's just get straight to it.

Cutting your sales cycle in half sounds pretty impossible, but that's exactly what Sandler training did with HubSpot. They use Breeze, HubSpot's AI tools to tailor every customer interaction without losing their personal touch. And the results were incredible. Click through rates jumped 25%, qualified leads quadrupled, and people spent three times longer on their landing pages. Go to HubSpot.com to see how Breeze can help your business grow.

How you doing again today, Joe?

I'm doing great, man. I'm happy we're back. And I feel like this week was packed with all sorts of different announcements. Obviously, we got a bunch here. I feel like we're going to work in, like, topics, so we'll probably be talking about, I don't know, a lot of news. But last week was great. If you like that kind of flow, it's just going to get better.

Yeah, yeah, yeah. So again, like you said, a lot of stuff happened this week. We've got brand new models that came out from a lot of the foundation labs. In fact, the only one we didn't really hear from this week much was OpenAI, which was interesting.

They had some other news, though, so we'll definitely be talking about them too.

Yeah, yeah, let's just jump straight into that one.

Let's do it.

So the big story out of OpenAI this week was that the OpenClaw founder, his name is Peter Steinberger, went to OpenAI. He developed OpenClaw. He was the sort of lead person on that OpenClaw open source project. If you're not familiar with OpenClaw, then you probably haven't been paying too close of attention to the AI world. Here's the tldr, though. It was originally called Claude Bottom. Then Anthropic went to them and said, hey, you can't call it that. It sounds Too much like Claude. So they went, okay, fine, they changed it to Multbot for like a day. And then the next day they changed it again to openclaw. And what this was was a AI agent that you can run on your computer, you can run it on a virtual server, but you can chat with it from wherever you like to send Messages. So Telegram, WhatsApp, Slack, Discord, any of those tools where you do messaging, you can actually have conversations with your AI chatbot. And it would go off and do actions on your behalf. And you can give it as much or as little access as you want. You can give it access to your emails and your calendars and your passwords and your API keys and whatever you want. And depending on the level of, you know, autonomy you want it to have, it can go off and do different actions on your behalf. And that thing just blew up like it was the talk of the AI world, still kind of is to this day. And the guy who created it, this Peter Steinberger, well, he grabbed the attention of a lot of companies and ended up going to OpenAI.

I thought it was interesting cause he came from a previous PDF company. I think it was like a decade ago or something.

Yeah, yeah. So he created a company called pspdf. And I think the PS stood for Peter Steinberger, if I had to guess. He created this like, PDF tool that is actually now used by Dropbox and a whole bunch of other big platforms use this sort of infrastructure that he built to manage PDFs.

Yeah.

And then he ended up getting like $100 million investment and then cashing out and then sort of going and becoming a hermit for a while before coming and stepping back into the limelight to develop Open Claw.

Yeah, I thought that was interesting. And when it came back to the world of AI, like in 2025, so like last year, but it started developing like all these different projects, and then Open Claw ended up being the one that got the most attention for good reason. I think people are even talking about, like, could this be the first one person, billion dollar company? And I think he mentioned something like, you know, I've done the whole, like, growing a company thing myself. I'm good.

Yeah, yeah.

Like, I want to go partner up with someone awesome. And I guess in his eyes.

Yeah, yeah. Well, I mean, all of the products that he built between when he sort of came back online to now were all sort of infrastructure pieces towards Open Claw. In fact, he actually put something out on X pretty recently, saying it's kind of funny, there's this Storyline that I had 40 failures before I finally made Open Claw. And he's like, that's not true. All of these other tools I built were the building blocks that became Open Claw makes. So like that narrative is just like a made up narrative for marketing. That's not the real narrative.

It's good story.

Yeah, it makes for a good story, but yeah. So this Open Cloth thing, it's interesting because he wasn't really making money off of it. In fact, he was saying he was paying out of pocket to keep this thing alive and going and he was sort of relying on donations through GitHub to sort of keep paying it and to pay some of the other workers that were working on this thing. So, like, I don't think he had any intention of turning this into $1 billion business.

No, I don't think so either. That was just the chatter out there I saw.

Look, we know AI moves fast. So we created your weekly cheat sheet on what's happening this week in AI with over 20 prompts for the newest tools that you can copy and paste a quick reference table that shows you exactly which model to use and tactical breakdowns of the newest AI releases. Click the link in the description. Now let's get back to the show.

But it was interesting. So he picked OpenAI. I think he was chatting with a couple of the other labs as well. Which is interesting because it started off with Claude behind the whole thing, basically powering this and they don't have him anymore. So that dude is now working with their biggest competitor, OpenAI.

Here's the thing. So when it comes to OpenClaw, everybody that was using OpenClaw in the early days, they were going and using their Claude Max plans, right? So Claude has a plan called the Clod Max plan. They have a hundred dollar plan and they have a $200 a month plan. And Claude Max basically lets you use like Claude code at almost like an unlimited level. Like you do have some limits, but you could get a lot of usage out of it for your hundred or $200 a month plan. And people found workarounds to use their Claude Max plan in tools like Open Claw. Anthropic went and said, no, Claude Max plan is designed for people who want to use Claude code specifically and use our tooling. So if you're using your Claude Max plan outside of our tooling, that's against our terms of service. And they started like essentially threatening people for using their $200 a month plan inside of OpenClaw, right? And so that's kind of what got people going, oh, so they don't even want us to pay their high end $200 a month plan if we're going to go use it somewhere else. Okay, well, we'll go to OpenAI and then OpenAI said, hey, feel free to use our $200 a month plan inside of Open Cloud. We welcome it. Right. So OpenAI went and had this narrative of like, Claude's going to be the company that's going to shut you down. If you're on their high end plan, we're going to be the company that's going to say, yeah, come to us, we'll take you.

Open arms.

So that is a big, big piece of why he ended up going to OpenAI. I don't know if Anthropic ever actually approached him about bringing him in. I know Meta did. He said he had a conversation with Zuckerberg and Zuckerberg tried to bring him in, but he ultimately decided to go to OpenAI because he felt that OpenAI would a keep Open Claw open source. He didn't want OpenAI to come in and say, you have to use OpenAI models inside of OpenClaw if you want to use OpenClaw. He wanted to keep it open source. Your choice. You get to use this however you want. OpenAI basically allowed that. So now what it sounds like is happening is he's going to OpenAI. He's finding somebody else to lead the development of Openclaw. Openclaw is going to continue sort of on its own within like a foundation away from Peter. Peter's going over to OpenAI to start developing their like more agentic systems at OpenAI.

Ah, interesting. Yeah, that's what I just read here is Open Claw will be a project under an independent foundation. Okay, yeah, I didn't know that.

That's what I'm here for, Joe.

Claude just validified that from an article that came out two hours ago. So you nailed it.

Yeah, so that's pretty much the storyline there. He's going to be building out whatever OpenAI's next version of an agent is going to be. And Anthropic really, really did kind of fumble the ball here. Right. Because a, it was called Clodbot and when they named it claudebot, everybody just sort of immediately associated it with Anthropics Claude. Right. And so everybody already had this sort of like tie in association to Anthropic. It brought more awareness to Claude. People were actually signing up to Anthropic's $200 a month max plan. So they could go use it inside of claudebot. So Anthropic was getting more customers from it. And then when Anthropic did their whole like, hey, you need to stop using this name. And he decided to go to OpenAI, a lot of people went, eh, screw Anthropic.

Then, yeah, I think they dropped the ball. I mean, especially with the momentum. And what does it say, like, on GitHub, it has 100,000 plus stars. I think it's like the biggest growing project ever on GitHub. So, yeah, it's just like the fact that like, dude, you would have had all that momentum. But now, I mean, we were talking last week about how Open AI with their new monetization model they put out there. We're like, what are they going to do? I was like, well, shoot, maybe it's agents, you know, and now you get the attention of all these people that are also paying attention to Open Claw and Peter. So I don't know, maybe that just helps shift stuff a little bit for them.

Yeah, well, I mean, I feel like we're still early in the year, right? Like, it's, it's still February, it's still only the second month of this year, and it's. This year's already shaping up to be the year where, okay, agents are finally feeling useful, right? Like last year, agents was a big buzzword, right? 2025, we did have that as like everybody was saying, oh, agents are here, we're using them. That's a big buzz. But an agent was essentially like, oh, look, it can click things on my browser for me, right? Like, look, my browser is now highlighted and it can move my mouse and type things on my keyboard without me having to move my mouse and type things on my keyboard. That's cool. It's doing it really slow, like slower than if I was just to do the damn thing myself. But I mean, I could open up like four tabs and have it doing four different things very slowly for me, right? Cloud Bot is sort of the first thing. And I'm going to keep calling it Cloud Bot, even though I know it's called Open Claw now. My brain hasn't caught up yet. It's like Twitter X, I still call it Twitter. So with Cloud Bot, this was like the first one that came along where you can just go and use whatever chat platform you like. If you, you know, if you use slack on a daily basis, you can use Slack and treat it just like a team member. Hey, I need this done. Five minutes later, your team member comes back and Says, cool, I did it for you. Right. That's totally different than you seeing your browser glow and then the mouse is moving around for you and doing things like that's just trying to get your computer to do something that a human can do already, where something like Claudebot is tapping into the APIs. It's building its own software behind the scenes to do things. It's basically handling a whole bunch of processes behind the scenes to make sure your outcome happens instead of like doing what a human would do to get that outcome to happen, you know?

Sure. I love following people on Twitter X. There's a lady that just came across. I think it was retweeted by. I don't know who was someone, some tech person. But like, she homeschools her kids and she's like this open claw, like fanatic and is awesome. I forget her name. But basically she's like, figured out how to train her. I don't know, she gave it a name. Everybody gives their Open Claw a name, it seems like. And it's now like printing stuff on a 3D printer from whatever she told it to do with the school curriculum. So it's like. I'm like, holy moly. And it's doing it by itself without any human intervention in the middle.

