WEBVTT 00:00:00.000 --> 00:00:02.020 Okay, let's start with the concept. It's something 00:00:02.020 --> 00:00:05.080 that usually you'd find in a sci -fi novel, not 00:00:05.080 --> 00:00:07.780 a Tuesday news headline. Recursive self -improvement. 00:00:08.060 --> 00:00:11.560 It's the big one. The holy grail of AI, really. 00:00:11.699 --> 00:00:14.099 The idea that a system can look at its own code, 00:00:14.220 --> 00:00:16.940 find the problems, and just fix them. Rewrite 00:00:16.940 --> 00:00:18.920 itself to be better without a human stepping 00:00:18.920 --> 00:00:21.359 in. And we're not talking theory here. This is 00:00:21.359 --> 00:00:23.519 what broke this week with OpenAI's new model. 00:00:23.600 --> 00:00:26.120 It wasn't just writing code. It was being used 00:00:26.120 --> 00:00:29.620 to debug and evaluate itself. doing its own training. 00:00:29.760 --> 00:00:31.739 That's a huge shift. I mean, we're moving from 00:00:31.739 --> 00:00:34.840 an era of tools we use to one of agents that 00:00:34.840 --> 00:00:37.759 build themselves. And it's honestly, it's a little 00:00:37.759 --> 00:00:40.179 airy. A little. It's kind of terrifyingly cool. 00:00:40.359 --> 00:00:42.359 Welcome back to the Deep Dive. Today is Thursday, 00:00:42.500 --> 00:00:45.899 February 5th, 2026. And I hope you're ready because 00:00:45.899 --> 00:00:48.700 the world feels different than it did just a 00:00:48.700 --> 00:00:50.299 week ago. Oh, it really does. The acceleration 00:00:50.299 --> 00:00:53.570 is just... It's noticeable now. So here's our 00:00:53.570 --> 00:00:55.469 roadmap. We had a lot to get through today. First, 00:00:55.570 --> 00:00:58.289 we're going to unpack just the absolute madness 00:00:58.289 --> 00:01:00.829 of this agendic arms race. Yeah. We had OpenAI's 00:01:00.829 --> 00:01:04.109 GPT 5 .3 codex and Anthropix's Claude Opus 4 00:01:04.109 --> 00:01:07.409 .6 drop, what, minutes apart? It felt like a 00:01:07.409 --> 00:01:10.189 coordinated strike. It was wild. Totally. Then 00:01:10.189 --> 00:01:12.349 we'll look at how this all spilled out into the 00:01:12.349 --> 00:01:13.969 real world. I'm talking about the culture war 00:01:13.969 --> 00:01:16.730 at the Super Bowl and the stock market just freaking 00:01:16.730 --> 00:01:18.489 out. Yeah. And then we have to talk about the 00:01:18.489 --> 00:01:21.569 elephant in the room, Google. While everyone 00:01:21.569 --> 00:01:24.469 was watching these shiny new agents, Google has 00:01:24.469 --> 00:01:27.189 been scaling something that is, frankly, hard 00:01:27.189 --> 00:01:29.790 to even comprehend. The financial reality check 00:01:29.790 --> 00:01:32.150 there is pretty staggering. And finally, we'll 00:01:32.150 --> 00:01:35.090 wrap up with some new tools in this mindset we're 00:01:35.090 --> 00:01:37.769 calling the cognitive gym. Because if the AI 00:01:37.769 --> 00:01:40.290 is getting smarter, we really need to make sure 00:01:40.290 --> 00:01:42.829 we aren't getting dumber. That's maybe the most 00:01:42.829 --> 00:01:44.670 critical part of this whole conversation for 00:01:44.670 --> 00:01:47.030 me. Okay, so let's dive in. All right, let's 00:01:47.030 --> 00:01:49.450 start with what everyone's calling Thursday AI. 00:01:49.930 --> 00:01:52.150 It feels like every Thursday the world changes. 00:01:52.209 --> 00:01:54.950 But this one, this one was special. OpenAI and 00:01:54.950 --> 00:01:57.269 Anthropic head to head. Oh, it was pure theater. 