WEBVTT 00:00:00.000 --> 00:00:02.819 We are watching something pretty profound happen 00:00:02.819 --> 00:00:05.339 right now. It looks like the systematic formation 00:00:05.339 --> 00:00:08.859 of, well, maybe a half trillion dollar monopoly. 00:00:09.199 --> 00:00:10.919 And it's not being built through the usual M 00:00:10.919 --> 00:00:13.140 &A playbook, right? It's pure platform architecture. 00:00:13.580 --> 00:00:15.960 Exactly. OpenAI, they're not just satisfied making 00:00:15.960 --> 00:00:19.019 better chatbots or language models. They seem 00:00:19.019 --> 00:00:21.920 to be systematically replacing whole startup 00:00:21.920 --> 00:00:26.620 ecosystems. It feels like they want ChatGPT to 00:00:26.620 --> 00:00:29.940 be the next sort of core. digital operating system. 00:00:30.039 --> 00:00:32.960 Uh -huh. Yeah. It's definitely aggressive, brilliant, 00:00:33.159 --> 00:00:36.700 maybe, in a ruthless kind of way. And their latest 00:00:36.700 --> 00:00:38.979 Dev Day announcements really just confirmed it 00:00:38.979 --> 00:00:41.439 all. Yeah. They've basically perfected this playbook 00:00:41.439 --> 00:00:43.240 we've started calling Watch, Learn, Replace. 00:00:43.420 --> 00:00:46.759 And the timeline, it's speeding up a lot. Which 00:00:46.759 --> 00:00:49.200 is exactly why we need this deep dive. We really 00:00:49.200 --> 00:00:50.960 have to look closely at the sources talking about 00:00:50.960 --> 00:00:52.640 this transformation. And maybe more importantly, 00:00:52.719 --> 00:00:54.740 figure out what actions we need to take right 00:00:54.740 --> 00:00:58.219 now to adapt and survive. Makes sense. So today, 00:00:58.649 --> 00:01:00.789 We're going to unpack the three big pillars of 00:01:00.789 --> 00:01:04.590 this strategy. The apps SDK, AgentKit, and ChatKit. 00:01:04.689 --> 00:01:07.950 Then we'll dig into the almost superhuman tech 00:01:07.950 --> 00:01:11.090 engine that's driving their speed. And finally... 00:01:11.340 --> 00:01:13.420 Look at the strategic response that's probably 00:01:13.420 --> 00:01:15.519 required. OK, let's start with the scale just 00:01:15.519 --> 00:01:17.420 to set the stage, because this whole strategy, 00:01:17.459 --> 00:01:20.079 it rests on a foundation that's already huge. 00:01:20.239 --> 00:01:22.939 We're talking a potential $500 billion valuation 00:01:22.939 --> 00:01:27.980 and maybe 800 million weekly active users. That 00:01:27.980 --> 00:01:30.560 user base, that's the leverage. It's the core 00:01:30.560 --> 00:01:32.700 piece for all their market domination plans. 00:01:32.959 --> 00:01:35.359 And that mechanism they use, the watch. learn 00:01:35.359 --> 00:01:37.480 replace playbook you mentioned it just sounds 00:01:37.480 --> 00:01:39.939 devastating for smaller players the sources seem 00:01:39.939 --> 00:01:41.819 to lay out a pretty clear pattern four steps 00:01:41.819 --> 00:01:44.519 yeah it's almost like clockwork step one release 00:01:44.519 --> 00:01:47.219 apis get developers building on their tech step 00:01:47.219 --> 00:01:50.560 two watch closely monitor usage see what features 00:01:50.560 --> 00:01:52.920 people actually want essentially let the market 00:01:52.920 --> 00:01:55.519 do the r d for them okay step three identify 00:01:55.519 --> 00:01:57.620 the winners the products that really take off 00:01:57.620 --> 00:02:00.420 find that product market fit And step four. Attack 00:02:00.420 --> 00:02:02.980 and replace. Build a native, probably better 00:02:02.980 --> 00:02:05.500 version right into the open AI platform. Game 00:02:05.500 --> 00:02:07.739 over for the original developer often. That is 00:02:07.739 --> 00:02:09.680 a tough blueprint to compete against. So, OK, 00:02:09.780 --> 00:02:11.479 let's look at the infrastructure they're actually 00:02:11.479 --> 00:02:13.360 using to pull this off these three new pillars. 