WEBVTT 00:00:00.000 --> 00:00:02.580 So today, we're gonna skip the complexity, we're 00:00:02.580 --> 00:00:07.599 diving into a really strange and frankly pretty 00:00:07.599 --> 00:00:09.759 lucrative corner of the tech world. Yeah, it's 00:00:09.759 --> 00:00:12.060 like, think about all those giant venture -backed 00:00:12.060 --> 00:00:14.419 companies you read about, the ones building battleships. 00:00:15.099 --> 00:00:17.079 We're not looking at those, we're looking at 00:00:17.079 --> 00:00:20.219 the simple, hyper -efficient speed boats. And 00:00:20.219 --> 00:00:23.120 that is the absolute core idea we're diving into. 00:00:23.539 --> 00:00:26.179 These are small, really focused mobile apps, 00:00:26.399 --> 00:00:28.980 sometimes built by just one person. And they're 00:00:28.980 --> 00:00:33.159 making, what, $50 ,000 to sometimes $300 ,000 00:00:33.159 --> 00:00:36.140 every single month. Every single month. And they 00:00:36.140 --> 00:00:38.000 fly completely under the radar. Because they 00:00:38.000 --> 00:00:41.159 only do one simple thing. But that focus, that's 00:00:41.159 --> 00:00:43.340 what makes them ridiculously profitable. It just 00:00:43.340 --> 00:00:45.759 flips the old scaling playbook on its head. Welcome 00:00:45.759 --> 00:00:48.310 to the deep dive. Today we're digging into the 00:00:48.310 --> 00:00:51.409 source material on these simple AI sauce models 00:00:51.409 --> 00:00:53.810 to pull out the exact patterns of repeatable 00:00:53.810 --> 00:00:56.570 stuff that makes them work. Our mission here 00:00:56.570 --> 00:00:58.549 is to really understand how these founders are 00:00:58.549 --> 00:01:02.270 building so much value with what looks like minimal 00:01:02.270 --> 00:01:05.510 complexity. Exactly. So first we'll unpack why 00:01:05.510 --> 00:01:07.930 those old barriers to entry, you know, needing 00:01:07.930 --> 00:01:10.510 to be a coding genius or having tons of cash, 00:01:10.890 --> 00:01:14.459 have just completely vanished. And then... We're 00:01:14.459 --> 00:01:17.739 going to go deep on five specific real -world 00:01:17.739 --> 00:01:20.280 examples. We'll break down the psychology, the 00:01:20.280 --> 00:01:22.920 business case, and even the simple AI prompts 00:01:22.920 --> 00:01:24.900 that are powering the whole thing. You know, 00:01:25.000 --> 00:01:28.159 if you look back at the last, say, 20 years of 00:01:28.159 --> 00:01:31.439 tech, starting a real software business... It 00:01:31.439 --> 00:01:33.939 felt like you needed a PhD in computer science. 00:01:34.260 --> 00:01:36.480 Oh, absolutely. I mean, just to get a prototype 00:01:36.480 --> 00:01:39.400 running, you were talking months of coding school, 00:01:39.780 --> 00:01:43.060 wrestling with syntax. Right. It was a nightmare. 00:01:43.180 --> 00:01:45.920 It was this huge technical wall. It kept the 00:01:45.920 --> 00:01:47.939 classic idea guy, the person who actually understands 00:01:47.939 --> 00:01:50.140 the customer's problem, completely out of the 00:01:50.140 --> 00:01:51.959 process. Totally. If you weren't a coder, you 00:01:51.959 --> 00:01:54.040 had to, what, beg one to be your co -founder 00:01:54.040 --> 00:01:56.000 or go out and try to find millions in funding. 00:01:56.079 --> 00:01:58.140 But those walls have just, they've crumbled. 00:01:58.400 --> 00:02:00.280 That technical barrier is, for the most part, 00:02:00.400 --> 00:02:03.879 gone now. Right. With tools like Cursor or Replet, 00:02:04.060 --> 00:02:06.939 all these AI developer assistants, you can basically 00:02:06.939 --> 00:02:09.280 just describe what you want in plain English. 00:02:09.699 --> 00:02:12.039 And it generates functional code. It's like a 00:02:12.039 --> 00:02:14.379 language translator for engineering. It just 00:02:14.379 --> 00:02:16.860 completely democratizes the whole process. The 00:02:16.860 --> 00:02:18.780 person with the vision, the one who really gets 00:02:18.780 --> 00:02:21.719 the market need, can finally be the builder. 