WEBVTT 00:00:00.000 --> 00:00:02.120 You know, we've all been there. Staring at that 00:00:02.120 --> 00:00:05.299 blink screen. The blinking cursor. Exactly. And 00:00:05.299 --> 00:00:08.199 you're trying to map out this perfect, complex 00:00:08.199 --> 00:00:10.939 prompt in your head. You have to plan the execution, 00:00:11.199 --> 00:00:13.679 the format, all of it before you even type a 00:00:13.679 --> 00:00:16.399 word. It's incredibly stressful. It is. And it's 00:00:16.399 --> 00:00:19.320 so inefficient. basically forces you to become 00:00:19.320 --> 00:00:21.420 the operating system. Well, the whole interface 00:00:21.420 --> 00:00:23.239 is moving away from that. It's shifting that 00:00:23.239 --> 00:00:25.300 entire intellectual burden. We're going from 00:00:25.300 --> 00:00:28.500 asking you to be a prompt engineer to just letting 00:00:28.500 --> 00:00:31.039 you delegate work. To a specialized AI teammate. 00:00:31.320 --> 00:00:36.299 Yeah. The blank text box, as we know it, is starting 00:00:36.299 --> 00:00:39.759 to die. Welcome to the Deep Dive. Our mission 00:00:39.759 --> 00:00:42.710 today is to give you a shortcut. really to understanding 00:00:42.710 --> 00:00:46.710 this next wave of ai workflow our main source 00:00:46.710 --> 00:00:50.350 is a pretty massive recent leak about claude's 00:00:50.350 --> 00:00:53.950 proposed ager mode and it signals a really fundamental 00:00:53.950 --> 00:00:55.829 shift in how we're going to interact with these 00:00:55.829 --> 00:00:58.109 systems and our mission for you is quick and 00:00:58.109 --> 00:01:00.710 thorough so first we're going to unpack these 00:01:00.710 --> 00:01:03.549 new delegation structures and and some revolutionary 00:01:03.549 --> 00:01:05.920 tracking features that are coming Then second, 00:01:06.040 --> 00:01:09.219 we'll dive into a really powerful like five minute 00:01:09.219 --> 00:01:12.760 framework for validating trends, high growth, 00:01:12.920 --> 00:01:15.480 low competition trends using a real world example. 00:01:15.700 --> 00:01:17.599 High rocks. The high rocks fitness competition. 00:01:17.620 --> 00:01:19.780 And then finally, we'll look at two seriously 00:01:19.780 --> 00:01:23.599 underrated tools, CREA and Notebook LM, and offer 00:01:23.599 --> 00:01:26.700 a clear tactical strategy for just moving past 00:01:26.700 --> 00:01:28.739 that initial vagueness and building something 00:01:28.739 --> 00:01:32.120 valuable. OK, so let's unpack this shift. The 00:01:32.120 --> 00:01:34.140 big problem, as we said, is the blank text box. 00:01:34.280 --> 00:01:37.299 Right. It forces you to plan and structure and 00:01:37.299 --> 00:01:40.560 explain this whole complex task up front. It 00:01:40.560 --> 00:01:43.079 leads to what the source calls prompt paralysis, 00:01:43.319 --> 00:01:45.760 which I think is a perfect term. Totally. The 00:01:45.760 --> 00:01:47.739 bottleneck was never the model's intelligence. 00:01:47.920 --> 00:01:51.099 It was just the sheer amount of work the interface 00:01:51.099 --> 00:01:53.019 made you do just to get started. So this delegation 00:01:53.019 --> 00:01:55.239 model just flips that on its head. It changes 00:01:55.239 --> 00:01:57.719 the relationship completely. The leak suggests 00:01:57.719 --> 00:02:00.659 AI is moving from being this reactive answer 00:02:00.659 --> 00:02:03.480 machine to a dedicated delegated worker. You 00:02:03.480 --> 00:02:05.400 assign the work type first, then you give it 00:02:05.400 --> 00:02:07.359 the input. So you're defining the shape of the 00:02:07.359 --> 00:02:09.659 output before you even start typing the details. 00:02:09.860 --> 00:02:13.740 You choose research or analyze or write. Exactly. 00:02:14.300 --> 00:02:16.379 you're leveraging their internal strengths. So 00:02:16.379 --> 00:02:18.840 when you click research, you're not just asking 00:02:18.840 --> 00:02:21.259 a search query, you're handing off a predefined 00:02:21.259 --> 00:02:25.360 multi -step process for compiling sources and 00:02:25.360 --> 00:02:28.580 deep factual validation. And the UI is splitting. 