00:02:13.500 --> 00:02:16.180 Right. Starting with the apps SDK. This is the 00:02:16.180 --> 00:02:19.439 big one that aims to turn chat GPT into like 00:02:19.439 --> 00:02:22.659 the next app store. It's a huge shift in how 00:02:22.659 --> 00:02:25.060 we might interact with services. How so? Well, 00:02:25.099 --> 00:02:28.490 think about it. You, the user, you might not 00:02:28.490 --> 00:02:31.629 bother going to, say, booking .com anymore. Okay. 00:02:31.689 --> 00:02:34.250 You'll just ask ChatGPT to handle your travel 00:02:34.250 --> 00:02:38.129 booking. And ChatGPT will use booking .com, but 00:02:38.129 --> 00:02:40.210 just as a back -end plug -in. You never actually 00:02:40.210 --> 00:02:43.030 leave the chat. But hold on. Haven't we seen 00:02:43.030 --> 00:02:45.449 things like this before? Google Assistants, Alexa, 00:02:45.770 --> 00:02:48.669 they tried to be that central hub. Why would 00:02:48.669 --> 00:02:51.110 this succeed where they kind of stalled? I think 00:02:51.110 --> 00:02:54.580 the difference is agency. And much, much better 00:02:54.580 --> 00:02:56.759 language understanding. The older platforms, 00:02:56.900 --> 00:02:59.639 they were pretty much limited to specific voice 00:02:59.639 --> 00:03:02.060 commands, right? Or these fixed skills. Very 00:03:02.060 --> 00:03:05.340 rigid. Right. Very limited scope. Yeah. ChatGPT 00:03:05.340 --> 00:03:08.479 can grasp intent across really complex multi 00:03:08.479 --> 00:03:12.159 -step tasks. That turns it into a genuinely flexible, 00:03:12.460 --> 00:03:15.199 well, operating environment, not just a voice 00:03:15.199 --> 00:03:18.060 remote. So the website itself, the destination, 00:03:18.360 --> 00:03:20.240 becomes less important than its function inside 00:03:20.240 --> 00:03:23.610 this AI gateway. Yeah. Interesting. Okay, what 00:03:23.610 --> 00:03:26.990 about Pillar 2? Agent Kit. Agent Kit. That's 00:03:26.990 --> 00:03:29.490 OpenAI's direct shot at the automation platforms. 00:03:29.629 --> 00:03:32.050 Think Zapier, Make .com, that sort of thing. 00:03:32.270 --> 00:03:34.689 It's a full platform for building AI agents, 00:03:34.830 --> 00:03:37.949 but the standout feature is this really powerful 00:03:37.949 --> 00:03:41.490 visual workflow builder. Visual. Yeah, like drag 00:03:41.490 --> 00:03:43.169 and drop. You're essentially stacking together 00:03:43.169 --> 00:03:45.750 different AI capabilities and data sources, like 00:03:45.750 --> 00:03:48.090 building with digital Lego blocks. It suddenly 00:03:48.090 --> 00:03:49.949 makes building complex automations accessible 00:03:49.949 --> 00:03:52.680 to almost anyone. Wow. Visual automation tools, 00:03:52.740 --> 00:03:54.960 but backed by the power and models of open AI. 00:03:55.120 --> 00:03:57.259 That's significant. And pillar three, ChatKit. 00:03:57.379 --> 00:03:59.539 ChatKit is all about embedding AI intelligence 00:03:59.539 --> 00:04:02.620 everywhere. It lets you put a customizable chat 00:04:02.620 --> 00:04:05.139 interface onto any website or app. Like a standard 00:04:05.139 --> 00:04:08.319 chatbot embed. Sort of, but the really revolutionary 00:04:08.319 --> 00:04:11.210 part is something they call widgets. These aren't 00:04:11.210 --> 00:04:13.849 just text bubbles. They're interactive UI components. 00:04:13.969 --> 00:04:17.290 Think calendars, charts, even payment forms that 00:04:17.290 --> 00:04:19.990 live and function directly inside the chat window. 