Yeah, well, I mean, theoretically there's sites out there that have 3D print templates already, right? Yeah, you've got printables and thingiverse and all sorts of sites where you can find anything to 3D Print. These tools could go in, search those sites, find the thing you need, download the STL file, send it to your printer for you, and print it out. You know, theoretically, we tried this last week. Theoretically you could ask tools like Gemini to go and create you a print file and it can send that to your printer and automatically print it out as well. Theoretically, we didn't have so much luck with that, but theoretically you can.

Sure, it'll change. But yeah, like, I don't know, I think of the different use cases. I mean, your buddy Matthew Berman seems to be like, way down the rabbit hole when it comes to the open claw.

Oh, dude, his videos are awesome. I mean, they're pretty in the weeds, but I've watched two or three videos where he's breaking down, like, these insane systems where, like every night while he's asleep, it's going and doing a bunch of research and finding talking points for his videos and making thumbnails for him and, you know, all sorts of just stuff. And then he wakes up in the morning and he's just got like his messenger app full with like, here's everything I did for you while you were asleep.

So I don't know, I see this and immediately, you know, I thought of the whole Agents are now coming to everybody. You know, like, everybody is going to be able to at least start interacting with agents. Seems like that's what OpenAI is pushing towards. It's still very early there. But then you have what Meta was it just this week, I believe, basically talking about how Manus, who they acquired is now it's all being usable on Telegram and essentially you can kind of do something similar. And I'm curious of your take here, Matt, of all the details. Of course.

Yeah. So this is Meta's play in response to claudebot. That's what it feels like. I don't know for sure. Like, the mana steel was probably in the works before the whole Clodbot stuff bubbled up. Had to have been, because it all seemed to happen a little too fast. But, you know, the way this is all angled and the way they're marketing, it feels very similar to what we're getting out of Clodbot. But yeah, basically the same thing. Like you can use Manus agents inside of Telegram, which I think is hilarious because Meta owns WhatsApp.

That is a good point. I didn't.

Okay. But you know, I actually think it works in WhatsApp as well, I believe. So I think you can use it like in any of the messaging apps. It's just interesting to me that like, the headline tool that they mention is Telegram. Telegram is the first supported channel. Maybe it's just Telegram right now and WhatsApp and the rest are coming. But it is kind of funny that it's like a Meta competitor that is one of the first ones that is rolling out into. Yeah, it's probably like the easiest API to deal with or something.

I don't know. There's had to be some relationship and they're just like, we got to rush to market now. Like, this is our chance.

Yeah. But I mean, it basically seems like this similar idea, but I don't think it's any sort of level close to what you can do with Open Claw yet.

Doesn't seem that way.

Like a lot of what Manus is still doing is a lot of like taking control of a browser browser and doing things the way a human would do them instead of just like operating directly in the back end via APIs and stuff.

Yeah, but it's cool to see this. And of course, when it's Kind of like the model race. Right? Which I know we'll probably talk about that here too. But it's like when all this stuff happens, you know, you have the other guy is going to bring something to market, even if it might not be the same or perfect or it's just support of one channel. It. It's just like now it's becoming a thing in all of our minds. And yeah, I don't know, we've been told, like you said, like, the agents, they've been around, but I feel like this is the year of the agents.

Well, this is the year that the agents actually start to feel useful. This is the year where an agent feels like an employee and not like a gimmick. Right. I feel like last year agents were very gimmicky. They were very take control of my browser and ooh, I can sit here and watch it do the things I would be doing anyway. It's just doing them a little bit slower. But I could open multiple tabs and have it do multiple of them, but now it's actually like, they're useful. Like, I can go away and have things happening for me while I'm away. That's what makes this current sort of iteration of agents useful.

Yeah.

They also are much, much better at checking their own work. That was kind of an issue with past agents is that they can go and act on your behalf, but they would screw stuff up and not realize they screwed stuff up. Current iterations of these agents, they can go and do some work. Look at the browser go, oh, that's not the way it should look. Go back, fix its own mistakes, look at the browser again and go, okay, now I got it right. I'm going to report back to my owner. Whatever you want to call the humans in the loop, I'm going to go report back to my owner and say that I've finished this now that I've double checked my work. That whole double checking your work thing is a big deal. Right. Because if you've ever tried to like vibe code stuff, that's the kind of headache you run into is that it'll say, oh, I fixed that. You go and open your browser and it's like, it's still not working. Fix it better. It's like, okay, well, I fixed it. Now it should definitely work now. You open your browser, still not working. Yeah. What the heck? You just said you fixed it. It's not working. Well, less of that is happening now. If you've ever used like the newer models like Opus 4.6, Claude, 5.3 codecs, things like that, they actually will look at their work and go, oh, it's throwing up an error, let me fix that. Oh, it's still throwing up an error. Let me fix that. Oh, now it's working. Okay, I'm going to report back and say I'm done with my task. They're actually double checking themselves now and that is a huge leap from what we had this time last year.

Right. Interesting. I saw Mark Cuban, I don't know if you saw this, and he brought up an interesting point. And you might know the math and the things more than I do, but it's an interesting little segue here where he basically put something out there. He was like, hey, so what happens? And it's a response to the all in podcast. But what happens when AI tokens cost more than your employees?

I did see this. Basically it's getting to a point now where you might be spending $300 a day on AI tokens, but $300 a day is more than what you probably would have paid an employee to do the same level of work. Right?

Yeah. And that's kind of the argument. And obviously he goes into the math and all that, but I thought, I've not totally thought about that before, but it's like a lot of people are so hyped up on agents and automations, but then we might not be doing the math on what it's going to take to actually complete the job effectively.

One thing that's like a key point though is he says in the short term. True, that is a very key point there in the short term. So is the aggregate token cost to do what an employee does plus fully encumbered developer and maintenance cost divided by fully encumbered employee cost, equal, less than or equal to productivity? If it takes eight clot agents at $300 for tokens per day plus $200 per day in dev slash maintenance to do what an employee does per day at a fully encumbered cost of $1200. That's $2600 divided by 1200. But then you need to factor in the productivity rate. Is it more than 2.16 productive? Are there qualitative issues like morale, morality, whatever, that can't be quantified that need to go into this decision? Yeah. So it's interesting. I have been seeing that more and more. I do a lot of vibe coding stuff myself. I code up a lot of things and I've definitely had days where I've spent, you know, $100 in tokens and it's like really added up quickly. I also think that is a very short term problem. We're gonna get into all of these model releases here and one of the big things about these model releases is that they're getting better and better and better and also getting cheaper and cheaper and cheaper. So right now the state of the art model for coding is Probably Cloud Opus 4.6, maybe ChatGPT 5.3, Codex. Right. It's kind of a toss up between those two. They're both really good coding models. However, Sonnet4.6 just came out and Sonnet 4.6 is doing almost Opus level coding. It's almost as good for like a third of the price. We'll get into the prices in a minute. I don't know exactly what they are, but the cost is like quite a bit less. But they're still getting opus 4.6 level quality out of it. So these costs are going down very, very quickly. So this is only a problem, like right now.

So it's a short term, very short term. Yeah. Interesting thought process, I guess. But yeah, I want to bring that up just because I was like, oh, it's interesting. But to your point, it's like these models, which is what we'll talk about. It's like there's so many getting pushed out all the time and it's wild.

Yeah, yeah. A couple quick things on this whole agent topic. Meta launches Manus AI Integration and Ads Manager. I know we've got like a large business audience for this podcast. Well, pretty soon you're going to be able to have. Pretty soon, might even be right now.

I think it's right now.

Meta has taken the next steps in its integration of Manus AI, with some Meta advertisers now being informed that they can access Manus functionality within the Ads Manager. So it's kind of like a rollout. It doesn't sound like everybody has it yet, right. But it sounds like you can go and tell like your Manus agent, go build me an ad for my product. Right? Go build me an ad for this cereal brand. And it will just go and like look at your site for you, learn everything about your product, probably pull screenshots for you of what the product looks like, use sort of copywriting principles on your behalf and then just go and create your ad in the Ads Manager without you ever even touching the Ads Manager you just prompted in telegram and not WhatsApp.

And probably keep an eye on the stats as well because it's all integrated in Meta's, you know, in their ad center there and then just make more creatives and. Yeah, cause that's the whole thing that I'm seeing from, I mean all these creatives. And I know you and I have run ads for a ton of years now and it's changed, but now it's almost like trust the AI, they want more creatives in general and then their AI engine, you know, metas will kind of do the hard labor now. Whereas back in the day it was the humans dialing the knobs, I. E. You, Matt, at least.

Harold, that is a very good point. Right. Like when we used to run a lot of Facebook ads ourselves, we had to run ads and then, you know, every day we were in that ads manager making sure like our cost per clicks weren't going over a certain level. And if this ad was underperforming and this one was overperforming, adjust the budget. So we're putting more money towards this ad and less money towards this one or turning this ad off altogether. And it was like this constant like course correction that you have to do to keep your ads on track. And if you have a bot that doesn't sleep right, it could keep your ads on track 247 without you even looking into it. It's just, it knows your criteria that you want it to stay within and it's constantly managing that. Now do you actually need an agent to do that? Not really. I feel like that should just be something that the ads platform does for you without the need of an agent.

It's taken some of the friction down, you know, it could maybe take some of the creat of creation out of the way. But even though I'm sure that might get even more integrated into the platform, you know, over time.