00:02:28.680 --> 00:02:30.860 It's going from that one conversation box to 00:02:30.860 --> 00:02:34.560 five core task categories. So there's research 00:02:34.560 --> 00:02:36.620 for the deep dives. And then if you need to analyze, 00:02:36.879 --> 00:02:39.520 you click that button, and the agent knows immediately, 00:02:39.740 --> 00:02:42.099 okay, I need to compare data, look for patterns, 00:02:42.240 --> 00:02:44.319 maybe run a forecast. And you define the output 00:02:44.319 --> 00:02:46.780 structure, like a table or a graph, before you 00:02:46.780 --> 00:02:49.080 even paste in the data. Yeah. Then there's write 00:02:49.080 --> 00:02:51.330 mode. which is way more than just generating 00:02:51.330 --> 00:02:54.330 text. It supports specific formats. Docs, slides, 00:02:54.629 --> 00:02:56.789 a spreadsheet. And it has automatic citation 00:02:56.789 --> 00:02:59.289 support built right in, tying directly back to 00:02:59.289 --> 00:03:01.229 the sources it just researched for you. And the 00:03:01.229 --> 00:03:04.930 last two are build for lightweight app development. 00:03:05.210 --> 00:03:07.719 And do more. for other autonomous things but 00:03:07.719 --> 00:03:09.500 the point is you don't write a thousand word 00:03:09.500 --> 00:03:11.599 mega prompt anymore you just choose the workflow 00:03:11.599 --> 00:03:14.039 but the thing that really separates this the 00:03:14.039 --> 00:03:17.120 mechanism that shifts user confidence is the 00:03:17.120 --> 00:03:20.949 new visibility you get two things first the progress 00:03:20.949 --> 00:03:24.490 tracker, a sidebar that literally shows the agent's 00:03:24.490 --> 00:03:27.490 current tasks in real time. You watch the AI 00:03:27.490 --> 00:03:30.330 work. And that, combined with the context manager, 00:03:30.530 --> 00:03:33.050 which is this list of all the documents and sources 00:03:33.050 --> 00:03:36.090 the agent is actively using, that is crucial. 00:03:36.310 --> 00:03:38.009 It gets rid of that nagging feeling, right? You're 00:03:38.009 --> 00:03:40.090 not wondering what the AI is thinking or if it's 00:03:40.090 --> 00:03:42.590 just making stuff up. Exactly. So if the AI is 00:03:42.590 --> 00:03:45.550 autonomous, how does this new mode increase the 00:03:45.550 --> 00:03:48.460 user's confidence in the output? Well, seeing 00:03:48.460 --> 00:03:50.539 that real -time progress tracker in the context 00:03:50.539 --> 00:03:53.039 window, it just gives you visibility into the 00:03:53.039 --> 00:03:55.139 agent's actual workflow. So it's building trust. 00:03:55.400 --> 00:03:57.319 It builds trust and saves you the effort of being 00:03:57.319 --> 00:04:00.360 a prompt engineer. That's a huge structural shift. 00:04:00.719 --> 00:04:03.520 And, I mean, if this delegation mode saves us 00:04:03.520 --> 00:04:06.620 all this intellectual time on execution, where 00:04:06.620 --> 00:04:08.949 should we spend that time? On spotting better 00:04:08.949 --> 00:04:11.270 opportunities. Right. In the market. Exactly. 00:04:11.469 --> 00:04:14.069 We need to pivot to validation. Right. How do 00:04:14.069 --> 00:04:16.649 you spot a real opportunity with momentum, not 00:04:16.649 --> 00:04:20.329 just noise? You don't need months of research. 00:04:20.569 --> 00:04:22.509 You just need to realize that a trend usually 00:04:22.509 --> 00:04:25.550 whispers before it shouts. That's it. So the 00:04:25.550 --> 00:04:28.329 process starts with what the source calls social 00:04:28.329 --> 00:04:30.829 signal detection. You're looking for repetition, 00:04:31.089 --> 00:04:34.569 not volume. Repetition in, like, short videos, 00:04:34.769 --> 00:04:37.430 comments. Yeah, casual mentions. Where are people 00:04:37.430 --> 00:04:40.029 venting or sharing their passion over and over 00:04:40.029 --> 00:04:42.529 again without being prompted? The source points 00:04:42.529 --> 00:04:44.610 to the fitness competition, High Rocks. Right. 