00:01:57.450 --> 00:02:01.650 So first you have OpenAI dropping GPT 5 .3 codecs. 00:02:01.689 --> 00:02:03.909 And, you know, for context, this is not just 00:02:03.909 --> 00:02:06.049 a slightly better chatbot. This is a model built 00:02:06.049 --> 00:02:08.689 to live inside your operating system. Right. 00:02:08.830 --> 00:02:11.930 It's powering a new macOS app. But I want to 00:02:11.930 --> 00:02:13.629 get this agentic part. We use that word a lot. 00:02:13.689 --> 00:02:15.689 What does it actually look like when you use 00:02:15.689 --> 00:02:18.569 this thing? Okay, so the old way, and by old 00:02:18.569 --> 00:02:21.610 I mean 2025, was you'd ask the bot, write me 00:02:21.610 --> 00:02:24.349 a function. It spits out code. You copy it, you 00:02:24.349 --> 00:02:27.449 paste it, you see if it works. Codex 5 .3 is 00:02:27.449 --> 00:02:29.830 different. It treats your entire computer like 00:02:29.830 --> 00:02:32.310 it's sandbox. The release notes talked about 00:02:32.310 --> 00:02:34.669 it handling complex app development, even game 00:02:34.669 --> 00:02:37.430 development, over multiple days. Multiple days. 00:02:37.590 --> 00:02:39.210 That's the part I can't wrap my head around. 00:02:39.389 --> 00:02:41.849 How does an AI remember what it was doing on 00:02:41.849 --> 00:02:44.150 Tuesday when it wakes up on Thursday? That's 00:02:44.150 --> 00:02:46.479 the leap. It's about more than just a big context 00:02:46.479 --> 00:02:49.439 window. They're calling it episodic memory. The 00:02:49.439 --> 00:02:51.939 model keeps a log of its own decisions, of its 00:02:51.939 --> 00:02:54.560 own intent, so it knows why it built the database 00:02:54.560 --> 00:02:56.439 that way yesterday so it can build the front 00:02:56.439 --> 00:02:59.020 end to match it today. And the headline feature, 00:02:59.219 --> 00:03:01.840 the self -debugging, you said it helped build 00:03:01.840 --> 00:03:03.780 itself. How does that really work? Is it just 00:03:03.780 --> 00:03:06.300 like a spellcheck for code? It's way deeper. 00:03:06.800 --> 00:03:09.580 So in traditional training, a human has to label 00:03:09.580 --> 00:03:12.419 the data, right? A person says, this code is 00:03:12.419 --> 00:03:15.120 wrong. That's slow. With this, the model generates 00:03:15.120 --> 00:03:17.539 code, tries to compile it, reads the error message, 00:03:17.800 --> 00:03:20.719 figures out the logical flaw, and then rewrites 00:03:20.719 --> 00:03:23.020 the code and tries again. It's a closed loop. 00:03:23.219 --> 00:03:25.360 So it's learning from its own failures in real 00:03:25.360 --> 00:03:28.210 time. At machine speed. Exactly. It's why it's 00:03:28.210 --> 00:03:31.509 25 % faster in execution. It's already failed 00:03:31.509 --> 00:03:33.750 a thousand times in a simulation before you even 00:03:33.750 --> 00:03:36.409 type your first prompt. Wow. Okay. So on one 00:03:36.409 --> 00:03:38.849 side, we've got OpenAI building the ultimate 00:03:38.849 --> 00:03:42.229 builder. But then Anthropic was not just sitting 00:03:42.229 --> 00:03:44.250 around. No. And this was the drama of it all. 00:03:44.550 --> 00:03:46.629 Anthropic pushed their launch by 15 minutes. 