00:04:20.149 --> 00:04:23.750 Whoa. Yeah. It transforms a basic chat bot into 00:04:23.750 --> 00:04:26.949 this dynamic little mini application, all contained 00:04:26.949 --> 00:04:28.990 within the conversation itself. Okay, putting 00:04:28.990 --> 00:04:31.740 it all together. Apps SDK keeps you in the chat. 00:04:31.899 --> 00:04:34.899 AgentKit lets them build replacements fast. And 00:04:34.899 --> 00:04:37.759 ChatKit embeds this experience everywhere. The 00:04:37.759 --> 00:04:39.939 goal seems to be containing the entire user journey 00:04:39.939 --> 00:04:42.540 inside their AI environment. Precisely. It's 00:04:42.540 --> 00:04:44.339 a ruthless strategy, all right? Yeah. But the 00:04:44.339 --> 00:04:47.579 execution, it requires... unbelievable speed 00:04:47.579 --> 00:04:49.959 if they're replacing entire successful products 00:04:49.959 --> 00:04:51.819 they must be shipping code at rates we haven't 00:04:51.819 --> 00:04:53.899 seen before yeah what's the engine behind that 00:04:53.899 --> 00:04:56.019 speed how are they moving so fast it really is 00:04:56.019 --> 00:04:57.839 the million dollar question isn't it or maybe 00:04:57.839 --> 00:05:00.000 the 500 billion dollar question and the answer 00:05:00.000 --> 00:05:03.220 is they're using their own extremely advanced 00:05:03.220 --> 00:05:07.720 internal ai tools to build everything ah so the 00:05:07.720 --> 00:05:10.980 ai is building the ai the public stuff is almost 00:05:10.980 --> 00:05:13.519 a side effect of their internal capability can 00:05:13.519 --> 00:05:16.189 you walk us through what codex is exactly Yeah. 00:05:16.269 --> 00:05:18.529 Codex is basically their secret weapon. It's 00:05:18.529 --> 00:05:20.990 their internal coding assistant or maybe partner 00:05:20.990 --> 00:05:22.949 is a better word. And it's reportedly running 00:05:22.949 --> 00:05:26.329 on a highly optimized version of GPT -5. Okay. 00:05:26.509 --> 00:05:29.509 And it's way beyond just generating code snippets. 00:05:29.509 --> 00:05:32.410 It functions like a really complex AI development 00:05:32.410 --> 00:05:35.110 team member. A superhuman team member from the 00:05:35.110 --> 00:05:37.750 sound of it. Pretty much, yeah. Think about what 00:05:37.750 --> 00:05:40.629 the sources say it can do. automated, complex 00:05:40.629 --> 00:05:43.370 code reviews. That alone massively speeds up 00:05:43.370 --> 00:05:46.050 quality control. It can handle refactoring tasks 00:05:46.050 --> 00:05:48.769 that might take a human developer hours, maybe 00:05:48.769 --> 00:05:52.149 even days, tedious, error -prone work, and do 00:05:52.149 --> 00:05:54.509 it in minutes. Multi -hour refactors in minutes. 00:05:54.649 --> 00:05:57.490 Wow. Get this. It can apparently manage over 00:05:57.490 --> 00:06:00.389 20 distinct development tasks asynchronously, 00:06:00.550 --> 00:06:02.790 working on different parts of a huge code base 00:06:02.790 --> 00:06:04.889 all at the same time. That's incredible. And 00:06:04.889 --> 00:06:06.769 they even released an open source SDK for it. 00:06:06.769 --> 00:06:08.529 Why would they do that? That's the strategic 00:06:08.529 --> 00:06:12.420 master. By open sourcing parts of Codex, they're 00:06:12.420 --> 00:06:14.920 essentially encouraging the entire industry to 00:06:14.920 --> 00:06:17.379 adopt their way of building software super fast, 00:06:17.540 --> 00:06:20.480 AI -assisted development. It standardizes their 00:06:20.480 --> 00:06:22.899 methods. And the proof is in the pudding, as 00:06:22.899 --> 00:06:26.220 they say. Agent Kit. Exactly. That whole sophisticated 00:06:26.220 --> 00:06:28.259 agent building platform we were just talking 00:06:28.259 --> 00:06:30.420 about, the one with the visual builder, built 00:06:30.420 --> 00:06:33.600 in six weeks. Beat. Yeah. Using these internal 00:06:33.600 --> 00:06:37.920 AI tools. Six. Yeah, that's the reality check 00:06:37.920 --> 00:06:40.019 for every developer out there. It hammers home 00:06:40.019 --> 00:06:43.259 the point. You probably can't outship AI if you're 00:06:43.259 --> 00:06:45.240 not using AI tools yourself. It's becoming table 00:06:45.240 --> 00:06:47.519 stakes. Okay, let's look a bit closer at Agent 00:06:47.519 --> 00:06:50.000 Kit then. People can access it now. Yep. It's 00:06:50.000 --> 00:06:53.620 available through platform .opi .com. And they've 00:06:53.620 --> 00:06:55.819 included some useful pre -built templates to 00:06:55.819 --> 00:06:58.399 get started things for data enrichment, a planning 00:06:58.399 --> 00:07:01.360 helper, customer service bots, document comparison. 