WEBVTT 00:00:00.000 --> 00:00:02.020 You know, I was looking at the labor market numbers 00:00:02.020 --> 00:00:05.459 this morning. It is February 2nd, 2026, by the 00:00:05.459 --> 00:00:08.779 way. And there's a $10 trillion figure staring 00:00:08.779 --> 00:00:11.279 back at me. Wow. That is the valuation of the 00:00:11.279 --> 00:00:14.339 AI labor market. It feels massive. Kind of like 00:00:14.339 --> 00:00:16.420 the entire world's been rewritten by code. It 00:00:16.420 --> 00:00:18.839 definitely feels that way. But then you look 00:00:18.839 --> 00:00:21.679 at the hard truth buried in those reports from 00:00:21.679 --> 00:00:25.359 IBM and MIT. And that truth is just staggering. 00:00:26.140 --> 00:00:30.320 Up to 95%. of AI implementations right now show 00:00:30.320 --> 00:00:34.700 zero measurable return on investment. Zero. Zero. 00:00:34.960 --> 00:00:37.399 That is a massive disconnect. You've got $10 00:00:37.399 --> 00:00:40.299 trillion of hype on one side, and on the other, 00:00:40.600 --> 00:00:42.299 business is just burning Pash with nothing to 00:00:42.299 --> 00:00:45.039 show for it. So why? Why is there such a gap? 00:00:45.159 --> 00:00:48.340 Well, the answer is actually pretty counterintuitive. 00:00:48.939 --> 00:00:50.960 It turns out the secret isn't being exciting 00:00:50.960 --> 00:00:54.600 or revolutionary. It's about being boring. Welcome 00:00:54.600 --> 00:00:56.020 to the Deep Dive. I'm really glad you're here 00:00:56.020 --> 00:00:57.759 with us. Today we are stripping away the hype. 00:00:57.939 --> 00:00:59.740 We are not talking about how to build the next 00:00:59.740 --> 00:01:02.700 chat GPT or, you know, the next sentient robot. 00:01:02.920 --> 00:01:05.400 We are looking at the current landscape of 2026 00:01:05.400 --> 00:01:08.040 to understand why most AI businesses crash and 00:01:08.040 --> 00:01:10.260 burn and how to build one that actually works. 00:01:10.379 --> 00:01:13.079 We are digging into a guide titled The Profitable 00:01:13.079 --> 00:01:16.480 Pivot. And the core argument here is just fascinating. 00:01:16.730 --> 00:01:18.989 It really is. We're going to unpack some concepts 00:01:18.989 --> 00:01:21.730 that might honestly change how you see your own 00:01:21.730 --> 00:01:24.090 work. We'll talk about the wrapper trap, why 00:01:24.090 --> 00:01:26.430 treating AI like a smart intern is the only way 00:01:26.430 --> 00:01:31.810 to survive, and why doing manual unsexy service 00:01:31.810 --> 00:01:34.709 work is actually the shortcut to building a massive 00:01:34.709 --> 00:01:36.969 software company. I love that. OK, let's start 00:01:36.969 --> 00:01:40.530 with that 95 % failure rate. That is a terrifying 00:01:40.530 --> 00:01:43.370 number if you're a founder or an investor. What 00:01:43.370 --> 00:01:45.549 is going wrong? It usually starts with what we 00:01:45.549 --> 00:01:48.120 call the wrapper trap. Even now, in 2026, you 00:01:48.120 --> 00:01:50.400 see founders who think an AI business is just, 00:01:50.400 --> 00:01:53.439 you know, slapping a user interface, a UI on 00:01:53.439 --> 00:01:56.459 top of OpenAI or Anthropix models. OK. They create 00:01:56.459 --> 00:01:58.859 a nice looking website that basically just passes 00:01:58.859 --> 00:02:01.599 text back and forth to a big brain in the cloud. 00:02:01.799 --> 00:02:03.700 Right. So you have a specialized chat bot, but 00:02:03.700 --> 00:02:05.540 under the hood, it's just the same engine everyone 00:02:05.540 --> 00:02:08.919 else has access to. Precisely. And the problem 00:02:08.919 --> 00:02:12.159 is that doesn't solve a deep business problem. 