Yeah, I mean I, I feel like what, what an agent is going to probably be best at in the, in terms of advertising would be going and looking at your website, reading like your sales copy on your website, learning what your product is all about, then deciding what the best imagery that's going to resonate with the most people is going to be, deciding what the best copy that's gonna resonate with the most people is going to be and then creating those ads for you. I would imagine you don't even need AI agents or AI at all really to keep your ads on track. Right. You should be able to go in there and say, hey, I want to generate leads at $2 or less per lead. If anything goes over that, turn it off. Right. Like that's just if this, then that statement in the back end, that's just like regular logic. That's even like AI necessary.

Yeah, yeah, yeah. So I don't know, we'll see how this rolls out. But again, this what's got pushed out on the 16th. So it was, you know, earlier this week. It just seems like, okay, this whole agent thing, you're going to start seeing this, I feel like integrated in all the different platforms in some way to kind of have their own flavor.

Yeah, no, it is interesting. So there was this big wave during, let's call it 2023. The year of 2023 was the year of every company put AI into their product, whether the product needed AI or not.

We're an AI first company now.

It's like every company did that, right? Hey, look, we're a company that does SEO services, but now with AI, we're a company that does, you name it, bookkeeping services. But now with AI, like every product just now has AI. And I feel like this year we might start seeing that same kind of trend. But with AI agents, right? Like, I don't really feel like Facebook ads should need an agent to manage your ads. They already have recommendation algorithms. They already have the if this, then that sort of logic algorithms. Like why does that need AI? I don't know. Why do I need to talk to an agent to make that happen? I don't know. Shouldn't I just be able to go in the back end and say here's the cost per lead I'm trying to get, go try to get that for me. And then it just goes and does it? Do I need an agent to do that for me? I don't know, top of my head, I don't really feel like you do.

But yeah, I know it's too much of a good thing. Maybe not necessary.

And then last thing, we'll talk about this real quick. 11 agents for support. I did briefly come across this one, but I haven't dug in too much. My guess without reading the website, is that because it's from 11 labs, it's customer support that you can like have it use your own voice or use one of their like pre existing, like realistic sounding voices to do customer support for you.

Right, right. That's what it seems like. And they're claiming, what does it have here that you know, you can load up things like your SOPs and then you can probably, you know, loop in your agents, maybe specified agents. Depending on what's going on. It improves with interactions, guardrails. I thought it was Interesting that it's emotionally and contextually aware of situations further down on the screen.

It's emotionally and contextually aware. So if you ever get on the phone with an 11 Labs agent for customer support, just start screaming at them. Just like rail at them, right? And they'll probably give you what you want because they're emotionally intelligent and can tell that you're pissed off. We will be right back, but first let me tell you about another podcast, DTC Pod. Hosted by Ramon Barrios and Blaine Bolas. The DTC Podcast is a podcast about all things direct to consumer. Ramon and Blaine cover everything for starting, growing and optimizing e commerce stores and DTC or D2C brands. They talk with founders, marketers, platforms, creators and marketing and growth agencies to cover topics like brand building, social media, paid media, email marketing, and more. If you're interested in the stories behind your favorite consumer brands, this podcast is for you. They recently did an episode about what every founder needs to know about compliance, claims and scale. Listen to DTC Pod wherever you get your podcasts.

Click into that. Let's see what it sounds like.

Hey there, I'm Jennifer with eleven Airlines. And how can I help you today, Jennifer?

My flight just got canceled and I'm

stuck here in Orlando.

At this rate, I'm going to be missing my daughter's birthday. This is ridiculous.

Yeah, I hear you and I'm so sorry about that. Let's figure out what's going on. Okay, could you please tell me which flight this was?

Yeah, this is flight MD412.

Great, thanks. Okay, just pulling that up now. Okay, yeah, I see it. Three cancellations on one trip. That's. Yeah, that's not the experience you should be having. I'm going to fix this. Okay, Let me just see what I've got.

Well, realistically, what kind of options do

you have for me?

Okay, so here's where we're at. I can book you on the same flight route tonight, but I'm going to be honest with you, that routes been a mess all week and I wouldn't trust it. Yeah, not for something like this. The other option is I reroute you through Charlotte first thing tomorrow morning. You'd land by noon. It's a longer day, but you're getting home.

Just give me the one that's actually

gonna get me there.

Will do. And yeah, again, I'm so sorry and completely get the frustration. But yeah, I think we can figure this out for you.

It's always Florida too. It's always Florida.

Okay, just locking that in for you.

This makes so much sense. I feel this guy's pain. I mean, it sounded pretty good, I thought.

Yeah, it does. It does. It sounds pretty realistic.

Yeah.

I just wonder how it would have been different emotionally if he was, like, screaming.

I know. I don't see an active way we can actually try this, but that would be fun.

I think that's interesting. I don't know. I have mixed thoughts about it. I don't know if you remember this story. This is probably almost a year and a half, two years ago, but there was a story, I believe it came out of Chevy, where they set up, like, an AI assistant on their website. It wasn't one that actually vocally spoke to you, but it was like a chatbot on their website, and somebody actually talked the agent into giving them, like, a Chevy Tahoe for free.

They're like, you said it.

So if you say it in the right way with the right sort of thing, like, oh, your boss said it was okay. I talked to him earlier. Like, are you going to be able to, like, convince these AI agents to, like, give you free stuff?

Interesting. I mean, it's worth a try.

Yeah. I'm sure they're trying to build guardrails in about with that kind of stuff, but I guarantee we're going to hear at least a couple of stories out of some of these support bots where people got them to, like, give them free cars and free flights and stuff like that, just because they used a certain word pattern that managed to get around some of the guardrails, you know, that'll be.

Yeah, testing the guardrails. I mean, like, it claims they have the guardrails in there, but. Or you have to maybe put them yourself. There's definitely going to be a pastime of people that are probably going to be testing guardrails. We'll just say, just for that reason alone.

Yep. All right, what do we got next up on the list here?

Well, Mr. Mateo, we mentioned how there've been all these releases of different models. Like, there's kind of like. I feel like a confusion with most people now. I don't know how many were released. Was it like 5, 6, 10? I don't know.

So, yeah, this week alone, we got Claude Sonnet 4.6, we got Gemini 3.1 Pro, and we got Grok420 or Grok 4.2, if you want to be lame about it.

Geez. Okay.

Yeah, these are the models that came out from the big sort of US Frontier Labs. We did also get a couple other Models out of the Chinese labs, we've got ByteDance, who is the creator of TikTok and Capcut. And you know, among other things, they released a new model called seed 2.0, which they claim to be fairly state of the art. And then we also got a model called Quinn with a whole bunch of numbers after it. I don't even remember what it was called, but that's an open source model that is almost catching up with the frontier models. So those are the other two that came out outside of the three that I just mentioned, but the three that I just mentioned, Claude Sonnet 4.6, that one came out first, 4.20 came out the same day. And then Gemini 3.1 came out on the 19th, which was Thursday.

Yeah, I found it interesting because there's been so much of this happening in obviously short period of time. I feel like it happens all the time. Right. It seems like the biggest single month. At least this is what I read. The biggest single month, model releases and AI history.

Quite possibly it's busy. I mean, I don't know exactly, but there have been weeks where it's like, hey, Google just released something, and the next day Anthropic just released something and the next day OpenAI just released something. And I do remember weeks that feel equally as hectic, but it does feel like a lot of models came out this week.

So my brain goes to what the heck does this mean for us, the users? Joe Schmo, Users like me. Hi, I'm Joe Schmo here and I'm not Matt Wolf for sure because you're tapping, but like, but when it comes to users, I don't know, I feel like the models are blending in. Like people have their favorites, you know, like OpenAI did not make the list like we talked about earlier. So, like.

Yeah, yeah, well, because they did GPT 5.3 Codex last week.

Yeah.

They actually released two models two weeks in a row, GPT 5.3 Codex, and then the following week, GPT 5.3 Codex Spark. I get exhausted trying to say these names.

I don't know. Yeah.

So I mean, what does this mean to the everyday normal consumer that just wants to use ChatGPT or Claude to ask it where to go to dinner or whatever? Not a whole lot, right. These models, they're getting better, they're getting smarter, but they're getting better and smarter in areas where the majority of people probably aren't going to notice or care that much, saying that they are also getting better, faster, cheaper, smarter. Right. They're getting better. So that has the benefit of bringing down costs over time. Clodsonic4 6 One of the big deals about this one is it's got that 1 million token context window which basically means you can feed it so much data, right? You can feed it like the entire Harry Potter book series and it's going to be able to read all of it and answer questions. Right?

Okay. So you're not going to get cut off if you're trying to shove a whole bunch of, I don't know, like I'm just thinking my use cases, it'll be like call transcriptions or all these other nodes.

Unless your call was, you know, a month long call that didn't stop. No, you're probably not going to get cut off on any sort of calls that you're going to have.

Nice, I like it.

But saying that the 1 million token context window is more for the API users, there's still limits. If you're just using the regular Claude AI website, you're not going to get that million token context window on their website. You're going to get it if you're using Claude via the API and sort of using it for development. So it's cool that we've got that. But your everyday Joe Schmo as you call them, I'm just calling myself, he's not going to really see the benefit of that. But Sonnet 4.6 is a lot better at coding. But it is the new model that's inside of the free plan, the $20 a month pro plan, all of the lower end plans, this is now the default model. Benchmarks don't really mean too much to too many people. But really what I look at em for is like the comparison. So you can see that when it comes to like agentic coding, Opus 4.6 is still the best. But Sonic 4.6 is still really, really good. Right? Opus 4.5 and 4.6 are like 80.8, 80.9%. This is 79.6%. Like it's basically as good. But this is a lot cheaper to use model.

You were saying this earlier how it's like, yeah, well now this new model, Sonic 4 point, even though it's not technically the best, now you're seeing how close it really is to the best and it's going to be dramatically cheaper like you're mentioning.