00:02:21.960 --> 00:02:23.740 And this is where the really fundamental shift 00:02:23.740 --> 00:02:25.639 happens. This is what makes these little six 00:02:25.639 --> 00:02:28.240 -figure businesses possible. AI isn't just a 00:02:28.240 --> 00:02:31.379 helper tool for the engineer anymore. The AI 00:02:31.379 --> 00:02:35.680 is the product. Yes. And the pattern is so simple. 00:02:35.900 --> 00:02:37.819 All five examples we're going to look at follow 00:02:37.819 --> 00:02:40.960 it perfectly. The user gives a little bit of 00:02:40.960 --> 00:02:43.360 input. you know, photo, voice note, whatever. 00:02:43.639 --> 00:02:47.020 Then the magic happens. The AI model crunches 00:02:47.020 --> 00:02:49.979 it. And then you get the output. Which is the 00:02:49.979 --> 00:02:52.379 valuable answer, the personalized result. Exactly. 00:02:52.580 --> 00:02:54.780 And all the heavy lifting, the really complex 00:02:54.780 --> 00:02:57.419 stuff, is being handled by these huge models 00:02:57.419 --> 00:03:00.960 like GPT -4 or Claude or stable diffusion. So 00:03:00.960 --> 00:03:03.039 these founders, they're not trying to build a 00:03:03.039 --> 00:03:05.319 new brain from scratch. No way. They're just 00:03:05.319 --> 00:03:08.800 connecting a very specific user to the powerful 00:03:08.800 --> 00:03:11.099 brain that already exists. The source material 00:03:11.099 --> 00:03:13.409 calls that connection the simple door. I love 00:03:13.409 --> 00:03:15.729 that. The founder's main job is just to design 00:03:15.729 --> 00:03:18.129 the easiest, most elegant door for people to 00:03:18.129 --> 00:03:20.949 walk through to get to that AI power. They solve 00:03:20.949 --> 00:03:24.150 that last mile problem. So if the core AI is 00:03:24.150 --> 00:03:26.930 doing all that heavy lifting, what's the one 00:03:26.930 --> 00:03:29.610 non -negotiable step a founder has to get right 00:03:29.610 --> 00:03:32.669 to hit that $50 ,000 a month floor? They have 00:03:32.669 --> 00:03:36.090 to design a door that instantly solves a specific, 00:03:36.610 --> 00:03:39.990 painful problem for a really narrow group of 00:03:39.990 --> 00:03:42.240 people. That distinction is everything. Yeah. 00:03:42.379 --> 00:03:43.780 OK, let's jump into the real data. We're going 00:03:43.780 --> 00:03:45.379 to look at these five businesses and see how 00:03:45.379 --> 00:03:47.379 they did it. First up, we've got Flash Loop. 00:03:47.699 --> 00:03:50.360 It's a viral video character creator, and it's 00:03:50.360 --> 00:03:52.639 reliably pulling in about 50 grand a month just 00:03:52.639 --> 00:03:56.180 by tapping into pure human vanity. The psychology 00:03:56.180 --> 00:03:58.780 here is what the source calls the vanity loop. 00:03:58.800 --> 00:04:00.960 Yeah. And it makes perfect sense. People love 00:04:00.960 --> 00:04:03.539 seeing themselves in, you know, funny or surprising 00:04:03.539 --> 00:04:05.780 situations. And more than that, they love the 00:04:05.780 --> 00:04:08.000 social currency they get from sharing it. Right. 00:04:08.020 --> 00:04:10.360 So what Flash Loop does is you give it a selfie, 00:04:10.879 --> 00:04:13.340 and it just plops your face onto a character 00:04:13.340 --> 00:04:16.500 in some funny, super shareable video. So you'd 00:04:16.500 --> 00:04:18.920 be the star of a movie trailer. Or, the more 00:04:18.920 --> 00:04:21.180 common one, your face is suddenly on a little 00:04:21.180 --> 00:04:23.199 baby who's dancing around in a business suit. 00:04:23.300 --> 00:04:25.560 It's ridiculous. And the technology behind this 00:04:25.560 --> 00:04:28.120 has shifted. You don't even need pre -film templates 00:04:28.120 --> 00:04:32.189 anymore. Nope. They use AI video tools like runway 00:04:32.189 --> 00:04:35.629 or maybe cling AI to generate the whole scene 00:04:35.629 --> 00:04:38.410 from scratch just from a text prompt Then they 00:04:38.410 --> 00:04:40.829 use face swapping to put you in it So the prompt 00:04:40.829 --> 00:04:43.069 is basically the entire product design pretty 