00:02:12.520 --> 00:02:15.340 It's a commodity. Think of it like this. Imagine 00:02:15.340 --> 00:02:17.659 opening a restaurant that claims to have a secret 00:02:17.659 --> 00:02:20.800 family recipe. Okay. But when you go into the 00:02:20.800 --> 00:02:23.580 kitchen, you realize they're just microwaving 00:02:23.580 --> 00:02:25.979 frozen meals from the grocery store. That's a 00:02:25.979 --> 00:02:28.939 great analogy. It tastes fine. But anyone can 00:02:28.939 --> 00:02:31.159 buy the frozen meal. There is no special sauce. 00:02:31.360 --> 00:02:33.780 Exactly. There's no moat. And then you have the 00:02:33.780 --> 00:02:35.780 corporate side of this failure, which is the 00:02:35.780 --> 00:02:39.020 magic button fallacy. The magic button. I'm guessing 00:02:39.020 --> 00:02:41.439 this means companies buying software expecting 00:02:41.439 --> 00:02:43.719 it to fix their problems instantly. That's the 00:02:43.719 --> 00:02:46.000 one. They buy the tool, they install it, and 00:02:46.000 --> 00:02:48.180 they just wait for the profits to roll in. But 00:02:48.180 --> 00:02:50.599 they don't change how they work. The MIT study 00:02:50.599 --> 00:02:53.259 we looked at showed that only 5 % of AI pilots 00:02:53.259 --> 00:02:56.039 actually make it to production. Only 5%. And 00:02:56.039 --> 00:02:58.669 Deloitte found that only a tiny fraction of organizations 00:02:58.669 --> 00:03:01.250 see real money coming back from these things. 00:03:01.449 --> 00:03:04.469 But, wait, hold on. You do have winners in this 00:03:04.469 --> 00:03:06.990 space. I mean, look at Klarna. They cut customer 00:03:06.990 --> 00:03:10.669 service costs by 40%. Or, Intercom, resolving 00:03:10.669 --> 00:03:13.789 millions of tickets, they're using the same underlying 00:03:13.789 --> 00:03:16.710 tech as the failures. So what's the difference? 00:03:17.009 --> 00:03:19.729 The difference is workflow. Klarna didn't just 00:03:19.729 --> 00:03:22.789 buy a tool. They fundamentally changed their 00:03:22.789 --> 00:03:25.370 internal processes. to accommodate the tool. 00:03:25.949 --> 00:03:29.009 The failures happen when you try to jam a probabilistic 00:03:29.009 --> 00:03:32.349 engine like an AI into a deterministic hole without 00:03:32.349 --> 00:03:34.710 changing the shape of the work. That is a really 00:03:34.710 --> 00:03:37.370 crucial distinction. So if the code isn't the 00:03:37.370 --> 00:03:41.110 problem, is the issue actually us, the humans? 00:03:41.189 --> 00:03:43.550 It's not a software bug. It's a workflow failure. 00:03:43.669 --> 00:03:45.729 That makes so much sense. And it brings us to 00:03:45.729 --> 00:03:48.090 the core of this deep dive, the three pillars 00:03:48.090 --> 00:03:50.389 of a model that actually works. If the standard 00:03:50.389 --> 00:03:52.969 model is broken, what replaces it? Well, the 00:03:52.969 --> 00:03:54.889 first pillar is what we just touched on, integrating 00:03:54.889 --> 00:03:58.069 into real workflows. You cannot just add AI as 00:03:58.069 --> 00:04:00.169 another layer on top of your existing stack. 