00:07:01.779 --> 00:07:03.680 Common use case. How does it connect to everything, 00:07:03.800 --> 00:07:05.779 though, to all the other apps and services people 00:07:05.779 --> 00:07:08.399 use? Ah, that's where this thing called MCP Model 00:07:08.399 --> 00:07:12.199 Context Protocol comes in. Think of MCP as like 00:07:12.199 --> 00:07:15.939 a universal translator and security layer. For 00:07:15.939 --> 00:07:18.819 AI agents. Okay. It standardizes how AgentKit 00:07:18.819 --> 00:07:21.379 can securely talk to thousands of different third 00:07:21.379 --> 00:07:23.379 -party apps without needing custom code for each 00:07:23.379 --> 00:07:25.660 one. So for instance, a service provider like 00:07:25.660 --> 00:07:29.579 Rube MCP gives you access to like over 500 apps, 00:07:29.800 --> 00:07:33.279 Gmail, Slack, GitHub, Supabase, you name it, 00:07:33.279 --> 00:07:35.639 all through one connection setup in AgentKit. 00:07:35.779 --> 00:07:38.910 So instant integration power. Huge library access. 00:07:39.110 --> 00:07:41.870 Bingo. Rapid, broad connectivity. That's crucial 00:07:41.870 --> 00:07:44.850 for building useful agents quickly. Now, something 00:07:44.850 --> 00:07:47.509 else that caught my eye in the chat kit documentation. 00:07:47.790 --> 00:07:49.069 You mentioned finding something interesting. 00:07:49.310 --> 00:07:51.389 Oh, yeah. It's kind of a hidden treasure buried 00:07:51.389 --> 00:07:53.430 in the docs. It's the entire prompt they use 00:07:53.430 --> 00:07:56.189 for their widget builder agent. And it is massive. 00:07:56.269 --> 00:07:59.589 We're talking 55 ,000 characters long. 55 ,000 00:07:59.589 --> 00:08:03.610 characters for one prompt. Beat. Wow. I mean, 00:08:03.629 --> 00:08:05.730 honestly, I still wrestle with prompt drift myself 00:08:05.730 --> 00:08:08.170 when I try to build anything complex. Just getting 00:08:08.170 --> 00:08:10.769 AI to behave reliably, consistently, it's a real 00:08:10.769 --> 00:08:13.810 challenge. It absolutely is. And this prompt, 00:08:13.949 --> 00:08:16.250 it's basically a masterclass in advanced prompt 00:08:16.250 --> 00:08:18.730 engineering. It's not just long for the sake 00:08:18.730 --> 00:08:21.389 of being long. Right. It's packed with layers 00:08:21.389 --> 00:08:23.550 of instructions for things like error handling, 00:08:23.810 --> 00:08:26.370 really sophisticated input validation, complex 00:08:26.370 --> 00:08:29.170 logic, even strategies for how different internal 00:08:29.170 --> 00:08:31.629 subagents should coordinate. It's essentially 00:08:31.629 --> 00:08:33.710 the... architectural blueprint for building a 00:08:33.710 --> 00:08:36.669 truly reliable, complex AI system that has to 00:08:36.669 --> 00:08:39.289 control interactive UI elements. So if these 00:08:39.289 --> 00:08:41.669 tools like AgentKit are becoming so powerful 00:08:41.669 --> 00:08:45.789 and fast, is mastering this kind of deep prompt 00:08:45.789 --> 00:08:48.009 engineering, maybe even specifically studying 00:08:48.009 --> 00:08:50.990 that giant 55 ,000 character example, is that 00:08:50.990 --> 00:08:53.370 now the most critical skill for developers who 00:08:53.370 --> 00:08:56.629 want to build serious AI applications? I'd say 00:08:56.629 --> 00:08:59.370 yes, absolutely. That deep mastery of prompting, 00:08:59.370 --> 00:09:01.090 understanding how to structure these complex 00:09:01.090 --> 00:09:03.509 instructions and constraints, it's becoming essential. 00:09:03.750 --> 00:09:05.289 It's really the difference between building a 00:09:05.289 --> 00:09:07.950 robust enterprise grade system versus just, you 00:09:07.950 --> 00:09:09.809 know, a fun little demo bot that breaks easily. 