00:04:43.069 --> 00:04:44.550 much you'd write something super descriptive 00:04:44.550 --> 00:04:47.769 like a cinematic shot of a cute chubby baby wearing 00:04:47.769 --> 00:04:51.069 a business suit Hosting a podcast in a professional 00:04:51.069 --> 00:04:53.589 studio and that just builds the viral growth 00:04:53.589 --> 00:04:55.709 right into the product every time someone shares 00:04:55.709 --> 00:04:58.209 their video on tik -tok It's a free ad for the 00:04:58.209 --> 00:05:00.910 app. Exactly. OK, so from pure vanity, we're 00:05:00.910 --> 00:05:04.569 going to pivot completely to utility and faith. 00:05:05.269 --> 00:05:08.569 Next up is the Bible Note Taker. It's a niche 00:05:08.569 --> 00:05:11.910 sermon recorder, quietly making $60 ,000 a month. 00:05:12.350 --> 00:05:14.810 This is just a textbook example of finding a 00:05:14.810 --> 00:05:17.129 recurring pain point in a very specific community. 00:05:17.329 --> 00:05:19.110 People go to church every week. They hear a sermon 00:05:19.110 --> 00:05:21.910 that really resonates. And then by Monday morning, 00:05:22.129 --> 00:05:25.199 all the key lessons are just gone. evaporated 00:05:25.199 --> 00:05:27.800 so the door is super simple you open the app 00:05:27.800 --> 00:05:31.379 on Sunday hit record the app uses AI transcription 00:05:31.379 --> 00:05:34.379 and they lean heavily on tools like open AI whisper 00:05:34.379 --> 00:05:38.180 which is just It's phenomenal at turning speech, 00:05:38.560 --> 00:05:41.579 even in a big, echoey room, into accurate text. 00:05:41.720 --> 00:05:43.579 But the real money isn't just in the transcription. 00:05:43.819 --> 00:05:45.339 It's in the synthesis. It doesn't just give you 00:05:45.339 --> 00:05:47.920 a wall of text. No, it uses something like GPT 00:05:47.920 --> 00:05:51.000 -4 to structure it all. The output is a bullet 00:05:51.000 --> 00:05:53.079 point summary of the main lessons. It suggests 00:05:53.079 --> 00:05:55.860 a prayer. And it gives you one actionable step 00:05:55.860 --> 00:05:58.720 for the week. It's immediate, personalized value. 00:05:59.160 --> 00:06:02.100 And the prompt is key here, too. It tells the 00:06:02.100 --> 00:06:05.019 AI to act as a helpful spiritual assistant and 00:06:05.019 --> 00:06:07.779 keep the tone encouraging and warm. And that 00:06:07.779 --> 00:06:10.920 focus on trust and a weekly habit. It's perfect 00:06:10.920 --> 00:06:13.579 for a subscription model. It's a need that the 00:06:13.579 --> 00:06:15.879 big, general note -taking apps would never even 00:06:15.879 --> 00:06:17.899 think about. OK, so let's move to an even bigger 00:06:17.899 --> 00:06:21.040 financial pain point. AI Home Decor, the room 00:06:21.040 --> 00:06:23.540 visualizer. This one's pulling in $100 ,000 a 00:06:23.540 --> 00:06:25.779 month. Yeah, and this solves that universal, 00:06:26.120 --> 00:06:29.600 very expensive problem. The fear of buying. The 00:06:29.600 --> 00:06:32.079 risk of spending $2 ,000 on a sofa you end up 00:06:32.079 --> 00:06:34.220 hating. Right. Or picking a paint color that 00:06:34.220 --> 00:06:37.680 just ruins the room. It's paralyzing. So for 00:06:37.680 --> 00:06:40.420 a $10 subscription, this app is basically cheap 00:06:40.420 --> 00:06:42.819 insurance. You upload a photo of your room, and 00:06:42.819 --> 00:06:44.819 you can instantly see it in any style you want. 00:06:45.060 --> 00:06:47.480 Japanese minimalist, art deco, whatever. It just 00:06:47.480 --> 00:06:50.079 removes that huge financial risk. And technically, 00:06:50.399 --> 00:06:53.060 this all hinges on a very specific technology 00:06:53.060 --> 00:06:56.040 called ControlNet. That's the secret sauce. OK, 00:06:56.060 --> 00:06:58.040 let's unpack that for a second, because ControlNet's 00:06:58.040 --> 00:07:00.660 precision is so important here. Normally, with 00:07:00.660 --> 00:07:03.660 an AI image generator, the results are kind of 00:07:03.660 --> 00:07:07.000 wild and unpredictable. Control net is different. 