00:04:00.270 --> 00:04:02.409 It creates friction. A ton of friction. You have 00:04:02.409 --> 00:04:04.789 to redesign the flow. The source material mentions 00:04:04.789 --> 00:04:07.530 a McKinsey study claiming that customizing workflows 00:04:07.530 --> 00:04:10.870 is the number one success factor. But let's get 00:04:10.870 --> 00:04:13.710 specific. I mean, customizing workflows sounds 00:04:13.710 --> 00:04:15.750 like corporate speak. What does that actually 00:04:15.750 --> 00:04:17.990 look like on the ground? Let's take a real estate 00:04:17.990 --> 00:04:21.610 agency. In the old world, a lead comes in. and 00:04:21.610 --> 00:04:24.170 a human agent has to read it, look up properties, 00:04:24.629 --> 00:04:27.569 type an email, update a spreadsheet. It's slow. 00:04:27.730 --> 00:04:30.529 Right. And boring. And prone to error if you 00:04:30.529 --> 00:04:32.209 haven't had your coffee yet. Extremely boring. 00:04:32.990 --> 00:04:36.069 Now, a successful AI integration using a tool 00:04:36.069 --> 00:04:39.449 like N8n, for example, turns that into a digital 00:04:39.449 --> 00:04:41.350 reflex. Pause there for a second. For people 00:04:41.350 --> 00:04:44.250 who might not know, what is NN? Good catch. NAN 00:04:44.250 --> 00:04:47.730 is basically a visual workflow builder. Imagine 00:04:47.730 --> 00:04:50.730 a whiteboard on your computer. You drag and drop 00:04:50.730 --> 00:04:53.529 little icons for different apps, Gmail, Slack, 00:04:53.689 --> 00:04:55.990 a database, and you just draw lines between them 00:04:55.990 --> 00:04:58.290 to connect them. OK, so it's the plumbing. Exactly. 00:04:58.449 --> 00:05:00.389 It's the digital plumbing. So back to real estate. 00:05:00.930 --> 00:05:03.009 The lid is captured. That's the trigger. The 00:05:03.009 --> 00:05:04.949 system automatically fetches properties matching 00:05:04.949 --> 00:05:08.089 the budget, say, within a 5 % variance. If a 00:05:08.089 --> 00:05:10.750 match is found, it instantly executes three actions. 00:05:11.360 --> 00:05:14.500 sends the email via Gmail, notifies the agent 00:05:14.500 --> 00:05:17.459 on Slack, and updates the database. So it's not 00:05:17.459 --> 00:05:19.899 just a chatbot answering questions. It's an engine 00:05:19.899 --> 00:05:22.000 performing actions across different platforms. 00:05:22.079 --> 00:05:24.759 It turns a manual tour into an automated reflex. 00:05:25.139 --> 00:05:28.420 But, and this is huge, that brings us to the 00:05:28.420 --> 00:05:31.319 second pillar, retraining the workforce. This 00:05:31.319 --> 00:05:34.079 is where the mindset shift has to happen. Yes. 00:05:34.259 --> 00:05:36.860 We've spent 30 years using deterministic software. 00:05:37.399 --> 00:05:40.839 If you click Save, In Microsoft Word, it saves 00:05:40.839 --> 00:05:43.259 every single time. It's binary. It works or it 00:05:43.259 --> 00:05:46.199 doesn't. But AI is probabilistic. It might give 00:05:46.199 --> 00:05:47.779 you a slightly different answer every time you 00:05:47.779 --> 00:05:50.500 ask. Which is, I imagine, terrifying for a manager 00:05:50.500 --> 00:05:53.060 who wants consistency. It is. That's why you 00:05:53.060 --> 00:05:55.360 have to train your team to be what we call rigorous 00:05:55.360 --> 00:05:57.439 editors. They have to stop trusting the screen 00:05:57.439 --> 00:05:59.519 blindly. If they don't learn to audit the output, 00:05:59.939 --> 00:06:02.079 they lose trust the moment the AI hallucinates 00:06:02.079 --> 00:06:04.589 or just makes up a fact. I have to admit, I still 00:06:04.589 --> 00:06:06.790 wrestle with this myself. I call it prompt grift. 