00:09:10.029 --> 00:09:13.710 This speed, this power, it obviously goes beyond 00:09:13.710 --> 00:09:15.909 just making software faster. It points towards 00:09:15.909 --> 00:09:18.549 much bigger societal changes. Job displacement 00:09:18.549 --> 00:09:21.679 comes up a lot. Yeah, and the sources. They don't 00:09:21.679 --> 00:09:23.620 shy away from it. It's the uncomfortable truth, 00:09:23.840 --> 00:09:25.240 right? What kind of jobs are we talking about? 00:09:25.360 --> 00:09:28.120 Well, the examples OpenAI themselves used were 00:09:28.120 --> 00:09:30.340 pretty direct. Things like camera operators, 00:09:30.840 --> 00:09:34.799 translators, even radiologists, potentially teachers. 00:09:34.919 --> 00:09:38.080 They demoed these hyper -personalized AI tutors. 00:09:38.379 --> 00:09:40.639 The analogy that really hit home for me was the 00:09:40.639 --> 00:09:43.340 real estate agent example. They showed how an 00:09:43.340 --> 00:09:45.919 agent could be created to give tours of, say, 00:09:46.000 --> 00:09:48.539 the Palace of Versailles. The implication is... 00:09:48.940 --> 00:09:51.139 Well, you could potentially replace every single 00:09:51.139 --> 00:09:54.080 human tour guide there with AI. Wow. It points 00:09:54.080 --> 00:09:56.580 to this idea that any job relying heavily on 00:09:56.580 --> 00:09:59.379 informational delivery or routine cognitive tasks, 00:09:59.679 --> 00:10:02.779 stuff that isn't truly unique human insight or 00:10:02.779 --> 00:10:04.919 interaction, is now potentially vulnerable. And 00:10:04.919 --> 00:10:06.779 this connects to an idea mentioned in the sources, 00:10:07.019 --> 00:10:09.779 the great divide. What's that about? It's this 00:10:09.779 --> 00:10:11.860 potential future where society kind of splits 00:10:11.860 --> 00:10:14.379 into two groups based purely on how people interact 00:10:14.379 --> 00:10:17.460 with AI. Okay, two groups. On one side, you've 00:10:17.460 --> 00:10:19.529 got the superhuman. These are the people who 00:10:19.529 --> 00:10:21.909 embrace the tools. They automate tasks, maybe 00:10:21.909 --> 00:10:24.870 build custom software 50 times faster using agent 00:10:24.870 --> 00:10:28.129 kit, use AI for strategic thinking. They become 00:10:28.129 --> 00:10:30.210 dramatically more effective, more productive. 00:10:30.330 --> 00:10:32.629 And the other side. The other side is potentially 00:10:32.629 --> 00:10:36.620 the addicted. Yeah, the risk here is people getting 00:10:36.620 --> 00:10:40.080 trapped, consuming endless streams of AI generated 00:10:40.080 --> 00:10:43.820 content, what some call digital slop. Think infinite, 00:10:43.919 --> 00:10:46.559 perfectly optimized, but ultimately empty SEO 00:10:46.559 --> 00:10:50.200 articles or those weirdly smooth, generic TikTok 00:10:50.200 --> 00:10:53.379 videos. Content that's increasingly stimulating 00:10:53.379 --> 00:10:56.279 but provides little real value or nourishment 00:10:56.279 --> 00:10:58.659 and potentially falling prey to sophisticated 00:10:58.659 --> 00:11:01.179 misinformation. And the digital swap problem 00:11:01.179 --> 00:11:03.179 is about to get way worse, isn't it? With tools 00:11:03.179 --> 00:11:06.919 like... Exactly. Open AI releasing Sora 2 via 00:11:06.919 --> 00:11:09.980 an API. That's a game changer for video. We're 00:11:09.980 --> 00:11:11.940 rapidly approaching the point where AI generated 00:11:11.940 --> 00:11:14.379 video will be truly indistinguishable from real 00:11:14.379 --> 00:11:17.000 footage for most people. Whoa, just imagine scaling 00:11:17.000 --> 00:11:19.279 that. A billion users each getting perfectly 00:11:19.279 --> 00:11:21.820 realistic, personalized video content on demand. 