Yeah, exactly. And here's the thing where this one actually starts to like stand out is that when it comes to actually like agentic use cases like going and doing work on your Behalf, it's like as good as the state of the art model here. So like a finance agent, it actually scores better than their 4.6. For office tasks, it actually scores better than 4.6. So like this model is almost designed for something like claudebot, which is where anthropic just totally effed themselves. Right. Because like this model is probably the best model to be running in something like Claudebot because it's a lot cheaper to use than Opus 4.6, but it's also really good at agentic stuff.

The thing we're talking about here.

Yeah, yeah. So it's kind of crazy here, but one of the big things about this is like in the API, if we look at their pricing here, Claude, Opus 4.6 is $5 per million token input, $25 per output tokens. Right. We just saw that Sonnet 4.6 is pretty much almost as good in most use cases, except for input tokens, it's $3 per million tokens instead of 5. And for output tokens it's 15 instead of 25. This is like the big deal here is that you have a model that's almost as good as the most state of the art model, but it's a lot cheaper to use. So developers are able to use this model a lot cheaper and get a lot more done with the same amount of cost. Now where does this matter to everyday Joe Schmo users? Here's where this gets really interesting and cool. We talked about this a little bit last week, but these models now, when you need it to solve a problem for you, if it doesn't know how to solve that problem, it will often go and write code for you that can then go solve that problem. Well, they're getting cheaper and better and faster at writing that code behind the scenes. So now when you need it to go solve a problem for you that maybe it just doesn't know because it's not in its like initial training data, well, it can go write a script that then solves that problem for you and it's getting better, faster, cheaper at that stuff.

That's going to get cool. And just what built right into Claude, I would imagine, because you can watch the thing literally code in real time, just talking natural language to it.

Yeah.

Not using cloud code or anything.

Yeah. And if you use Claude Cowork, you know, the one that can actually take control and basically access your terminal and do things on your computer on your behalf, if you ask it to do something for you that it doesn't just know how to do natively it will go and write a script that can then go do that thing for you.

That's cool.

And it used to be a lot more expensive to go and write those scripts, but now that's getting cheaper because models like Sonnet are actually able to do similar level of coding but for less cost.

Yeah, and we're not even talking about perplexity here, which I was going to bring up, but I used Perplexity just now to search up why does this matter for non tech users specifically? But it's like, I guess Anthropic deliberately put Sonnet 4.6 in the free tier by default and included features like file creation, connectors, skills, context, compaction at no cost. So basically it's opus level intelligence for 1/5 of the price. Which means yeah, the casual users that, yeah, basically now it can have that top tier AI performance without paying anything for premium. Yeah, take that chat GPT with your ads. I'm just kidding, I'm not playing sides.

Yeah, yeah, yeah. So I mean, yeah, that's pretty much it. And then, you know, moving over, we also got Gemini 3.1 Pro this week. This is now Google's not their best model because they also have their deep think model as well, which is the one we were playing with last week, you know, getting it to solve the three body problem and whatnot. But this model is now like their best sort of widely available, inexpensive to use model. Right. So same kind of idea. Right. It's getting better and smarter. This one, it seems to have really leveled up in like sciencey kind of areas. Some of the ones that are the most interesting here, Arc AGI, they scored 77.1%. The next closest is Opus at 68%, almost like 10 percentage points lower. And what Arc AGI is, is it's like these logic reasoning puzzles that a lot of them are like visual that'll be like, okay, so if I have this pattern of images here and then you have this next image and it's got this pattern of images. What's missing from this one based on what you know about this one. And it tries to find like the visual pattern recognition for example. The models don't really go and like train on this stuff in advance. They have to actually look at the thing, decide what the next sort of sequence in the pattern is. And models have been really sort of bad at this over time. Well, this model is now the best at doing that. This one's also the best at scientific knowledge, competitive coding, scientific research, coding. So this one it seems like, they really leveled up with like the sciencey stuff, you know, like, if you, if you have a lot of like really deep research, science type questions. This one seems to be the best at that kind of stuff. And it also got really good at making SVG images.

I saw that.

That's, that's another thing.

I mean, that's great for what? Just like just website graphics in general. It's like now you can animate some really cool stuff instead of some static image and it looks pretty dang cool. Yeah, they're like more gradient, a little bit more texture smoother. So I'm curious, like, is this rolling out within? Basically, yeah.

I don't believe it's in Gemini yet.

Okay.

If I just go to like the regular gemini.google.com here. I don't believe this is the model we're using.

I'm seeing Pro.

Oh, yeah, it is, it is now. When they first rolled it out yesterday, it wasn't, but it is at the time of this recording now. Okay. So now you can switch to the Pro model and use this new 3.1.

Sweet. And I like that new button. This is a little off topic, but that creep music button, now we can have more fun with you. It's only 30 seconds though. It's not the Suno match, but I could see what they're doing.

Yeah. Yeah. I made a song about the Padres and it kept on calling them the Paiges.

Yeah.

Kept saying Paige Rees or something. Yeah.

Like, oh boy. Yeah. Pronunciation not too great, but I see it's fun. It's fun.

Yeah, yeah, yeah.

Cool.

So, yeah, that's pretty much the Google model. I did some tests on my recent video on my channel where I made it. I had it do an SVG of a wolf playing basketball and it did. Okay. Okay.

The numbers are a little, little crooked, it seemed falling off.

Yeah, yeah, there's. There's a little funkiness going on.

Yeah. I don't know, it's interesting because, you know, like, things like Perplexity for instance, has a whole bunch of models baked in. You can kind of choose or have it do the selection. And I don't know, I feel like, yeah, most users are probably not going to go and single out a specific model, but it's just kind of interesting to see how they're all matching up and they're all baked into something that we're using or you can at least look for it.

Yeah.

I would say again, for most people's everyday use case of like, you know, help me write this email to my boss that I don't know how to word it or you know, help me find a great carne asada recipe or I want to go find a great place to have dinner on a Friday night at 5pm in San Francisco and I don't live here, so where should I go? Like all of those kinds of just like random use cases that most people use a lot of these chatbots for. Yeah, they've been good at that for a while. And you're probably not going to notice a huge difference. You know, it's more like you're going to notice the difference when you're getting into real researchy, sciencey, mathy stuff and coding stuff and you know, even like SVGs, which it's actually sort of writing code in the background to create those SVGs. So it's a lot of that kind of stuff where you're going to notice the biggest difference. But if you're just trying to get like advice or brainstorm ideas or figure out where to eat or that kind of stuff, like you're probably not going

to notice business as usual. Just stick with your LLM and have fun. Just move on with your life.

Whatever one you like the vibes of the best, you could probably stick with.

There you go.

There was one more. Elon did release Grok 4.2. The interesting thing about this one, they didn't really do any sort of like big announcement around it. There was no blog post. Elon just tweeted about it saying hey, it's out there.

That's his blog post by the way.

Just that's look. Yeah, yeah, here it is. But the big difference between this one is it uses like what they call the council of experts. So whenever you ask it a question, it uses four different models that are sort of fine tune in four different ways. One for like fact checking, one for like research, one. I don't remember what all four of them are, but there's like basically four models. It asks the question to all four of those models. The four models essentially like debate between themselves about the best answer and then give you back the one best answer. That all that was sort of like a consensus answer that all four of them kind of came up with collaboratively, if that makes sense. Yeah, this isn't exactly an apples to apples comparison, but let's say you went to Gemini, you went to Claude, you went to ChatGPT and you went to Deepseek and you went to all four of these models and you asked all four of them that identical question and all four of Them gave you an answer that had some similarities but some differences. And then you took all four of those models, put it into one other model, and said, based on all of these answers, give me, like, a consensus answer. That sort of sums this all up kind of thing. Right. That's kind of what's happening.

I mean, it's a cool, interesting. It's basically agents doing the. I mean, the work for that specific question or query. Is that happening? Like, without, I guess, that wrapper, the agent wrapper, like on the other ones? I mean, it seems to, I guess, based on the model you're using. Right.

Yeah. So, I mean, if we were to really, really get into the weeds, there's been an architecture for large language models called the mixture of experts architecture, which is kind of similar, where you look at these models that have, you know, 500 billion parameters, but then you might see them, like, if you read about them, that might say there's only 17 billion active parameters. Well, what that means is that it's using this mixture of experts architecture where there's 500 billion parameters sort of baked into the model. But when you ask it its question, it's deciding which sort of subset of parameters to go and ask that question to, instead of sort of asking the entire corpus of parameters, if that makes sense. But that's getting, like, really into the weeds, and I don't really think we need to, like, go too deep down that rabbit hole. So, like, this mixture of experts concept has existed where you have these big models, and it will go ask sort of separate chunks of parameters inside of the model and then come to a consensus. And that's. That's the mixture of experts kind of concept. Got it. This is a little bit more like straight up, like in. In your face. Like, hey, we're asking this agent, this agent, this agent and this agent, and then they're all sort of coming together on a consensus at the end.

Yep. And. And the four agents, to wrap it up, I looked it up. There's a research, a retrieval agent, reasoning and solver, critic and adversary, agent, and a writer stylist agent.

So.

Yeah.

Yeah. So there you go. So there you go. It'll. It'll sort of, like, optimize on those four things and kind of give you the best of all four of those worlds, theoretically.

The four brain team, they call it.

Yeah.

All right, cool. Model race happening.

Yeah. Who's going to win? Joe, what's your prediction? Who's the big winner at the end of the race? Who gets the gold medal?

Who's The. I like anthropic. I like what they're doing.

Yeah.

But I know there some things are happening in the Pentacon where I don't think we're going to even go in there. But like with the government, we'll see how that shakes out. I'm always betting on Elon too, with the robots too. You know, there's infrastructure there.