00:03:46.810 --> 00:03:48.949 It really felt like they were waiting for OpenAI 00:03:48.949 --> 00:03:50.629 to hit publish just so they could steal the news 00:03:50.629 --> 00:03:53.750 cycle. And they dropped Cloud Opus 4 .6. And 00:03:53.750 --> 00:03:56.069 they're calling it vibe coding? I saw that all 00:03:56.069 --> 00:03:57.930 over X. What does that even mean? Is that just 00:03:57.930 --> 00:04:00.830 marketing? It's a fun name, but it actually describes 00:04:00.830 --> 00:04:03.030 a real thing. It's about coding with natural 00:04:03.030 --> 00:04:06.210 language intent, not strict syntax. So you don't 00:04:06.210 --> 00:04:08.729 tell it how to write the code. You tell it the 00:04:08.729 --> 00:04:11.550 vibe of the app. I want a retro dashboard that 00:04:11.550 --> 00:04:14.710 feels like a 90s hacker movie. And it just, it 00:04:14.710 --> 00:04:17.680 translates that feeling into code. So it's bridging 00:04:17.680 --> 00:04:20.160 the gap between sort of art school instructions 00:04:20.160 --> 00:04:23.300 and computer science execution. Precisely. And 00:04:23.300 --> 00:04:26.500 while Codex is digging deep into the OS, Anthropic 00:04:26.500 --> 00:04:30.459 is going wide on data. Opus 4 .6 can pull info 00:04:30.459 --> 00:04:33.740 from huge document sets, run full financial analyses. 00:04:33.860 --> 00:04:36.100 I saw the stat here. It's number one on the finance 00:04:36.100 --> 00:04:39.199 agent benchmark. Beating OpenAI. Which is huge 00:04:39.199 --> 00:04:41.759 for the corporate world. While OpenAI is chasing 00:04:41.759 --> 00:04:44.100 developers, Anthropic is going right after the 00:04:44.100 --> 00:04:46.920 white -collar workflow. The analysts, the researchers, 00:04:47.259 --> 00:04:49.480 the people living in spreadsheets. So we have 00:04:49.480 --> 00:04:52.240 Codex building itself and Claude basically replacing 00:04:52.240 --> 00:04:55.680 the analyst. It makes me wonder, if the AI is 00:04:55.680 --> 00:04:59.139 doing all this, where do we even fit in? It's 00:04:59.139 --> 00:05:00.860 a multi -billion dollar question, isn't it? No, 00:05:00.860 --> 00:05:03.160 really. If the model can fix its own errors, 00:05:03.300 --> 00:05:05.459 where does a human developer actually fit into 00:05:05.459 --> 00:05:08.000 that loop? I think we stop being writers, you 00:05:08.000 --> 00:05:10.379 know, writers of code, writers of reports. We 00:05:10.379 --> 00:05:12.839 become architects of intent. We define the what 00:05:12.839 --> 00:05:15.740 and the why. And the agents handle the how. You 00:05:15.740 --> 00:05:17.519 don't lay the bricks anymore. You're designing 00:05:17.519 --> 00:05:20.779 the cathedral. So if the AI debugs itself, the 00:05:20.779 --> 00:05:23.360 human moves from being a writer to an architect. 00:05:23.639 --> 00:05:26.860 Yeah. We stop typing syntax and we start designing 00:05:26.860 --> 00:05:30.959 outcomes. But this tech rivalry, it didn't just 00:05:30.959 --> 00:05:33.360 stay on servers. It went right to the biggest 00:05:33.360 --> 00:05:37.389 stage in America. The Super Bowl, Super Bowl 00:05:37.389 --> 00:05:40.009 LX, I usually watch for the game, but the commercials 00:05:40.009 --> 00:05:43.970 this year were aggressive. Aggressive is one 00:05:43.970 --> 00:05:46.750 word for it. It was a culture war in 30 -second 00:05:46.750 --> 00:05:49.709 spots. Let's start with Anthropic's ad. It felt... 00:05:50.160 --> 00:05:52.860 Very pointed. It was a subtweet on national television. 