00:11:22.019 --> 00:11:24.850 That's... Mind boggling. It is. And you see major 00:11:24.850 --> 00:11:27.470 creators, people like Mr. Beast, publicly saying 00:11:27.470 --> 00:11:31.110 this stuff poses an existential threat to traditional 00:11:31.110 --> 00:11:34.450 human content creation. How do you compete with 00:11:34.450 --> 00:11:37.830 infinite, free, perfectly tailored video? So 00:11:37.830 --> 00:11:40.509 what's the survival strategy then for creators, 00:11:40.629 --> 00:11:43.190 for anyone whose work involve content? the sources 00:11:43.190 --> 00:11:46.309 suggest a mandatory pivot focus exclusively on 00:11:46.309 --> 00:11:49.409 education and unique human insight meaning meaning 00:11:49.409 --> 00:11:51.730 content that offers genuinely novel analysis 00:11:51.730 --> 00:11:55.190 deep expertise born from real experience or a 00:11:55.190 --> 00:11:58.230 truly authentic human perspective stuff ai can't 00:11:58.230 --> 00:12:00.769 replicate easily anything less generic information 00:12:00.769 --> 00:12:02.870 surface level commentary is likely going to be 00:12:02.870 --> 00:12:05.509 drowned out or replaced okay but if ai can generate 00:12:05.509 --> 00:12:08.379 entire indistinguishable videos or articles. 00:12:08.700 --> 00:12:11.519 How does a new entrepreneur even start? How do 00:12:11.519 --> 00:12:13.500 you build a defensible business model if anything 00:12:13.500 --> 00:12:15.740 you create can potentially be copied instantly 00:12:15.740 --> 00:12:18.840 by AI? That's the crux of it. The defense isn't 00:12:18.840 --> 00:12:21.500 just the content itself anymore. It's about focusing 00:12:21.500 --> 00:12:23.860 intensely on a specific niche where you have 00:12:23.860 --> 00:12:27.320 real expertise and, crucially, building a human 00:12:27.320 --> 00:12:30.159 -centric community around that expertise. Building 00:12:30.159 --> 00:12:32.539 trust. That's much harder for AI to replicate 00:12:32.539 --> 00:12:36.039 than just raw information output. mid -roll sponsor 00:12:36.039 --> 00:12:39.299 break. All right, let's shift gears now to the 00:12:39.299 --> 00:12:42.419 strategic response. We've painted a picture of 00:12:42.419 --> 00:12:45.799 a very powerful, fast -moving platform. Sounds 00:12:45.799 --> 00:12:47.940 intimidating, but maybe its rise doesn't have 00:12:47.940 --> 00:12:50.940 to be a death sentence for everyone else. Right. 00:12:51.039 --> 00:12:53.720 The survival guide, based on the sources, seems 00:12:53.720 --> 00:12:56.139 pretty clear. Don't try to compete head -on with 00:12:56.139 --> 00:12:58.159 OpenAI's core infrastructure. That's likely a 00:12:58.159 --> 00:13:00.639 losing battle. Instead, use the platform. Build 00:13:00.639 --> 00:13:02.850 on it. Okay, so for entrepreneurs, what does 00:13:02.850 --> 00:13:04.629 that look like practically? It means seriously 00:13:04.629 --> 00:13:07.250 considering building ChatGPT native apps using 00:13:07.250 --> 00:13:10.090 that app's SDK. Go where the 800 million users 00:13:10.090 --> 00:13:13.169 are. Get discovered there. And critically, use 00:13:13.169 --> 00:13:16.570 AgentKit for rapid prototyping. Build your minimum 00:13:16.570 --> 00:13:19.789 viable product in days, not months. Test ideas 00:13:19.789 --> 00:13:23.690 incredibly fast. But, and this is key, maintain 00:13:23.690 --> 00:13:27.330 extreme agility. Be ready to pivot quickly because 00:13:27.330 --> 00:13:30.049 OpenAI will likely move into successful niches 00:13:30.049 --> 00:13:33.490 eventually. assume that. Adaptability is paramount. 00:13:33.509 --> 00:13:35.889 Yeah. And for developers specifically, what's 00:13:35.889 --> 00:13:38.870 the advice? It gets quite technical. master agent 00:13:38.870 --> 00:13:41.389 kits visual builder that way of thinking visually 00:13:41.389 --> 00:13:44.769 about workflows is becoming crucial really study 00:13:44.769 --> 00:13:47.389 that 55 000 character widget prompt we talked 00:13:47.389 --> 00:13:49.370 about not just the code but the architecture 00:13:49.370 --> 00:13:51.570 of the prompt the error handling the control 00:13:51.570 --> 00:13:54.129 mechanisms it's an education in itself right 00:13:54.129 --> 00:13:56.970 and integrate mcps those model context protocols 00:13:56.970 --> 00:14:00.149 for broad connectivity building isolated agents 00:14:00.149 --> 00:14:02.389 isn't enough they need access to the wider digital 00:14:02.389 --> 00:14:05.090 ecosystem and mcps are the standardized way to 00:14:05.090 --> 00:14:07.529 do that now use code export features to build 00:14:07.559 --> 00:14:09.960 more custom things when needed. And open AI strategy 00:14:09.960 --> 00:14:12.039 around open source plays into this too, right? 