Yeah, yeah, yeah. I mean, at the end of the day, I don't think there's going to be any one winner. I think we're going to see Google, Anthropic, OpenAI and XAI all doing big things and I think we're also going to see open source really sort of catch up and there'll be a lot of other alternatives. And because there's a ton of alternatives outside of the big foundation labs that's going to sort of force the big foundation labs to do better and keep their costs down for users because they have alternatives and competition is healthy because it keeps everybody trying harder and not sort of screwing over the consumers. So that's what I hope continues to happen.

Yeah, I like it. So next topic is we have something way more visual from Google. They did a drop. Pomeli. So Matt, what is this thing? The way I looked at it when I started tinkering around, I was like, holy crap. Like, okay, basically marketing agencies, this is a quick way to have them at least be disrupted completely. Or you have now a brand new tool that you can quickly iterate and create some, some cool brand assets, marketing collateral, all that fun stuff that is cool for the visuals.

Yeah. So what we're talking about here is called Photoshoot in Pomelli. And what this is, is you can give it like an image of your product. And this is for physical products in the real world. Like the examples they're showing are sunglasses and purses and sodas and things like that, Right. You give it these real products that exist in the real world, images of them and then it will actually create product photo shoots of them. So there'll be scenes that are like inside of a studio or scenes of somebody actually holding your, like your can of soda or a scene of it like sitting on a counter in a kitchen, things like that. You know, to me it seems like a, a sort of implementation of Nano Banana, right? Like they, they have this technology with Nano Banana. I can go to Nano Banana, upload a product shot of my product, upload a picture of somebody holding a different cereal box and say put my cereal box in that person's hand and nanobanana will do that for me. But this Pomelli is basically a sort of shortcut to doing that. Right. It's basically them saying, we've already got a bunch of studio templates and people holding product templates. You throw your product image in. We'll create the image of like people holding them and things like that. So you can sell them on your Shopify store or whatever.

Yep. I mean even tinkering around with it, I think it's still just like in this beta tester experimental thing. But I mean you can use it right now. And there's one screen right before this that I'm not showing. But basically you can plug in your website address. In this case, I plugged in our favorite baseball team, so the theme continues. So I plugged in this mlb.com padres and it took, it claims like 10 minutes. I think it took more like four minutes or something. But basically what it did is it came up with the business DNA of the Padres based on just the URL that I gave it. And you can see down here, it's pulling in the title, their logo, the font they use or something maybe close to it. I didn't double check all this stuff. You can see it's pulling in the fonts. So it's like basically it's a brand kit. It's a style guide. Right here. Yeah, you can see what it's trying to do. It's pulling in the brand values, the aesthetic, the tone that's being used in the full overview, which is pretty damn cool. Just right off the rip, like it's just pulling all this info in along with the images that it's pulling from its website, its brand. You can see it has quite a bit. You could upload your own as well. So that's kind of like where you start. And then over here in under campaigns is where you could start building out marketing campaigns. It gives you some suggestions like, so it created these without me doing anything and then these two up here I'll go to next. But basically they're ones I created based off of some things.

So yeah, these features, these campaigns, this is the stuff that's been in there. The photo shoot is the new feature. But this is just to clarify, this is like the stuff that has been in Palmelli for a little bit. The photo shoot features the one that like they just rolled out this week.

Got it. Okay, so you've been a little deeper and this is where I was just like, oh, this is, this is pretty cool.

And the photo, it is, I mean, Palmelli is still sick. Like it's still awesome like that it goes and it figures out your business DNA and we'll make all these banners and things for you. It's. It's pretty sick.

Animate it and all that stuff. Yeah, I've been a little slow on the Pimelli train. Sorry. But it ain't perfect. But it does pretty damn well the photo shoot feature. So I actually took a hat is one of the images. So this little hat here. Yeah, I took and then it placed it on this dude as a model. So yeah, rendered. All it started off was a hat that was in here and basically now you can see it looks pretty good. Basically something that you can go out and post has called the actions. If you wanted to put it on here. I'm just going to try create photo shoot. So yeah, it does take a couple minutes.

Yeah. But they overstate that too. I did it yesterday and it took like less than a minute, even though it says three minutes.

Okay. So I mean, look, it started off with this kind of top view of this hat. Put it on this dude, put this guy in Petco Park.

Yeah. That actually does legit look like Petco park in the background.

Yeah, I'm like, we've sat right about there before. And then you got this woman over here that's hanging out with a bunch of Easter eggs.

Cool.

But yeah, I mean, and then from this you have the ability to change up this. I think I can even animate this if I go back to my campaigns.

I don't know.

This is cool. Like as someone who's worked with a bunch of businesses and it's like, I mean, you could do this the manual way or you could just start off quick and have like a whole bunch of ideas.

So what you're saying is if I wanted to start a super quick drop shipping business where I get cheap junk on AliExpress, set it up on a Shopify store.

Great idea.

And sell it to people quickly. I can use ply to go and just download the picture from AliExpress, throw it into Pumeli. It'll show people using that product. I can put that on my Shopify site and then I can go over to Manus and say, hey, Manus, create some ads for me that sell my Shopify junk and I got a business just like that.

Yeah, you got everything done for you. It's done. Yeah.

That was seed dance 2.0 that we were looking at last week. We still haven't gotten to play with it, but yeah, I mean, honestly, how Easy is it today versus like five years ago to create an E commerce business, like especially a drop shipping business where you just find a product online and you sort of resell it. You use a tool like this to make all your product imagery. You use, you know, Shopify to like actually set up a storefront. You use HubSpot to set up your CRM. Right?

There you go.

Yeah, you go and set up Manus and your Facebook ads to go and drive traffic to the ad and boom, you've got a business.

Like overnight, you literally have everything.

Yeah, dude, there's never been an easier time in history to like spin up a business and if it doesn't work well, you didn't put too much time and effort into it. Right? You spin up a business, you test it for a couple of weeks, you go, okay, this one's not working. Let's go spin up another business.

Yeah, I looked at this. I mean, like, I ain't gonna lie. I have a client that pays me. The thing is like, we create a lot of these type of assets and the first time I saw this, I was like, this is crazy. Like, that just made what we're doing kind of irrelevant. Not completely because there's more to it. But I'm like, shoot, if you're just a company developing assets like this, you better be thinking of how else can you be valuable to that company and not saying this is going to just replace what you're doing overnight. If you're like a Canva ninja or whatever, social media, whatever you're creating your assets media for. But I'm just saying, like, maybe try to elevate a little bit higher up than just the visuals at this point, because it's coming for your lunch.

I've sort of talked about this quite a bit in the past is people need to start to take more of the conductor mindset versus the trumpet player mindset. Right? Like, you can't just be the one person in the band that does this one. You need to move out a layer and be that conductor who's saying, okay, I'm the guy who now creates the ads. I'm the guy who creates the brand image. I'm the guy who creates the Shopify store. I'm the guy who creates, who sets up our CRM. I'm the guy who goes and does the ads now because AI is here and it can make all of that stuff feel effortless. So now you're just this conductor that's saying, go do this, go do this, go do this, go do this. And the business sort of happens around it. But saying all that I think it is. And Joe, you and I have had many, many talks about this on our various coffee dates. There is a short window of time where this is going to be true. Because as this gets more and more democratized and more and more people can access all this stuff and it gets easier and easier and easier. Well, the benefits that this brings you becomes more of a commodity. Like any. Once anybody can do this, everybody will do this, which makes it sort of not as valuable anymore. So, like, the window of time to be capitalizing on using these tools and like bringing them to whatever business you work on. The time is right now, but that window's going to close.

Mm. I'm a big believer of figure out how to be a really good human in this era of technology and know the tools. Exactly what you just said, Matt. Like, gotta be aware, gotta be using them. Like, even if it's just playing with this for 10 minutes, you know, it's like you will know more than most, but at the end of the day, it's like, yeah, figure out how you can use your human side and use AI to augment how good you might be or just get better and smarter, faster, stronger. Back to that song again. But, like, I feel like that's gonna win out. People are gonna want that connection in some way.

Yeah, I actually feel like this is a fairly decent segue into the next topic we wanted to talk about.

Let's go. So you put a tweet out, Matt, that we're looking at in the screen here. I thought was kind of funny. You're like, I'm gonna buy two copies of this Time magazine. I was like, what's Matt talking about here? And essentially this magazine has a big old article, but the COVID is the People versus AI. So it's obviously different stories of people behind the growing backlash of AI and the build out of data centers and what energy, water, all the consumptions. And you read your tweet, Matt, and you can elaborate on your own words.

All right, so here's the Time article tweet. It says, this is Times new cover. In a deeply divided nation, a new coalition is forming around one belief. AI is moving too fast inside the stories of nine Americans across ideologies and professions determined to slow down the technology reshaping daily life. My tweet in response to that was, I'm going to buy a copy for two reasons. One, I genuinely do want to hear both sides of the argument and understand where they're coming from that is 100% true. Number two, these will be used as the memes 10 years from now and read like the people who tried to slow down the growth of the Internet. Right? Like, if you look now, when people talk about AI, how often do you see them referencing, like, you know, articles from the 90s that say, what is this new Internet fad that everybody's talking about?

Right. Well, Internet superhighway.

Yeah, the Internet superhighway, the World Wide Web. And we can all see how that all aged. Everybody looks at that now and goes like, it was inevitable. Like, why were you talking about it that way? That's sort of how I feel about articles. Like, this is 10 years from now. People are going to be looking at this like, these were the people that said that we should slow down the advancements of the Internet. Like, that's what it's going to read. Like now, when I posted this, I. I got like, mixed comments. I always mute my posts after, like a day because it's like, after a day, trolls just jump in and it's just pointless. But some of the early comments were like, I saw one comment that said something like, I'm going to pick up this magazine so I can read the arguments of the people that I've already decided I disagree with. Right?

Yeah.