00:05:53.160 --> 00:05:56.139 They ran this ad that subtly made fun of AI platforms 00:05:56.139 --> 00:05:59.079 cluttered with ads. It was a direct shot at OpenAI, 00:05:59.279 --> 00:06:01.139 who've been playing with sponsored results. And 00:06:01.139 --> 00:06:03.120 it definitely got a reaction. Oh, for sure. Sam 00:06:03.120 --> 00:06:05.339 Altman was on X, like, immediately. He called 00:06:05.339 --> 00:06:07.759 the ad clearly dishonest. You know the industry's 00:06:07.759 --> 00:06:09.540 getting serious when the CEOs are beefing during 00:06:09.540 --> 00:06:12.459 the halftime show. It felt very 90s Cola Wars. 00:06:12.980 --> 00:06:15.079 But then Google. Google took a totally different 00:06:15.079 --> 00:06:17.399 path. They didn't attack anybody. No, Google's 00:06:17.399 --> 00:06:21.740 Gemini ad was warm. It had bicycles, families, 00:06:22.040 --> 00:06:24.579 the Madden Bowl. They were just selling helpful 00:06:24.579 --> 00:06:27.360 AI to everyone. It was like they were saying, 00:06:27.480 --> 00:06:29.579 hey, let the startups fight. We're just here 00:06:29.579 --> 00:06:31.240 to help you find a cookie recipe. That's their 00:06:31.240 --> 00:06:34.920 entire play. Normalization. Make AI feel as safe 00:06:34.920 --> 00:06:37.639 and boring as a Google search. But while the 00:06:37.639 --> 00:06:40.100 ads were going on, the stock market was reacting 00:06:40.100 --> 00:06:43.120 to the actual tech. And that was a lot less warm 00:06:43.120 --> 00:06:45.100 and fuzzy. Yeah. Let's talk about the panic. 00:06:45.459 --> 00:06:49.379 Anthropic drops 11 Claude Cowork plugins. And 00:06:49.379 --> 00:06:52.220 Wall Street just loses its mind. Why did software 00:06:52.220 --> 00:06:54.839 stocks tank so hard? It's the implication of 00:06:54.839 --> 00:06:57.160 those plugins. The Cowork tools let the AI plug 00:06:57.160 --> 00:07:10.100 directly into your company. So it was a sell 00:07:10.100 --> 00:07:12.300 -off based on the idea that these AI agents could 00:07:12.300 --> 00:07:14.079 just kill the whole software as a service model. 00:07:14.300 --> 00:07:17.100 Exactly. Why rent a tool for a human when you 00:07:17.100 --> 00:07:20.100 can just rent an agent to do the job? But... 00:07:20.439 --> 00:07:22.759 You know, Jensen Hong from NVIDIA called the 00:07:22.759 --> 00:07:25.839 whole thing illogical. Well, of course he did. 00:07:26.040 --> 00:07:28.360 He sells the chips that run everything. He has 00:07:28.360 --> 00:07:31.759 a bias, for sure, but he's got a point. The market 00:07:31.759 --> 00:07:34.399 is reacting to the fear of being obsolete, not 00:07:34.399 --> 00:07:37.620 the immediate reality. But then, to counter that 00:07:37.620 --> 00:07:41.029 fear, OpenAI launched Frontier. Right there. 00:07:41.069 --> 00:07:42.810 Enterprise platform. How does that calm things 00:07:42.810 --> 00:07:45.790 down? Well, it turns agents into coworkers that 00:07:45.790 --> 00:07:48.769 can use files and run code, but securely inside 00:07:48.769 --> 00:07:50.649 a corporate environment. They're trying to say, 00:07:50.750 --> 00:07:53.589 hey, we are destroying enterprise software. We're 00:07:53.589 --> 00:07:56.350 becoming the platform it runs on. It feels like 00:07:56.350 --> 00:07:58.870 we're in this limbo where the technology is moving 00:07:58.870 --> 00:08:00.970 way faster than the business models can adapt. 