00:14:12.080 --> 00:14:14.480 This hybrid approach. Absolutely. It's quite 00:14:14.480 --> 00:14:17.080 clever, really. They open source the things that 00:14:17.080 --> 00:14:19.779 encourage widespread adoption and lock in the 00:14:19.779 --> 00:14:23.120 chat kit, front end components, React, the codex, 00:14:23.120 --> 00:14:26.860 SDK, even those huge complex prompts get everyone 00:14:26.860 --> 00:14:29.559 using their tools and methods. But they keep 00:14:29.559 --> 00:14:33.320 the core strategic assets proprietary. The underlying 00:14:33.320 --> 00:14:36.659 cloud platform for agent builder, the most advanced 00:14:36.659 --> 00:14:38.960 versions of their models, the core infrastructure. 00:14:39.340 --> 00:14:41.600 It ensures everyone builds on their platform 00:14:41.600 --> 00:14:43.960 while they maintain control and the ultimate 00:14:43.960 --> 00:14:46.559 technical edge. Which brings us back to the monopoly 00:14:46.559 --> 00:14:49.220 question. We need to look at this whole shift 00:14:49.220 --> 00:14:51.960 with balance. There are good aspects, surely. 00:14:52.360 --> 00:14:55.179 Undeniably. The good part is that it's democratizing 00:14:55.179 --> 00:14:57.820 access to incredibly powerful AI agent development 00:14:57.820 --> 00:15:00.980 tools. It enables innovation at a speed and scale 00:15:00.980 --> 00:15:03.820 we haven't seen before. Small teams, even individuals, 00:15:03.980 --> 00:15:05.779 can now build things that were impossible just 00:15:05.779 --> 00:15:08.200 a year or two ago. Everyone benefits from these 00:15:08.200 --> 00:15:10.340 tools becoming more accessible, often free to 00:15:10.340 --> 00:15:12.799 start. But the flip side. The flip side is significant. 00:15:13.139 --> 00:15:15.600 It concentrates an immense amount of power and 00:15:15.600 --> 00:15:18.399 influence into one single company. It creates 00:15:18.399 --> 00:15:21.259 a huge dependency for potentially millions of 00:15:21.259 --> 00:15:24.000 developers and businesses worldwide. And it raises 00:15:24.000 --> 00:15:26.700 really serious long -term questions about data 00:15:26.700 --> 00:15:29.460 privacy, algorithmic bias, and whether this level 00:15:29.460 --> 00:15:32.460 of centralization ultimately stifles competition 00:15:32.460 --> 00:15:35.059 and innovation in the broader tech landscape. 00:15:35.340 --> 00:15:38.159 So given that concentration of power, what's 00:15:38.159 --> 00:15:40.360 the most resilient strategy for businesses or 00:15:40.360 --> 00:15:43.139 developers looking long term? It seems to be 00:15:43.139 --> 00:15:46.220 heading towards what the sources call a multiplatform 00:15:46.220 --> 00:15:49.409 future. Or maybe a hybrid approach. Meaning? 00:15:49.710 --> 00:15:52.409 Use ChatGPT and its ecosystem strategically, 00:15:52.809 --> 00:15:55.549 especially for discovery, user acquisition, maybe 00:15:55.549 --> 00:15:58.289 for simpler tasks. Leverage their reach. Well, 00:15:58.309 --> 00:16:00.169 don't put all your eggs in that basket. Exactly. 00:16:00.509 --> 00:16:02.610 Maintain your own native applications for your 00:16:02.610 --> 00:16:04.730 core, high -value, specialized functionality. 