00:06:07.189 --> 00:06:08.990 I'll get a great result three times in a row, 00:06:09.009 --> 00:06:11.189 so I stop checking. And then the fourth time, 00:06:11.449 --> 00:06:13.649 it's completely off the rails and I don't touch 00:06:13.649 --> 00:06:15.769 it until it's too late. It is so easy to get 00:06:15.769 --> 00:06:18.329 lazy. We all do it. It's human nature to conserve 00:06:18.329 --> 00:06:20.610 energy. That's why specific training prompts 00:06:20.610 --> 00:06:23.610 are so valuable. The source suggests literally 00:06:23.610 --> 00:06:26.550 giving your team a prompt that says, act as a 00:06:26.550 --> 00:06:30.120 rigorous editor. Identify hallucinations, incorrect 00:06:30.120 --> 00:06:34.180 facts, or tone mismatches. Explain why the AI 00:06:34.180 --> 00:06:35.939 might have made that mistake. You're training 00:06:35.939 --> 00:06:39.019 the human to grade the machine. Exactly. That 00:06:39.019 --> 00:06:41.319 leads right into the third pillar, but I have 00:06:41.319 --> 00:06:44.500 to ask, if the human is the grader, aren't we 00:06:44.500 --> 00:06:46.639 just trading one job for another? I thought the 00:06:46.639 --> 00:06:48.939 point was to do less work. That's the common 00:06:48.939 --> 00:06:52.439 trap. Pillar three is owning operations. Think 00:06:52.439 --> 00:06:55.500 of the smart intern analogy. AI is like a brilliant 00:06:55.500 --> 00:06:58.300 intern from a top university. They're fast, they're 00:06:58.300 --> 00:07:01.079 well read, but they're inexperienced. If you 00:07:01.079 --> 00:07:03.519 don't watch them, they will make very strange, 00:07:03.639 --> 00:07:06.139 confident mistakes. So the human role shifts 00:07:06.139 --> 00:07:10.220 from doer to babysitter. Or operator. The operator's 00:07:10.220 --> 00:07:12.839 job is updating prompts when the business logic 00:07:12.839 --> 00:07:16.279 changes and handling the edge cases, those weird, 00:07:16.459 --> 00:07:19.120 rare situations the AI hasn't seen before. Right. 00:07:19.259 --> 00:07:21.540 If you don't have an operator, Your system is 00:07:21.540 --> 00:07:23.459 like a Ferrari with a brick on the gas pedal 00:07:23.459 --> 00:07:26.079 and no one at the steering wheel. That is a vivid 00:07:26.079 --> 00:07:29.040 image. It raises a tough question, though. Does 00:07:29.040 --> 00:07:31.879 this mean the dream of set it and forget it is 00:07:31.879 --> 00:07:34.980 dead? For now, yes. You can't autopilot a car 00:07:34.980 --> 00:07:37.519 with no driver. That is a hard pill to swallow 00:07:37.519 --> 00:07:39.860 for people who wanted that easy passive income. 00:07:40.279 --> 00:07:42.759 But it seems like this necessity for human involvement 00:07:42.759 --> 00:07:45.720 is actually creating whole new business models. 00:07:45.740 --> 00:07:49.319 It is. This is where we see the rise of service 00:07:49.319 --> 00:07:52.100 -led AI. And this is probably the most important 00:07:52.100 --> 00:07:54.800 shift happening in 2026. We usually think of 00:07:54.800 --> 00:07:57.519 business as so binary. You're either a sauce 00:07:57.519 --> 00:08:00.399 company software as a service, or you're an agency 00:08:00.399 --> 00:08:02.639 selling hours. But you're saying that line is 00:08:02.639 --> 00:08:05.139 blurring. Completely blurring. In fact, the most 00:08:05.139 --> 00:08:07.240 successful companies are erasing that line. Even 00:08:07.240 --> 00:08:08.680 the biggest startups, the ones coming out of 00:08:08.680 --> 00:08:11.279 Y Combinator, are hiring roles like forward deployed 00:08:11.279 --> 00:08:13.800 engineers. Forward deployed. That