00:08:01.230 --> 00:08:03.529 That is a massive understatement. So with all 00:08:03.529 --> 00:08:06.389 this volatility, is the stock market panic justified 00:08:06.389 --> 00:08:09.750 or is it just noise? It's panic for sure, but 00:08:09.750 --> 00:08:12.449 it's a signal the SaaS era is ending. Speaking 00:08:12.449 --> 00:08:14.850 of ending eras, let's pivot. Let's talk about 00:08:14.850 --> 00:08:16.629 the giant that everyone keeps underestimating 00:08:16.629 --> 00:08:18.290 because they aren't always the coolest. Let's 00:08:18.290 --> 00:08:20.649 talk about Google. The empire strikes back. I'm 00:08:20.649 --> 00:08:23.350 serious. We get so caught up in the open AI and 00:08:23.350 --> 00:08:25.889 anthropic drama. Yeah. But then you look at Google's 00:08:25.889 --> 00:08:29.069 Q4 earnings call. Yeah. And it's just brute force. 00:08:29.350 --> 00:08:31.970 It really is the difference between product elegance 00:08:31.970 --> 00:08:34.490 and just logistical dominance. Let's run the 00:08:34.490 --> 00:08:36.330 numbers because they're insane. Yeah, lay them 00:08:36.330 --> 00:08:41.509 on me. Okay. The Gemini app. 750 million monthly 00:08:41.509 --> 00:08:44.480 active users. That's more than TikTok in some 00:08:44.480 --> 00:08:47.320 places. 750 million. That's a lot of people asking 00:08:47.320 --> 00:08:49.840 how to bake bread. And it's not just that. Enterprise. 00:08:50.200 --> 00:08:54.580 8 million paid seats in four months. 95 % of 00:08:54.580 --> 00:08:57.340 the top 20 SaaS companies are using it. Over 00:08:57.340 --> 00:09:01.379 5 billion interactions in Q4 alone. 5 billion. 00:09:01.500 --> 00:09:03.299 And there was that detail about how people are 00:09:03.299 --> 00:09:04.759 using it, right? It wasn't just text. Right. 00:09:04.840 --> 00:09:07.960 One in six AI mode searches now use voice or 00:09:07.960 --> 00:09:10.240 image. That's a huge behavioral shift. Yeah. 00:09:10.320 --> 00:09:12.059 People are showing Google a picture of a broken 00:09:12.059 --> 00:09:14.659 pipe. and asking, how do I fix this? In the infrastructure, 00:09:14.940 --> 00:09:17.419 I mean, OpenAI has Microsoft, but Google basically 00:09:17.419 --> 00:09:20.120 is the internet. That's their moat. Gemini 3 00:09:20.120 --> 00:09:22.679 Pro is the fastest adopted model in their history. 00:09:22.940 --> 00:09:25.960 Their new workbench, anti -gravity, got 1 .5 00:09:25.960 --> 00:09:28.580 million users in two months. But here's the number, 00:09:28.659 --> 00:09:30.840 the one that stops the conversation, capital 00:09:30.840 --> 00:09:33.360 expenditure. The money they're spending to build 00:09:33.360 --> 00:09:36.519 all this. Google's planned CapEx for 2026 is 00:09:36.519 --> 00:09:42.549 between 175 and 185. Billion dollars. 185 billion. 00:09:42.690 --> 00:09:44.429 I can't even. What does that money even buy? 00:09:44.549 --> 00:09:46.990 It's not just servers. It's energy. It's physics. 00:09:47.190 --> 00:09:49.889 They're securing fusion power contracts, upgrading 00:09:49.889 --> 00:09:52.509 national power grids, building custom chips. 00:09:52.629 --> 00:09:54.909 This is nation building money. This is the Manhattan 00:09:54.909 --> 00:09:57.370 Project times 10. It puts everything in perspective. 00:09:57.690 --> 00:10:01.750 Open AI feels like magic. Google feels like gravity, 00:10:01.909 --> 00:10:04.789 an unstoppable force. Open AI has elegance. Google 00:10:04.789 --> 00:10:07.529 has distribution. Basically, infinite cash. So 00:10:07.529 --> 00:10:09.149 does product elegance even matter if you can 00:10:09.149 --> 00:10:12.289 spend $185 billion on infrastructure? Distribution 00:10:12.289 --> 00:10:15.370 beats elegance every time. History repeats itself. 