That's not true. That's not what I said. I didn't say I disagree with these people. I'm not saying, hey, we shouldn't slow this down. All I'm saying is that's not going to happen. Right? Like, I want to understand why people want this slowed down, and I do understand a lot of the reasons they want it slowed down, and for a lot of the reasons they want it slowed down. I actually agree with many of them. I do think that there are big cons with what is happening in the world of AI and how fast this is advancing. I do believe that, and I do want to read about it, and I do want to hear those perspectives. And a lot of them I probably actually agree with. Yeah, but does me agreeing with them or does time writing a magazine about this topic, is that going to effectively slow it down? Nope. Not even in the slightest. It's. It's just not right. There's multiple reasons here, and probably the biggest one is if we say we need to slow down AI, let's just say we all in the US Agree, let's slow this down. It's all moving too fast. Do you think China's gonna be like, yeah, we're gonna slow down also Right. No, it needs to be like a global agreement if we're gonna slow down. There's mistrals in France, you've got the bite dances and the Alibabas. Over in China, you've got other countries developing this stuff as well. Russia's trying to develop their own as well as you've got other countries that are trying to push the advancement of this stuff. So if we do say we want to slow it down, you need a global agreement that we're going to slow it down, not a sort of localized country agreement that we need to slow it down. Yeah, Right. Like, when it comes to, like, military advancements, do we want the militaries of these other countries to have the upper hand? No. Right. When it comes to. I was talking about that Quinn model, which is an open source model that's almost caught up with the frontier models. Do we really think the open source models that are coming out of China are going to slow down? Do we think that the people in the US aren't going to be able to get their hands on the open source models and further develop them under the radar outside of these foundational labs? It's not a feasible thing in my mind. It's not. I mean, I could probably hear arguments that might change my mind, but I have a really, really, really hard time seeing it as a feasible possibility. That doesn't mean that I disagree with the people who are saying it should slow down. They're two different things.

It's true. Elon. I know he had a point and I was trying to find the tweet just now, but I can't quite yet. But you probably know what I mean, where like, for a while he was saying the same thing. He was kind of sitting on the sidelines, more or less, but then he just realized, you know, this train's moving so fast and he has his own way of how he wants to do things with AI and in the future he's like, I just gotta jump into the race because, you know, like, it's not slowing down and it doesn't seem like anybody else at the helm is able to do that either. So he's in it, you know, so that was his decision. But then, you know, there was this documentary mini doc, it was only like 10 minutes long on YouTube and came out like two years ago. But I just watched it randomly like yesterday. Matt, you said you've seen it before. It's called two Scale time and basically just shows the evolution of humans over the scale of 13.8 billion years. And just Looking at it and then you know, as this timelines this thing out of the scale of human history. But it shows it was trying to make a point of like how there's so much advancement that's happened recently. Also the point was like we're so minute, we're like a razor. Like half of a hair length of our of our time is actually what we'll ever know. Right. But on this timeline just shows how fast things are progressing. And this was before AI was really doing much like that wasn't really mentioned. So I'm sure people resisted technology growth over millennia, over 13 billion years. But now we're starting to see so much more change in such a rapid period of time. It's almost like we're probably not even used to even fathoming what is happening anymore. Because you can just see that timeline right there. Yeah. It's like. And they basically went to the Mojave Desert and scaled out over I don't

know how long, miles.

It had to have been miles. It's really cool because it is like drone shot from the beginning of time. Basically fire you can see and over the span of this whole period to where it shows now the present. But this was looking backwards, not forwards. And I thought it would be also kind of cool for you and I, Matt, since we're throwing out predictions all the time. We're like, what the heck is going to happen ahead of time? You know, like on the other, the left hand side of this scale here. What are your thoughts, I guess initially Matt, on all this?

Because the human timeline thing.

Yeah. But also like this whole pushback, it's just, I don't think it's going to slow down. It just seems inevitable that everything's going to speed up in the timeline. If we were to extract this timeline out further, it would just be insane amount of development.

I think this meme accurately describes how I feel. It says Time magazine, the people versus the wheel.

That's the point.

Grunts from the silent majority. Dragging rocks and rolling is a drag.

There you go. So there you go. The evolution of time. It's like really, are we going to fight this? I guess we. Maybe someone did. I'm sure someone fought electricity as well. Like I'm sure there's a lot of stories on that.

Yeah. I mean again, my thoughts on this whole thing are I actually do see both the pros and cons. I've been very vocal on both my YouTube channel and on this podcast and on my X account about like there is pros and there are cons, there are positives, there are negatives. There's a lot that I love that AI could potentially do. There's a lot that I hate that people are using AI for. I agree with a lot of things on both sides and on X and a lot of these places, nuance is very hard for people. They think you need to be polarized. They think you need to be on one side or the other. Like, if you use X as like your sort of barometer for how people feel about AI, you would think that you're either on the side of I absolutely love AI, we need to advance it and nothing's going to stop us, or I absolutely hate AI, we should shut it down, bomb the data centers, all that kind of stuff, right? That would be what the feeling you get on X would be. You're either on one side or the other. But the reality is the majority of the human race, the majority of the people I talk to on a daily basis are more in the middle. When I go and have dinner at my neighbor's house, right, one of my neighbors is a teacher, one of them's a graphic designer, one of them's a painter, One of them, he's like some sort of engineer, right? And all of these guys, whenever we all hang out together, AI comes up quite often because they know what I do for a living. So they all kind of want to bring it up and talk to me about it. And they all sort of live in the middle. Most people live in the middle. Most people are like, I really, really hate the fact that when I get on Facebook, I can't tell if what I'm looking at is real or fake anymore. But I really, really like that. I don't have to like read like my boss's 2000 word email anymore. I could just have AI summarize it for me, right? Like, yeah, most people fall like in this center. And the people that are doing this whole like Time magazine people, they're on one side of that spectrum. They're the ones that are saying, I'm on this side of the polarization. I'm one of those people that thinks we need to shut it down. And I have a really, really, really hard time getting my head into the headspace of the people that are on either side of that polarization. Like, I can't get inside the head of the person that says, accelerate at all costs. I don't care what it means for humanity. But I also can't get into the headspace of the people that say I hate AI. We need to stop it. We need to slow it down. We need to end this. This has gone too far. My opinion is that most people are probably in the middle. My guess is that most people are somewhere in the middle of like. ChatGPT is kind of cool. It helps me a lot. But I hate all the AI Slop videos I'm seeing lately.

Yeah.

Sorry about the rant. I have a lot of rants.

I like the rant. I sparked it. It worked. Takes a great. Yeah, man. Let's keep an eye on this. I feel like maybe this is a recurring thing of like, how can we. Not always the people versus AI but the idea of like predictions and where we maybe we can even, I don't know, somehow visualize this thing too. Kind of like how that video did it.

Yeah.

I don't know if we'll be going to the Mojave Desert to do all that. That seemed like a lot of work, but it's really impressive. Yeah, yeah, yeah.

I've seen it done on the scale of a football field as well. And it was sort of like the same thing where like, from science to present, like right here in this little teeny gap, when they laid it out on a football field, it wasn't even like the one yard line. It was like the one inch line, you know?

Right. I think right now it's just a big education phase. And I agree most people are probably in the middle, especially the more they learn about what's actually happening.

Yeah. You know, and there is another sort of thing I want to talk about that's in the same vein and that's the whole idea of like, creativity and like, can AI be creative? Like, there's obviously a lot of like freakouts going on in Hollywood right now. We. We even briefly touched on it this week and last week about the whole seed dance thing where Hollywood is kind of panicking right now because it can create Disney ip. It can create actors that look exactly like Tom Cruise and Brad Pitt and put them in a movie. And that scares people. Right. There's that whole discussion, which I think is like a really sort of fascinating rabbit hole as well. And are you familiar with Fei? Fei Li?

No.

They call her like the. The godmother of AI.

Oh, no.

And she has like a really good clip. She went on TVPN and it's a three minute clip. I'll just play it. And if, you know, we want to edit out just down to like the good bits, we can do that. But I think she has like a really good take on creativity. But Then also Matthew McConaughey has been in the news recently around his take on this and Ice T has been in the news the RA about his take on this. And so I want to share some of those takes as well because I think that's kind of like an interesting discussion. But I want to start with Fei Fei Li, because Fei Fei Li, she invented this thing called the imagenet and it's like this giant database of images that were labeled of what the images were and then those labeled images became what now are like image generators, right? Like it became a lot of the training data that now a lot of these image generations things came out of. And she now has a company called World Labs where she's creating basically spatial intelligence. She's teaching AIs to understand the physical world and how, you know, depth works and height works and gravity and motion and like actual physic like putting AI into the physical world and really, really fascinating person. I read her autobiography, it was awesome. But I want to share what she has to say about creativity because I think she's got a really good take on it.

I consider myself a scientist in my heart and I actually really don't like hyping as a scientist. I think world model is underappreciated because it's so new. And in terms of overheated, I do think that Silicon Valley as a whole tend to mistaken clear vision with short distance. For example, creativity industry, which we are also engaging in, I don't call it hype, I call it a misleading sentiment which actually really bothers me is that we don't want to replace human creators. You know, human creativity is precious. It tells the story of our species, of our culture, of our society, of our community, of each one of us. What we want to do is to superpower and augment creators capabilities. And sometimes the communication of technology is a little skewed towards oh, the model can do all the work in creativity. And I personally really want to highlight human creativity as something that's so fundamental to who we are. And AI is here to help and augment.