00:04:44.829 --> 00:04:47.209 It went from these niche gym conversations to 00:04:47.209 --> 00:04:49.589 being everywhere in short video content. That 00:04:49.589 --> 00:04:51.990 repetition, that's your first clue. Culture moves 00:04:51.990 --> 00:04:54.089 before the hard data does. And once you hear 00:04:54.089 --> 00:04:56.589 that whisper, you move to the quick data check. 00:04:56.970 --> 00:05:00.129 Yep. Use a tool like Perplexity to verify the 00:05:00.129 --> 00:05:02.980 interest fast. And the stats for HIROX workout 00:05:02.980 --> 00:05:05.459 are pretty compelling. It's about 3 .9 million 00:05:05.459 --> 00:05:09.120 monthly searches. Which is up 266 % year over 00:05:09.120 --> 00:05:12.779 year. That's huge. It is, but the engagement 00:05:12.779 --> 00:05:15.699 metrics are even more startling. ClassPass saw 00:05:15.699 --> 00:05:20.279 a 506 % increase in HIROX class searches globally 00:05:20.279 --> 00:05:24.089 in the last year. That velocity is just undeniable. 00:05:24.310 --> 00:05:26.910 It really is. So what's the biggest mistake people 00:05:26.910 --> 00:05:29.290 make when they spot a growing trend like that? 00:05:29.490 --> 00:05:31.850 Oh, they try to build something massive and vague. 00:05:32.149 --> 00:05:34.310 Instead of a narrow solution. Instead of a narrow 00:05:34.310 --> 00:05:37.269 solution for a specific group's immediate painful 00:05:37.269 --> 00:05:39.470 struggle. So you see this golden opportunity. 00:05:40.269 --> 00:05:43.569 Massive growth, low competition, probably a cheap 00:05:43.569 --> 00:05:46.490 cost per click. A rare combo. It's a signal to 00:05:46.490 --> 00:05:49.420 act, but the lesson is to build narrow. Don't 00:05:49.420 --> 00:05:51.220 try to build the next social network for high 00:05:51.220 --> 00:05:53.819 rocks. No. You ask, what is the one specific 00:05:53.819 --> 00:05:56.480 weekly struggle this group has? Do they need 00:05:56.480 --> 00:05:58.839 a training tracker? A specific gear guide, like 00:05:58.839 --> 00:06:01.019 best shoes for high rocks. Or an AI check -in 00:06:01.019 --> 00:06:02.740 buddy. Something like that. You just respond 00:06:02.740 --> 00:06:05.699 to the existing obvious demand. That focus on 00:06:05.699 --> 00:06:07.860 solving these narrow pain points and reducing 00:06:07.860 --> 00:06:10.439 friction, it leads us right to the creation tools 00:06:10.439 --> 00:06:13.180 themselves. If the agent is getting rid of prompt 00:06:13.180 --> 00:06:15.660 paralysis, what tools get rid of setup paralysis? 00:06:16.160 --> 00:06:18.139 That's the real -time sync now, isn't it? Yeah. 00:06:18.180 --> 00:06:20.500 It's not bad prompts. It's tool hopping. It's 00:06:20.500 --> 00:06:23.480 exporting and relearning UIs. Yeah. So tool number 00:06:23.480 --> 00:06:27.449 one is CREA. That's K -R -E -A dot A -I. You 00:06:27.449 --> 00:06:29.689 can think of it as like the creative AI Swiss 00:06:29.689 --> 00:06:32.629 army knife. Okay. It solves the integration problem. 00:06:32.790 --> 00:06:35.569 It bundles a bunch of strong models, Flux and 00:06:35.569 --> 00:06:39.350 Gemini for images, Runway and Luma for video 00:06:39.350 --> 00:06:42.209 all into one workspace. You can stop juggling 00:06:42.209 --> 00:06:44.689 five different subscriptions. And the game isn't 00:06:44.689 --> 00:06:47.410 just saving money on subs. It's about workflow 00:06:47.410 --> 00:06:49.730 velocity. Right. If you need 10 variations of 00:06:49.730 --> 00:06:51.709 a social media ad with a logo, a little video 00:06:51.709 --> 00:06:54.750 clip with CREA, it's one integrated process. 00:06:54.990 --> 00:06:57.389 The friction is just gone. And tool number two? 00:06:57.649 --> 00:07:01.009 Is Google Notebook LM. A lot of people are using 00:07:01.009 --> 00:07:03.730 it for, you know, basic summaries, but they are 00:07:03.730 --> 00:07:06.290 totally missing its most powerful feature, the 00:07:06.290 --> 00:07:08.910 slide generator. How does it work? Like, give 00:07:08.910 --> 00:07:11.149 me a high leverage scenario. Okay. You upload 00:07:11.149 --> 00:07:13.550 a 50 -page quarterly report, the CEO's unedited 00:07:13.550 --> 00:07:15.350 transcript from an earnings call, maybe a few 00:07:15.350 --> 00:07:18.250 competitive Intel PDFs. Notebook LM doesn't just 00:07:18.250 --> 00:07:21.069 summarize them one by one. It identifies the 00:07:21.069 --> 00:07:23.730 core arguments, structures them into a cohesive 00:07:23.730 --> 00:07:27.250 story intro problem solution, and generates a 00:07:27.250 --> 00:07:30.350 ready -to -pitch slide deck. Whoa. Okay, imagine 00:07:30.350 --> 00:07:32.970 scaling that. Going from a long transcript to 00:07:32.970 --> 00:07:35.949 a fully structured presentation in, what, minutes? 