sounds like 00:08:13.800 --> 00:08:16.279 military terminology. It does, doesn't it? Yeah. 00:08:16.639 --> 00:08:19.040 But it's really just fancy talk for engineers 00:08:19.040 --> 00:08:20.959 who go into the client's office and fix their 00:08:20.959 --> 00:08:23.579 data manually. Look at a company like Harvey 00:08:23.579 --> 00:08:27.300 AI. They're massive in the legal space. Now, 00:08:27.560 --> 00:08:30.100 lawyers are notoriously resistant to new tech. 00:08:30.199 --> 00:08:31.639 Right. They don't like changing how they work. 00:08:31.740 --> 00:08:34.460 And they bill by the hour. Efficiency isn't always 00:08:34.460 --> 00:08:37.820 their top goal. Sure. Exactly. So Harvey AI didn't 00:08:37.820 --> 00:08:40.179 just hand over a login and say, here's your chat 00:08:40.179 --> 00:08:42.799 bot. Good luck. They hand hold the deployment. 00:08:43.179 --> 00:08:45.299 They send people in to ensure the model actually 00:08:45.299 --> 00:08:48.519 works for that specific firm's messy data. They're 00:08:48.519 --> 00:08:50.440 doing service work to make the software stick. 00:08:50.639 --> 00:08:53.460 Because if they don't, the client churns. The 00:08:53.460 --> 00:08:55.700 software just sits there. Unused and then it's 00:08:55.700 --> 00:08:58.320 worthless spot -on. We're seeing the emergence 00:08:58.320 --> 00:09:00.840 of the full stack automation partner. It's not 00:09:00.840 --> 00:09:03.919 just building a bot It's a four -step cycle audit 00:09:03.919 --> 00:09:06.899 the process build the automation train the staff 00:09:06.899 --> 00:09:10.480 and then support it on a retainer That support 00:09:10.480 --> 00:09:13.399 piece is key. That's recurring revenue and it's 00:09:13.399 --> 00:09:16.480 sticky You aren't just a contractor you become 00:09:16.480 --> 00:09:20.299 the AI officer That is a booming role right now. 00:09:20.539 --> 00:09:23.159 Someone who sits between the CEO and the developers 00:09:23.159 --> 00:09:26.820 to translate business needs into technical prompts. 00:09:27.220 --> 00:09:29.679 Whoa! Imagine scaling that. You aren't just selling 00:09:29.679 --> 00:09:32.440 a login. You are selling a whole new way for 00:09:32.440 --> 00:09:34.840 a company to exist. You are. And that creates 00:09:34.840 --> 00:09:37.960 a dependency that pure software can't match. 00:09:38.200 --> 00:09:40.620 But wait a minute, why are investors suddenly 00:09:40.620 --> 00:09:43.240 loving services again? We spent a decade hearing 00:09:43.240 --> 00:09:45.860 services don't scale, revenue per employee is 00:09:45.860 --> 00:09:48.720 too low, tech multiples are better. Why the U 00:09:48.720 --> 00:09:51.720 -turn? Because in an AI world, services ensure 00:09:51.720 --> 00:09:54.539 the software actually gets used. Services protect 00:09:54.539 --> 00:09:57.179 the revenue. Without that service layer, the 00:09:57.179 --> 00:09:59.429 churn rate is just too high. That makes perfect 00:09:59.429 --> 00:10:01.610 sense. It's a moat. But there's another angle 00:10:01.610 --> 00:10:03.809 here that I found really compelling. The idea 00:10:03.809 --> 00:10:06.149 that starting with services is actually the fastest 00:10:06.149 --> 00:10:08.750 way to build a product. This is the x -ray vision 00:10:08.750 --> 00:10:11.509 concept. Break that down for us, because usually 00:10:11.509 --> 00:10:13.370 people think you build the product, then you 00:10:13.370 --> 00:10:16.210 sell it. OK, so everyone wants the passive income 00:10:16.210 --> 00:10:18.750 dream, right? Build the software once, sell it 00:10:18.750 --> 00:10:21.570 a million times while you sleep. But if you start 00:10:21.570 --> 00:10:24.009 there, you're usually guessing at what people 00:10:24.009 --> 00:10:27.100 need. You're building in a vacuum. Exactly. You 00:10:27.100 --> 00:10:29.559 build as a solution in search of a problem. But 00:10:29.559 --> 00:10:31.940 if you start as a service provider, let's say 00:10:31.940 --> 00:10:34.539 you're manually managing SEO for 10 different 00:10:34.539 --> 00:10:37.340 clients, you get x -revisioned into the market. 00:10:37.679 --> 00:10:41.220 You see that client A, client B, and client C 00:10:41.220 --> 00:10:43.759 all struggle with the exact same workflow step. 00:10:44.139 --> 00:10:46.120 Maybe it's categorizing keywords or formatting 00:10:46.120 --> 00:10:49.580 headers. That repetition. That is the signal. 00:10:49.740 --> 00:10:52.320 That is the blueprint. Yes. You aren't writing 00:10:52.320 --> 00:10:54.440 code based on a hunch anymore. You're writing 00:10:54.440 --> 00:10:56.440 code based on the pain you felt doing it manually. 00:10:56.539 --> 00:10:59.679 Yeah. Even Andreas and Horowitz A16Z, their reports 00:10:59.679 --> 00:11:02.639 confirm this. The long -term AI winners often 00:11:02.639 --> 00:11:04.679 start with a heavy service layer. So we should 00:11:04.679 --> 00:11:07.480 essentially embrace the manual grind. Yes. The 00:11:07.480 --> 00:11:09.240 grind is what generates the blueprint for the 00:11:09.240 --> 00:11:12.139 software. I love that. It really validates the 00:11:12.139 --> 00:11:14.799 hard work. Okay, let's get practical. We have 00:11:14.799 --> 00:11:17.240 listeners who are employees, listeners who are 00:11:17.240 --> 00:11:19.919 founders, and listeners running agencies. Let's 00:11:19.919 --> 00:11:22.360 give them a roadmap. Start with the employee, 00:11:22.659 --> 00:11:24.860 the intrapreneur. If you have a job right now, 00:11:25.059 --> 00:11:28.159 do not quit yet. Use your job as a laboratory. 00:11:28.500 --> 00:11:31.299 A laboratory, I like that. Look at your own day. 00:11:31.899 --> 00:11:34.919 Find the tasks that drain you. The boring stuff. 00:11:35.440 --> 00:11:38.679 Use tools like Zapier or NATU to automate just 00:11:38.679 --> 00:11:41.080 your own desk. Once you save five or 10 hours 00:11:41.080 --> 00:11:42.539 a week, you have a prototype. You go to your 00:11:42.539 --> 00:11:44.700 boss and say, look, I made myself faster and 00:11:44.700 --> 00:11:47.500 more accurate. You prove the value first. Exactly. 00:11:47.679 --> 00:11:49.519 And then you get paid to learn. You're upgrading 00:11:49.519 --> 00:11:51.600 your skills on the company dime. OK, what about 00:11:51.600 --> 00:11:54.299 the new founder, the person starting from scratch 00:11:54.299 --> 00:11:57.299 today? The biggest warning here is, do not be 00:11:57.299 --> 00:11:59.820 the AI guy for everyone. The general is trapped. 00:11:59.940 --> 00:12:02.500 I can automate anything for anyone. It's deadly. 00:12:02.679 --> 00:12:05.779 You need to niche down hard. Be the AI automation 00:12:05.779 --> 00:12:08.879 expert for dental clinics or inventory specialist 00:12:08.879 --> 00:12:11.240 for local warehouses. How do you even find that 00:12:11.240 --> 00:12:13.740 niche if you don't already have one? Use AI to 00:12:13.740 --> 00:12:16.769 find it. You can literally ask ChatGPT. List 00:12:16.769 --> 00:12:19.470 10 industries with high average customer value 00:12:19.470 --> 00:12:22.789 and heavy manual admin tasks. It'll give you 00:12:22.789 --> 00:12:26.269 things like legal intake or tax document labeling. 