00:10:15.690 --> 00:10:18.470 Okay, let's bring this back down to earth. We've 00:10:18.470 --> 00:10:20.990 talked about billions of dollars in global infrastructure, 00:10:21.250 --> 00:10:23.330 but because they're spending all that money, 00:10:23.450 --> 00:10:27.350 the cost for us, for the regular person to create 00:10:27.350 --> 00:10:29.830 something, is dropping to almost zero. That's 00:10:29.830 --> 00:10:31.429 the other side of the coin. It's a golden age 00:10:31.429 --> 00:10:34.220 for creative tools. Yeah. Despite the Titans 00:10:34.220 --> 00:10:36.059 fighting, there's some incredible stuff coming 00:10:36.059 --> 00:10:38.639 out for creators. Let's do a quick rapid fire 00:10:38.639 --> 00:10:42.259 on some new tools. First up, Kling 3 .0. Huge 00:10:42.259 --> 00:10:45.559 for video. We're talking native 4K outputs, longer 00:10:45.559 --> 00:10:47.879 video times. You can basically make short films 00:10:47.879 --> 00:10:50.759 with this now. It's not just a four -second blurry 00:10:50.759 --> 00:10:53.320 clip anymore. Webflow. Big one for designers. 00:10:53.419 --> 00:10:55.720 You give it a prompt, it builds a multi -page 00:10:55.720 --> 00:10:58.649 production -ready website. Not a mock -up. a 00:10:58.649 --> 00:11:01.690 real site higgs field vibe motion yeah motion 00:11:01.690 --> 00:11:04.529 images from a single prompt it gives you incredible 00:11:04.529 --> 00:11:07.809 control if you need a specific asset and superboard 00:11:07.809 --> 00:11:10.789 this one's amazing connects over 600 data sources 00:11:10.789 --> 00:11:13.009 it's like having a team of data analysts in your 00:11:13.009 --> 00:11:15.049 pocket you connect your stripe youtube whatever 00:11:15.049 --> 00:11:17.470 and just ask it questions across all of them 00:11:17.470 --> 00:11:19.970 these tools are incredible but they bring us 00:11:19.970 --> 00:11:23.309 to the uh the philosophical part of this the 00:11:23.309 --> 00:11:25.610 thing we're calling the cognitive gym I love 00:11:25.610 --> 00:11:28.110 this concept because there's a real danger here. 00:11:28.250 --> 00:11:30.950 The danger that we all just get. You lazy. Worse. 00:11:31.289 --> 00:11:34.570 That we get functionally dumber. If we outsource 00:11:34.570 --> 00:11:37.590 all our thinking. If we let Codex write the code 00:11:37.590 --> 00:11:40.309 and Claude do the research. What happens to our 00:11:40.309 --> 00:11:42.490 own brains? It's like using GPS all the time. 00:11:42.509 --> 00:11:44.830 Nobody knows how to navigate anymore. I have 00:11:44.830 --> 00:11:46.789 to admit something here. A bit of a vulnerable 00:11:46.789 --> 00:11:49.610 admission. I've felt this myself. I call it prompt 00:11:49.610 --> 00:11:52.549 drift. I've been letting the AI summarize articles 00:11:52.549 --> 00:11:55.210 for me instead of reading them. And last week, 00:11:55.230 --> 00:11:57.629 I realized I couldn't actually remember the details 00:11:57.629 --> 00:12:00.850 of something I'd supposedly learned. I retained 00:12:00.850 --> 00:12:02.629 less information because I did less of the work. 00:12:02.730 --> 00:12:04.889 That's the trap. Exactly. The friction is where 00:12:04.889 --> 00:12:07.269 the learning happens. The cognitive gym idea 00:12:07.269 --> 00:12:10.149 is the answer. You have to use the AI to increase 00:12:10.149 --> 00:12:12.409 the difficulty, not take it away. How does that 00:12:12.409 --> 00:12:14.269 even work? Give me an example. How do I bench 00:12:14.269 --> 00:12:16.990 press with an AI? Instead of asking the AI for 00:12:16.990 --> 00:12:20.190 the answer, you ask it to quiz