00:07:36.050 --> 00:07:38.589 It's an insane leverage point. That's incredible. 00:07:38.889 --> 00:07:41.829 It completely reverses the normal workflow. Most 00:07:41.829 --> 00:07:44.009 tools want you to write everything, then design. 00:07:44.740 --> 00:07:47.920 Notebook LM takes your content, structures it, 00:07:48.040 --> 00:07:49.920 and then presents it back to you for polishing. 00:07:50.120 --> 00:07:52.519 So beyond just getting faster output, what mindset 00:07:52.519 --> 00:07:55.699 does using these bundled tools encourage? It 00:07:55.699 --> 00:07:58.040 really encourages using tools for speed first, 00:07:58.180 --> 00:08:00.839 then polishing somewhere else. It just cuts down 00:08:00.839 --> 00:08:02.959 drastically on the time you spend on that initial 00:08:02.959 --> 00:08:05.199 setup and planning, which brings us from tools 00:08:05.199 --> 00:08:07.279 to high -level strategy. And let's start with 00:08:07.279 --> 00:08:09.300 a criminally simple startup idea from the source 00:08:09.300 --> 00:08:12.259 material, the Digital Hotel Concierge. Okay, 00:08:12.300 --> 00:08:14.269 what's the leverage here? And how does it connect 00:08:14.269 --> 00:08:16.870 back to delegation? Well, the problem is repetitive 00:08:16.870 --> 00:08:20.750 human labor. Hotel staff waste hours every single 00:08:20.750 --> 00:08:23.149 day repeating the answers to the same simple 00:08:23.149 --> 00:08:25.990 questions. Wi -Fi password, breakfast hours, 00:08:26.329 --> 00:08:29.389 checkout time. All of it. The solution is a simple 00:08:29.389 --> 00:08:32.409 QR code in the room that leads to a no -app digital 00:08:32.409 --> 00:08:35.509 guest guide. It handles the FAQs, has short walkthrough 00:08:35.509 --> 00:08:38.529 videos, local tips, all instantly. And the value 00:08:38.529 --> 00:08:41.559 prop is simple. Staff time saved is money saved 00:08:41.559 --> 00:08:44.240 and fewer basic complaints mean higher satisfaction 00:08:44.240 --> 00:08:46.740 scores. And the business model. It's designed 00:08:46.740 --> 00:08:50.899 to win. Low friction sauce, maybe $29 to $79 00:08:50.899 --> 00:08:53.460 a month per property, plus affiliate revenue 00:08:53.460 --> 00:08:55.679 from booking local tours and restaurants. It 00:08:55.679 --> 00:08:58.000 works because it solves a specific repetitive 00:08:58.000 --> 00:09:01.480 pain point. In a low cost, clear way. And that 00:09:01.480 --> 00:09:04.139 specificity is so important. You know, even with 00:09:04.139 --> 00:09:05.980 great tools like these, I still wrestle with 00:09:05.980 --> 00:09:08.490 prompt drift myself. Oh, yeah. especially when 00:09:08.490 --> 00:09:11.049 I haven't clearly defined the target person I'm 00:09:11.049 --> 00:09:14.190 building for. It's so hard to start if your audience 00:09:14.190 --> 00:09:16.750 is vague. This is exactly why the next strategic 00:09:16.750 --> 00:09:19.370 framework, the 1 ,000 people framework, is so 00:09:19.370 --> 00:09:21.769 powerful. Your goal is not millions of users. 