I really, really like her take on creativity. First of all, I think it's fascinating to think about the autonomous vehicles, right? The first self driving car, 2006. We're just now getting Waymos in 2026. One of the points that she was obviously making is like a lot of people think that this is moving a lot faster than it, it actually is. But also exponential growth curves, right? I do think a lot of the large language model stuff is moving faster than what the pace of, you know, 2006 to 2026 was with self driving cars. I think, you know, robotics and the house cleaning robots and stuff will take not 20 years to get to from now. Right. We're probably only more of a few years off due to exponential growth curves. But the whole creativity aspect is something that, like, I feel like I've been trying to say for a long, long time, but everybody wants to argue on the creativity aspect and that's that like, humans should be the creative force, but AI gives human superpowers to make it so that whatever their creative vision is, they can actually achieve that. Not everybody is going to be able to have like these crazy CGI skills or the ability to draw anything they want or the ability to create music. But now you can have a creative vision and not having the knowledge or skills that used to be gained over 20 years, you can now sort of put your creative vision into the world because these tools exist. And I think that is the way to be looking at AI when it comes to creativity is that it's not taking away human creativity, it's taking away the hard part of getting that creative vision out of your head into the real world.

Mm. And I see it as. Cause I just heard another example the other day of, of someone who was using AI for one of the first times to create like a business. It was in the brewing space and basically he was saying how like this guy who's like, man, I have the best bulletproof business model and all this stuff, I can build it all. And basically it was the AI telling him how smart he was and not pushing back on him with any critiques and all that. Until our buddy basically said, okay, I want you to ask. I don't know what he was using. Let's say Claude or. Yeah. And say like, okay, so point out all of the potential flaws in this business model and like, how do you poke all the holes in this whole thing? And then he's like, oh God, yeah. And basically it's the AI talking back saying like, yeah, this model would never work and here's why, blah, blah, blah. And it basically, like essentially brought the guy to tears. I wasn't in the room, but this is what I kind of the essence of it and what my takeaway was there, which I think a lot of folks will kind of lean on AI for maybe creativity and all that is like just to have it do the thinking and do all of that stuff without coming to the table and using it more as a thinking Partner and like you said, maybe run a whole bunch of ideas that we as humans are coming. The unique perspectives, the, you know, the flavors that we want to bring into whatever our work is. And you know, maybe we pull all that context out of ourselves somehow. AI can help with that and then it has a better picture of how we're thinking creatively and then it can expand and help us get that out of our head, out of our mouth, whatever it is, and then iterate from there.

Yeah, well, I mean like how much of the AI videos that we're seeing circulate are just like kind of slop memes that, that don't really have great story. That don't really have great. Like, yeah, like they're very throwaway feeling content. Right. They're like one shot prompts. Because you have access to the ability to make videos and because you have access to the ability to make music and to write articles that like, are we seeing an abundance of amazing stuff put out into the world as a result? No, because you still need that like a good human, creative person to go and orchestrate that stuff for you. And, and I think that's the part that a lot of people are missing is that like you still have to have taste, you still have to have creativity, you still have to have ideas. And it turns out we could give people all the tools in the world to create whatever they want and people just don't know what to make with them still. Right. Anybody can go and create any app they can imagine right now you can get into Claude code, tell it an idea for an app and it will go and make that app and maybe you go back and forth with it a little bit, but 10 hours from now you can have a working app that does what you want. Why isn't everybody a millionaire selling SaaS products yet? Why not? Because nobody has good ideas. You still need to have the taste, you still need to have the ideas. You still need to have like this sort of implementation mindset of like, okay, this isn't working, let's do this. You need to have a feel for user interface and user experience and test your own apps. And turns out most people, even when you give them all the tools in the world to make anything they can possibly ever imagine, they're still too lazy to do it.

There's a reason why the word of the year last year was slop.

Yeah, yeah. So absolutely. Here's another thing. This is from Ice T the rapper and along the same lines, right? So somebody wrote on X. I wish more older Rappers realized that AI shouldn't touch the music industry in any way that impacts the creative work, including music videos. Right. And then Ice T basically says, I disagree. Fans want us to make and produce the music, then shoot an expensive video. Then they get it for free if they have an Apple subscription or Spotify. I would argue if you have an Apple subscription or Spotify, you're not getting it for free. But whatever. That's beside the point. The days of the expensive videos are over. There isn't even mtv. AI is the only sensible way to add visuals to a song. You can hate it all you want, but it's the future. Right. This is somebody that has money, has resources, has means to go and make music videos if he wanted to. And he's still saying, yeah, but AI just makes it so much easier and cheaper. So we're all going to go that route.

Yep. That's interesting. I think this is cool to see the perspective from creators and people like McConaughey in Hollywood as well.

Yeah. So Matthew McConaughey. So there's a video that's going viral that is from Variety. I don't think the actual question is asked in this video, but basically he's telling people, get on board because it's here, whether you like it or not. So here's his take on it.

It's coming. It's already here. Don't deny it. It's not enough. It may be for you, but it's not going to be enough to sit on the sidelines and make the moral plea. The moral plea that, no, this is wrong. It's not going to last. There's too much money to be made and there's.

It's just. It's too productive.

It's.

It's tear. All right, So I say own yourself, voice likeness, etc, trademark it. Whatever you got to do.

So you did that.

Yeah. Own yourself. So when it comes, not if it comes, no one can steal you, but they're gonna have to come to you to go, can I? Or they're going to be in breach and you'll have the chance to be your own agency and go, yeah, for this amount or no. Okay, it's coming. Is it going to be another category or is it going to infiltrate our category? It's damn sure going to infiltrate. Infiltrate our category. I think it'll end up. Does it become another category? Will we be in five years having films. The best AI Film. The best AI Actor. Maybe. I think it might be. That might be the thing is that becomes another category. I'M not sure it's going to be in front of us in ways that we don't even see it. It's going to get so good, we're not going to know the difference. That's one of the big questions. What we're doing right now is the question of reality. That's. It's more hazy than ever in a very exciting way, I think, but also a scary way.

So nuanced take, right? Exciting and scary. Not a lot of people have that nuance.

I just wish everything was voiced in his voiceover. I mean, he's got a nice cadence, and he really hammers in on that. Whatever.

Well, his voice is trademarked now, so you can probably, for the right price, you can use it. Well.

See, that was my biggest takeaway is like, he made. Because I saw. Yeah, he did this whole trademark of his ip, basically. Genius. And this goes back to Fei Fei Li and how she was talking about creators. It's like, well, this might actually be the time, if we're thinking it correctly through. It's like, okay, well, own your ip, put your stake and basically protect yourself so you can become your own agency person as a creator. And that doesn't just mean, like, YouTuber, but it could just be an actor, an artist, anything out there.

Yeah. One of the things that he was talking about is, like, you're trying to stand on, like, this moral high ground of, like, this is wrong. You're trying to replace us. It's all of this, like, moral thing. But at the end of the day, and there was a really good Colin and Samir podcast episode where they talked about this AI stuff. And I actually went and left this comment on their video because I felt so strongly after watching the video, is that you could stand on your moral high ground right now. But the reality is, one, you might say, like, hey, I'm never gonna watch AI videos. I don't want to see what an AI video looks like. If a movie was fully made with AI and it comes with theaters, I'm not gonna go watch it. I don't wanna see it. Okay, that's cool. You do that. Here's the thing. Give it a few years. Nobody's gonna be able to tell us AI Right. And who knows if the studios are even gonna tell you it was generated with AI at that point either. Right? So we're gonna get to a point where you have no idea that this was made with AI you might be watching the next episode of South Park. They did the who and they just didn't even Tell anybody they used AI but the whole thing was just generated for them and you're watching it without even knowing it. That's going to be a future reality. Whether you like it or not, that is going to happen. These studios are going to figure out how to cut costs, how to crank out movies faster, how to crank out TV shows faster. How do I pay actors less, how do I pay voice actors less? How do I pay animators less, how do I pay CGI artists less? All of these studios want to figure out how to cut costs. They're going to do it. It's going to happen. It's inevitable, guaranteed. That's point number one. Point number two right now. Maybe our generation is romantic about the idea that we have movie stars. We have our Tom Cruise's and our Brad Pitts and our Angelina Jolie's and the people we grew up watching on the big screen. And these are the stars in our eyes that these are the famous people that we want to go see in movies right now. And we're romantic about an era where that was the reality. Yeah, my kids, Your kids, Joe, they will not give a shit.

That's right.

They do not care. They are growing up in a world where AI Slop is going to be everywhere. By the time they're on social media and on the Internet and watching tv, it's going to be just everywhere. Like, they won't be able to get away with it. They're growing up in a world that's native to AI it will just always be there. They're not going to have that same sort of romantic sentiment about like, oh, I need to go see Matt Damon in the new movie, because it's freaking Matt Damon. They don't care and they won't care. That is not the future of the younger generations. So our generation may not want to go to a movie and see AI Slop, but the next generation following us up won't give a damn.

Yeah, it makes me. You brought up south park and we know how, like, with what Seed Dance, how fast it is to make something. And this reminds me of the documentary Six Days to Air. Yeah, yeah, yeah, I know. You know? Yeah. It's the whole idea of, hey, you got six days from an idea to the final cut that's actually pushed out to Comedy Central. And it's like, dude, that's ain't six days anymore. That's. I mean, I'm not saying South Park's doing this at all or ever will, but what you're saying it's possible well,

they did have an episode they wrote with ChatGPT, and the episode was about ChatGPT. So it has happened at least once.

But visually, in the whole nine, though, you know. Yeah.

Yeah. I just think that the people are stuck in the way it is and the way it used to be instead of looking at where it's going to go. And that's why I think Ice T and Matthew McConaughey and some of these people are looking at this and going, look, it's going there whether you like it or not. So we have two options. We can stick our head in the sand and say, look, I am not getting on board with AI. I am not going to let companies use my face, my likeness, whatever. I am going to boycott it. I am going to write an article for Time magazine and tell people how much I hate AI. Right. You can go and do that thing, but it's still going to steamroll over you and the world's not going to care.

It's going to keep spinning and it's going faster. Yeah.