you. Instead of 00:12:20.190 --> 00:12:22.570 asking it to write the draft, you write it, and 00:12:22.570 --> 00:12:25.149 then you ask the AI to completely tear your logic 00:12:25.149 --> 00:12:27.590 apart. You force it to debate you. You have to 00:12:27.590 --> 00:12:29.850 lift the weight yourself. So the AI becomes the 00:12:29.850 --> 00:12:32.409 spotter, not the weightlifter. Exactly. And we're 00:12:32.409 --> 00:12:34.730 seeing this trend with solo founders. They're 00:12:34.730 --> 00:12:37.970 using AI to do the work of 10 people, but they're 00:12:37.970 --> 00:12:40.250 still the one driving the vision. So how do we 00:12:40.250 --> 00:12:43.350 stop from becoming just passive consumers of 00:12:43.350 --> 00:12:46.509 AI output? You have to treat the AI as a sparring 00:12:46.509 --> 00:12:49.509 partner, not a butler. We'll be right back. After 00:12:49.509 --> 00:12:54.009 a quick word. And we're back. Okay, let's recap 00:12:54.009 --> 00:12:56.470 the big ideas from today. It's been a heavy week. 00:12:56.529 --> 00:12:59.129 A historic one, I think. First, the agent wars. 00:12:59.610 --> 00:13:02.389 Codex is literally building itself. Glaude is 00:13:02.389 --> 00:13:04.389 aiming to replace entire white -collar workflows. 00:13:04.870 --> 00:13:08.070 We are moving well past chatbots. Second, the 00:13:08.070 --> 00:13:10.769 culture is splitting. The Super Bowl ads showed 00:13:10.769 --> 00:13:12.490 us the tension. The market showed us the fear. 00:13:12.649 --> 00:13:15.710 The entire SaaS model is under threat. Third, 00:13:15.850 --> 00:13:20.059 Google. The quiet giant. While we watch the startups, 00:13:20.320 --> 00:13:24.259 Google is spending $185 billion to build an infrastructure 00:13:24.259 --> 00:13:26.720 mode that might be impossible for anyone else 00:13:26.720 --> 00:13:29.909 to cross. And finally, the cognitive gem. The 00:13:29.909 --> 00:13:32.230 tools are getting incredibly powerful, which 00:13:32.230 --> 00:13:34.110 means we have to be more disciplined. We have 00:13:34.110 --> 00:13:36.409 to make sure we stay the architects of intent. 00:13:36.750 --> 00:13:39.490 We are absolutely moving from chatting with bots 00:13:39.490 --> 00:13:42.789 to managing swarms of agents. That's the new 00:13:42.789 --> 00:13:45.470 reality. If you want to try out Kling or Superboard 00:13:45.470 --> 00:13:47.590 or any of the tools we talked about, check the 00:13:47.590 --> 00:13:49.190 show notes. We've got links for you there. They're 00:13:49.190 --> 00:13:51.000 definitely worth experimenting with. I want to 00:13:51.000 --> 00:13:52.799 leave you with one final thought. We mentioned 00:13:52.799 --> 00:13:56.600 11 labs raised over 400 million euros, more than 00:13:56.600 --> 00:13:59.120 all of Europe combined in AI funding recently. 00:13:59.559 --> 00:14:02.779 And Google is spending 185 billion. The scale 00:14:02.779 --> 00:14:05.700 of the numbers is just, it's astronomical. So 00:14:05.700 --> 00:14:07.919 if all that capital is concentrating so intensely, 00:14:08.139 --> 00:14:10.799 are we just watching the consolidation of human 00:14:10.799 --> 00:14:13.299 intelligence into three or four zip codes? And 00:14:13.299 --> 00:14:16.279 if that's true, what does that mean for your 00:14:16.279 --> 00:14:19.100 solo business in 2027? That's the question that 00:14:19.100 --> 00:14:21.009 keeps me up at night. Thanks for listening to 00:14:21.009 --> 00:14:22.690 the Deep Dive. We'll see you next time.