00:09:21.929 --> 00:09:24.909 It's finding 1 ,000 very specific people with 00:09:24.909 --> 00:09:27.629 the exact same painful problem. So step one is 00:09:27.629 --> 00:09:30.330 just radical specificity. You move from a vague 00:09:30.330 --> 00:09:33.090 term like founders to something hyper -specific 00:09:33.090 --> 00:09:36.399 like... Italian restaurant owners in mid -sized 00:09:36.399 --> 00:09:38.860 cities who lose money from online booking no 00:09:38.860 --> 00:09:41.320 -shows. Now you can picture their day, their 00:09:41.320 --> 00:09:43.759 stress points, their budget. Exactly. And once 00:09:43.759 --> 00:09:45.759 you have them, you research. You can simulate 00:09:45.759 --> 00:09:48.679 interviewing them, even use AI role -play agents 00:09:48.679 --> 00:09:50.960 to really understand their pain. Then you ask 00:09:50.960 --> 00:09:53.299 the big question, the practical one. What would 00:09:53.299 --> 00:09:57.059 all 1 ,000 of these people happily pay $50 to 00:09:57.059 --> 00:09:59.940 $100 annually for without even thinking about 00:09:59.940 --> 00:10:02.330 it? That price point is so crucial. Why that 00:10:02.330 --> 00:10:05.169 number specifically? Because that price bypasses 00:10:05.169 --> 00:10:08.330 99 % of internal procurement hurdles for a small 00:10:08.330 --> 00:10:11.070 business. A manager can just expense it. It's 00:10:11.070 --> 00:10:13.429 low enough for that, but high enough to feel 00:10:13.429 --> 00:10:15.940 valuable. And if you can't answer that $100 question 00:10:15.940 --> 00:10:18.399 in one sentence, you aren't ready to build. You're 00:10:18.399 --> 00:10:21.799 not. And then finally, step four is reach strategy 00:10:21.799 --> 00:10:24.279 clarity. You have to name the exact distribution 00:10:24.279 --> 00:10:27.440 channel, not just social media ads. It's all 00:10:27.440 --> 00:10:30.100 email the 5 ,000 restaurant owners on OpenTable 00:10:30.100 --> 00:10:32.240 in this region. We're all post daily in these 00:10:32.240 --> 00:10:35.039 two specific Facebook groups. Distribution clarity 00:10:35.039 --> 00:10:38.019 beats product polish every single time. So how 00:10:38.019 --> 00:10:41.480 does defining a specific $100 annual offer help 00:10:41.480 --> 00:10:43.899 the building process? It just forces you to focus 00:10:43.899 --> 00:10:47.820 on solving a core recognized pain point instead 00:10:47.820 --> 00:10:50.919 of building a collection of cool but ultimately 00:10:50.919 --> 00:10:53.379 vague features. So let's bring this all together. 00:10:53.500 --> 00:10:55.179 We've covered a lot, but there are two massive 00:10:55.179 --> 00:10:58.080 takeaways for you today. First, the AI workflow 00:10:58.080 --> 00:11:00.279 is changing. It's moving from conversational 00:11:00.279 --> 00:11:04.080 input to structured delegation. Using specialized 00:11:04.080 --> 00:11:07.509 agents with real -time tracking. The era of prompt 00:11:07.509 --> 00:11:10.649 engineering is ending. And second, success in 00:11:10.649 --> 00:11:12.870 this new landscape is all about finding extreme 00:11:12.870 --> 00:11:16.000 clarity in who you serve. using frameworks like 00:11:16.000 --> 00:11:18.320 the 1 ,000 people framework, and then validating 00:11:18.320 --> 00:11:20.799 their pain with quick data checks, just like 00:11:20.799 --> 00:11:22.879 we saw with the Hyrox trend. The cloud leak is 00:11:22.879 --> 00:11:25.440 really just a preview of what's going to be table 00:11:25.440 --> 00:11:28.519 stakes across all platforms very soon. The real 00:11:28.519 --> 00:11:30.519 competition won't be about who has access to 00:11:30.519 --> 00:11:32.460 the best models. It's going to be about who moves 00:11:32.460 --> 00:11:35.039 the fastest and uses these delegation tools with 00:11:35.039 --> 00:11:38.080 the most clarity. Absolutely. And that opportunity 00:11:38.080 --> 00:11:41.220 window to leverage this agent shift, it's shrinking. 00:11:41.710 --> 00:11:44.149 So if you understand these principles of structured 00:11:44.149 --> 00:11:47.049 delegation now, what complex workflow could you 00:11:47.049 --> 00:11:48.950 automate for yourself by this time next year? 00:11:49.549 --> 00:11:51.769 That's the strategic question to think about. 00:11:51.929 --> 00:11:54.029 We really encourage you to test this agent mode 00:11:54.029 --> 00:11:56.769 mindset. Try running that 1 ,000 people exercise 00:11:56.769 --> 00:11:59.009 on a problem you're facing. And try generating 00:11:59.009 --> 00:12:02.230 a quick slide deck and notebook LM. Momentum 00:12:02.230 --> 00:12:04.370 comes from action. We'll catch you on the next 00:12:04.370 --> 00:12:05.110 Deep Dive.