00:12:26.490 --> 00:12:29.429 So the strategy is find a boring problem like 00:12:29.429 --> 00:12:32.769 missed patient calls, build a specific solution 00:12:32.769 --> 00:12:35.669 like a voice agent to book appointments, and 00:12:35.669 --> 00:12:38.990 then charge a setup fee plus a retainer. Precisely. 00:12:39.370 --> 00:12:42.370 You solve an expensive Boring problem. And finally, 00:12:42.409 --> 00:12:44.769 for the existing agency owners, the marketing 00:12:44.769 --> 00:12:47.230 firms, the design studios. For them, it is all 00:12:47.230 --> 00:12:50.070 about silent margin expansion. That sounds profitable. 00:12:50.190 --> 00:12:53.230 It is. You use AI internally to do the work 10 00:12:53.230 --> 00:12:56.710 times faster. But, and here's the secret, you 00:12:56.710 --> 00:12:58.710 keep your prices the same. So your delivery cost 00:12:58.710 --> 00:13:00.610 just drops through the floor. And your profit 00:13:00.610 --> 00:13:03.190 margin skyrockets. Plus, because a client already 00:13:03.190 --> 00:13:05.610 trusts you, you can sell them trust upsells. 00:13:05.690 --> 00:13:07.769 Sell them an AI training workshop. It's an easy 00:13:07.769 --> 00:13:09.450 sell because the relationship is already there. 00:13:09.580 --> 00:13:11.759 When you look at all these strategies, the employee, 00:13:11.919 --> 00:13:14.379 the founder, the agency, what do they all have 00:13:14.379 --> 00:13:17.120 in common? They all solve unsexy, boring problems 00:13:17.120 --> 00:13:20.360 instead of chasing hype. Unsexy, boring, profitable. 00:13:20.639 --> 00:13:23.100 That's the trifecta. Let's recap the big idea 00:13:23.100 --> 00:13:25.639 here. We started with the failure of the wrapper 00:13:25.639 --> 00:13:28.639 model. We talked about the need for a smart intern 00:13:28.639 --> 00:13:31.259 mentality. Right. And the realization that the 00:13:31.259 --> 00:13:33.899 winning business model isn't about the newest 00:13:33.899 --> 00:13:37.120 shiny model from Google or OpenAI. It's about 00:13:37.120 --> 00:13:39.960 customizing workflows, training humans to audit 00:13:39.960 --> 00:13:43.100 the machines, and acting as the operator. The 00:13:43.100 --> 00:13:46.080 money's in the gap. The gap between what AI can 00:13:46.080 --> 00:13:49.399 do and what businesses actually do. That's the 00:13:49.399 --> 00:13:51.659 best way to put it. Before we go, we always like 00:13:51.659 --> 00:13:54.299 to leave you with something to do, not just something 00:13:54.299 --> 00:13:57.419 to think about. Here's your challenge. Pick one 00:13:57.419 --> 00:14:00.299 manual process in your daily life today. Just 00:14:00.299 --> 00:14:02.879 one. Spend 30 minutes mapping it out on paper, 00:14:03.120 --> 00:14:05.799 step by step. And then just ask yourself, how 00:14:05.799 --> 00:14:08.039 can I help a human do this faster? That is where 00:14:08.039 --> 00:14:11.879 it starts. The year is 2026. There are no true 00:14:11.879 --> 00:14:14.399 experts yet. Everyone is figuring this out in 00:14:14.399 --> 00:14:16.899 real time. If you are willing to do the unsexy 00:14:16.899 --> 00:14:19.240 work, the future belongs to you. Couldn't have 00:14:19.240 --> 00:14:20.600 said it better. Thanks for diving in with us. 00:14:20.740 --> 00:14:21.480 We'll see you on the next one.