00:16:05.210 --> 00:16:07.809 The stuff where you need complete control, deep 00:16:07.809 --> 00:16:10.429 integration, or unique user experiences. Wow. 00:16:10.490 --> 00:16:12.909 And you can power these native apps using AgentKit 00:16:12.909 --> 00:16:15.450 on the back end for the AI logic and then embed 00:16:15.450 --> 00:16:18.250 custom ChatKit. components for the UI UX within 00:16:18.250 --> 00:16:20.809 your own properties. It's about building a hedge 00:16:20.809 --> 00:16:24.289 using open eyes power without becoming completely 00:16:24.289 --> 00:16:26.710 dependent on them. OK, let's try to wrap this 00:16:26.710 --> 00:16:29.289 up. What's the big idea of the core takeaway 00:16:29.289 --> 00:16:32.009 from this deep dove? I think the core takeaway 00:16:32.009 --> 00:16:34.990 is simply this open eye hasn't just launched 00:16:34.990 --> 00:16:38.389 some new products. They have fundamentally redefined 00:16:38.389 --> 00:16:40.710 the landscape for software development, for content 00:16:40.710 --> 00:16:43.269 creation, potentially even for knowledge work 00:16:43.269 --> 00:16:46.299 itself. beat. And the choice facing basically 00:16:46.299 --> 00:16:48.700 everyone, developers, entrepreneurs, creators, 00:16:48.940 --> 00:16:50.919 maybe even employees, is becoming increasingly 00:16:50.919 --> 00:16:54.299 binary and immediate. You either adapt to this 00:16:54.299 --> 00:16:57.299 new reality and learn to use these tools or you 00:16:57.299 --> 00:16:59.840 risk getting replaced or left behind. Engagement 00:16:59.840 --> 00:17:01.860 isn't optional anymore. So for you listening, 00:17:01.940 --> 00:17:04.000 what's the immediate action plan if this feels 00:17:04.000 --> 00:17:05.980 urgent? What should be the focus for the next, 00:17:06.059 --> 00:17:09.140 say, 30 days? The sources suggest three clear 00:17:09.140 --> 00:17:11.160 things. One, if you haven't already, create an 00:17:11.160 --> 00:17:13.160 OpenAI platform account. Get familiar with the 00:17:13.160 --> 00:17:15.450 interface. face the options just explore today 00:17:15.450 --> 00:17:18.930 right to build your first simple agent kit workflow 00:17:18.930 --> 00:17:21.529 doesn't have to be complex just test it out see 00:17:21.529 --> 00:17:23.230 how the visual builder works connect it to maybe 00:17:23.230 --> 00:17:27.390 one simple API using MCP feel the speed get hands 00:17:27.390 --> 00:17:31.299 -on exactly and three Spend some real time digging 00:17:31.299 --> 00:17:34.400 into that 55 ,000 character chat kit widget prompt. 00:17:34.920 --> 00:17:37.279 Don't just glance at it. Try to understand its 00:17:37.279 --> 00:17:39.019 structure, the techniques they're using. It's 00:17:39.019 --> 00:17:41.880 like a free advanced course in practical industrial 00:17:41.880 --> 00:17:44.900 scale AI engineering. A challenging but valuable 00:17:44.900 --> 00:17:47.019 assignment. For sure. Look, we're witnessing 00:17:47.019 --> 00:17:49.789 a transformation that feels... kind of inevitable 00:17:49.789 --> 00:17:52.390 at this point. The real question isn't if open 00:17:52.390 --> 00:17:55.230 AI will consolidate power over significant parts 00:17:55.230 --> 00:17:57.190 of the internet. It feels like that's already 00:17:57.190 --> 00:17:59.190 happening. The question is whether you are going 00:17:59.190 --> 00:18:01.690 to be actively participating, building on this 00:18:01.690 --> 00:18:04.049 powerful new ecosystem, learning the new skills, 00:18:04.190 --> 00:18:06.910 or finding yourself watching from the sidelines 00:18:06.910 --> 00:18:09.779 as the advantage shifts decisively. So maybe 00:18:09.779 --> 00:18:11.460 the final thought to leave you with is this. 00:18:11.779 --> 00:18:14.700 Take a moment after this. Really consider where 00:18:14.700 --> 00:18:16.640 your unique skills, your specific knowledge, 00:18:16.700 --> 00:18:19.099 your human insight fits into this rapidly evolving 00:18:19.099 --> 00:18:22.319 AI native world. Because that irreplaceable perspective, 00:18:22.519 --> 00:18:24.720 that might just be your most valuable asset moving 00:18:24.720 --> 00:18:26.299 forward. Out to your own music.