So why not set yourself up for a world where you are placing yourself in a good position for when it is here? That's what he's doing. He's saying, look, I am trademarking myself. I am trademarking my voice. I am trademarking some of my sayings. Like, I'm probably not allowed to say. All right, all right, all right. Sorry, I won't say the third. All right. Because that's probably trademarks now. Right?

Oh, man, I'm so fun to say

it, but he did that in a way where even after he's gone, his estate can still make money off of his face, his likeness, his voice, his sayings. Right. Like, he can go on vacation and have a movie shot that he's starring in while he's sipping pina coladas on the beach. Right. Like, why not choose that path instead of the path of, I'm going to resist this as much as possible, but then be irrelevant in a few years.

Actors like, yeah, my kids. Your kids. Like, so I guess what is the future of, you know, a person sharing content, value, something to the world? Like, do they not care about that at all? So it's like, you know, so our creators actually more in jeopardy than anything because they're kind of like actors, if you think about it. Like, someone trusts them. Right. And they want to hear from them.

I still lean on what I said last week.

Yeah.

I think it comes down to taste and opinions and things that are uniquely human. Right. Like, if an AI is Giving an opinion. That opinion is basically a general consensus. Right. That it's based on all of the training data that is collected from all around the Internet. So it's sort of like leaning on the mean. It's leaning on the average of what most people would say about that thing that you asked it. But it's like an average opinion of the corpus of data that it was trained on. Right. It's not like a real human opinion. I think that's going to continue to be valuable to people. I think taste is going to continue to be valuable to people. If Steven Spielberg went and had access to AI tools and prompted all the scenes to put together an amazing movie, it would come out a hell of a lot better than if I went and I prompted all of the scenes to make a movie. Because he knows what goes into good storytelling, he knows what goes into good visuals. He know like he has all of that experience where if he now sort of leaned into the tools that are available to him and use those tools and to put his process around it, but with these tools that can make it faster, he's going to do a hell of a lot better job at making a movie than I will having access to the exact same tools.

There you go. Taste, us being human are going to bring out the best of this stuff with the tools, at least for now.

Absolutely. We can circle back around to this topic because I have a feeling it'll be an ongoing topic often. So we can circle back around. I think we wrap this one up pretty quickly here. But before we do, we show off some cool robot ninjas.

That's really how we should wrap up everything, right?

I think every episode should just wrap up with robot ninjas.

This is crazy. Yeah. So Chinese robots have Spring festival gala. I guess it happened this week. It's going viral on X. There's this video that you're showing now. So it looks like it's. Yeah, there's humans dancing and also doing crazy like martial arts moves. I mean, I got a little freaked out when part of this, I think it had swords. So I was like, oh God, yeah.

Yeah.

And my guess, and I think I read this correctly, they're all pre programmed. I mean, they're obviously synchronized extremely well.

Yeah.

So like if we're thinking Optimus, you know, with Elon Musk and his robots that are obviously different from this, his are not like this.

What I think about when I look at this, I think this is like the future of military. Right. Like, why would we go and send like humans into battle? When we can just go and send robots in and. And they can hold onto the weapons and go and do the business. Right.

I already heard they're doing that in Ukraine. Like, they're doing. They're sending robots that are kind of similar into buildings to, like, you know, search them, but also be on the front lines to do reconnaissance, where that's probably the most dangerous part of things, you know?

Absolutely. And so, like, I look at this, and we see a lot of stuff in China about robots lately. And I was at CES earlier this year, and robots like this were just freaking everywhere. They were everywhere. And what I learned during ces, seeing the robots everywhere, is that when it comes to, like, mechanical engineering, making robots that can do anything, there's not a bottleneck there. Look, they're doing. They have robots that can do backflips, that can swing swords around, that can use nunchucks, that can jump seven feet into the air, jumping backwards over them. We're there with robots. There's no issue with getting robots to do any sort of movement that a human can do. The problem is still on the software side. The problem is still the robots don't necessarily understand the environment around them. We don't have robots that can really go and do our dishes for us, because, like, when a human goes and does their own dishes, they have different strengths with which they might grab different things. A heavy plate, you grab it and lift it at a different sort of rate than you would go and grab a little plastic cup. You might give it a lighter touch. Or if something was glass, you won't squeeze it as hard because you don't want to squeeze it and shatter that glass under your hand. Right. Robots, on the other hand, they're just as capable of a hand going like this and closing around a glass. They're just as capable of going and picking up a plate and lifting that weight. But it's the difference between all of those little potential nuances and everything that might be in your sink that needs to go to the dishwasher, that it can't sort of switch between all of the nuanced items. Yet that's a software issue, not a hardware issue. Right. So what I've sort of come to learn about robots is that from a engineering mechanical standpoint, we can make robots that do literally, like, anything. Now, from a software perspective, they don't have an understanding of the real world yet. That's also what Fei. Fei Li is working on when she's trying to build these world models, is trying to help these Models understand fei physical intelligence, physical world around them, understanding depth, understanding how much to grab this thing if it's moving eggs from an egg carton, you don't want to just go and pick up, like, the whole thing and squish all the eggs inside.

Right? Yeah.

The robots don't have that level intelligence yet to understand the nuance of every possible thing that it could potentially interact with.

Totally agree. That's my takeaway here is, like, when I first saw this, I was like, dang. Like, the movements, I mean, they're just. It's clean, you know. But on the intelligence side, it seems like that's where, like, Elon Musk with everything he's doing and, you know, he just. What? They stopped producing the Tesla X and the S to produce Optimus robots in the Fremont factory and maybe elsewhere too. But, like, it seems like he's going more for the intelligence side of things, of these robots. Not sure exactly the mechanics and how if he has. If they have the possibility to do all this yet. But it's interesting exactly what you're saying. They have, like China with the mechanical side, obviously, dial that in and production, we all know they can crank stuff out, but when it comes to the intelligence side, that's the gap right now it seems like Elon's kind of coming from that angle. I mean, I'm sure he's going to come from the production side too. He's got that covered pretty well.

Yeah. Yeah. Well, so right now, if you look at where China has advantages, China has advantages on the mechanical side. Like, they have a culture of industrial manufacturing. Right. Why do you think the US So much stuff around your house has made in China? It's because the China has all of the manufacturing infrastructure to manufacture tons and tons and tons of stuff. What China doesn't have is all of the compute. We have Nvidia over here. We have amd, we have intel, we have Qualcomm, we have, you know, you name it. All of the companies that are developing and engineering the chips, they're actually not making the chips. Almost all the chips are made in Taiwan, but. But they're actually sort of engineering and architecting the chips that are then made at the manufacturing facilities in Taiwan. And so we have, like, an abundance of compute here in the US So we can make things smarter and smarter and smarter. You know, things like large language models and image models and agents and all that kind of stuff. They have a large manufacturing base, so they can actually make machines over there a lot, lot better and faster and more efficiently than we can make machines, but they don't have the same level of compute and intelligence to put in those machines. The irony is, if the US and China just decided to work together, we would probably have the smartest robots in the world, like, already, because they would be using the US's ability to create this insane good compute with China's ability to manufacture these robots really, really quickly and a lot less expensively in the us. Elon's probably not a good example because he's probably manufacturing a lot of it here. But when companies in the US are trying to manufacture their own humanoid robots, most of the parts to make those robots are made in China and they have to import them into the us. So when somebody wants to make a humanoid robot themselves here in the us, they need to go and figure out what parts they need. They need to order them from China. There's, you know, multiple weeks of getting over here, going through customs, getting those products here, getting them here, manufacturing them, assembling them all in house, right? In China, they don't have those same bottlenecks to actually manufacture these robots. They have the facilities. They can just make the robots really quick. So what does that do for China? That means that they can fail a lot faster. They can make a robot go, oh, that one didn't work. That one doesn't do backflips. Let's try again. All right. They spin one up, two weeks later, this one can do backflips. Cool, right? They can iterate a lot faster because they have the manufacturing culture and infrastructure to manufacture robots at a highly rapid rate. Where in the us, every time you fail, you have to go through that whole process of getting parts from across the world back to the US to make your next iteration. That's sort of the difference between the US and China and why China is so far advanced in robotics and the US is so far advanced in the actual, like, chips and intelligence. But China is begging to get the smarter and smarter chips in. The US wants the robotics, right? And it's like, that's why there's that sort of like massive difference in capabilities between the two countries right now.

You know what's interesting is I just looked up and Elon Musk's Optimus, the humanoid robot currently relies significantly on China for its key parts and materials. They're just assembled here.

So there you go.

So that's still the bottleneck. I mean, maybe not for a long time, but still bottleneck. And even Musk said, like, you know, he's affected by some of the things that happen in China and, you know, stuff in between the. The countries.

Yeah. I mean, he'd be able to iterate on his Tesla robots a lot faster if he had a facility in China just cranking them out for him. But he's getting the parts over here. They have to build and engineer them over here. And then if it doesn't work well, then they're going back and forth to get all the things. And it's sort of a difference in, I guess, this sort of like business processes of the two countries. Like China just put a huge, huge, huge emphasis on manufacturing and the US Has a lot less of an emphasis on manufacturing. Meaning when you do manufacture something in the US It's a hell of a lot more expensive than when you manufacture it in China. Yeah.

And it still says even the future, like Elon's probably going to keep reliance with China on the parts and all that stuff. So.

Pretty wild.

Yeah, we'll see where it shapes up. We always got to end with some robots. Robot fun.

We got to end with some robots, especially ninja robots. I call those danger robots. Scary. Well, cool. This was a. Definitely a long episode, but we covered a lot of ground and got a couple of good rants in there. And if you like this kind of stuff and you want to stay looped in and hear me and Joe rant on what happened last week in the world of AI well, make sure you like this episode, subscribe to this channel. We'll make sure more of this kind of stuff shows up for you. Thanks again for tuning in. Really appreciate you and hopefully you will catch you in the next one. Sat.