00:07:07.160 --> 00:07:10.000 It acts like a digital stencil. It locks the 00:07:10.000 --> 00:07:12.699 geometry. So the walls, the windows, the light 00:07:12.699 --> 00:07:15.040 fixtures, they all stay in the exact same place. 00:07:15.680 --> 00:07:18.879 Then the AI, like stable diffusion, just repaints 00:07:18.879 --> 00:07:21.100 everything inside those lines. If you didn't 00:07:21.100 --> 00:07:23.339 have that, you'd get a new room design with windows 00:07:23.339 --> 00:07:25.620 on the ceiling. Exactly. So the prompt has to 00:07:25.620 --> 00:07:28.230 command that. Something like, redesign this room, 00:07:28.569 --> 00:07:30.670 but keep the window placement and room structure 00:07:30.670 --> 00:07:33.269 exactly the same. You know, I have to admit, 00:07:33.689 --> 00:07:36.550 even with these tools, getting that control just 00:07:36.550 --> 00:07:39.670 right is a battle. I still wrestle with prompt 00:07:39.670 --> 00:07:41.790 drift myself when I'm using control net. Yeah. 00:07:41.910 --> 00:07:44.189 Yeah, you know, you try to change one small detail, 00:07:44.629 --> 00:07:47.290 like the wood grain, and suddenly the whole structure 00:07:47.290 --> 00:07:50.610 gets a little wonky. It takes real skill to get 00:07:50.610 --> 00:07:53.009 that consistency. But when you nail it, apparently 00:07:53.009 --> 00:07:56.889 you get to seven figures. OK, moving on. Our 00:07:56.889 --> 00:07:58.850 fourth example taps into something we do all 00:07:58.850 --> 00:08:02.009 day, every day. It's called MojiLab, a custom 00:08:02.009 --> 00:08:05.009 sticker maker, also making 100K a month. This 00:08:05.009 --> 00:08:07.569 one is just pure frequency plus social status. 00:08:07.750 --> 00:08:10.709 We're on messaging apps constantly, and standard 00:08:10.709 --> 00:08:13.410 emojis get old fast. So custom stickers of your 00:08:13.410 --> 00:08:15.990 friends or your cat or yourself, that's like 00:08:15.990 --> 00:08:17.870 high value currency in a group chat. It creates 00:08:17.870 --> 00:08:19.970 this automatic viral loop. Right, because if 00:08:19.970 --> 00:08:22.230 I send a sticker of my dog dressed as a pirate. 00:08:22.589 --> 00:08:24.779 Everyone in the chat immediately asks, Wait, 00:08:24.800 --> 00:08:26.699 how did you make that? And that question is all 00:08:26.699 --> 00:08:29.680 the marketing you need. The tech behind it uses 00:08:29.680 --> 00:08:34.000 powerful APIs like Deli 3 or Mid Journey. But 00:08:34.000 --> 00:08:36.879 the real trick isn't just making the image. No, 00:08:36.919 --> 00:08:39.639 it's making it look like a real usable sticker. 00:08:39.840 --> 00:08:41.820 So the key detail on the prompt is all about 00:08:41.820 --> 00:08:44.600 the packaging. It has to demand white background, 00:08:44.940 --> 00:08:47.139 thick white border around the character, skicker 00:08:47.139 --> 00:08:49.039 pack effect. Without that, it's just a picture. 00:08:49.220 --> 00:08:52.230 With it, it's a product. OK, finally, we're turning 00:08:52.230 --> 00:08:55.169 to the world of high -intent collectors. Vinyl 00:08:55.169 --> 00:08:57.269 Snap, the collector's price guide. This one's 00:08:57.269 --> 00:09:00.269 making $70 ,000 a month. This targets a really 00:09:00.269 --> 00:09:02.690 specific high -stakes moment. You're a collector. 00:09:02.870 --> 00:09:04.490 You're at a garage sale. You're holding a record. 00:09:04.629 --> 00:09:08.070 And you need to know right now, is this a $5 00:09:08.070 --> 00:09:12.269 common pressing, or is this a rare $500 first 00:09:12.269 --> 00:09:14.580 edition? The seconds matter. They absolutely 00:09:14.580 --> 00:09:17.700 do. So the door is your phone's camera, you scan 00:09:17.700 --> 00:09:20.480 the album cover, and the app instantly identifies 00:09:20.480 --> 00:09:23.320 the artist, the album, the specific catalog number. 00:09:23.419 --> 00:09:25.299 Which is the critical detail for collectors. 00:09:25.480 --> 00:09:27.840 Right. And then it estimates the condition and 00:09:27.840 --> 00:09:30.080 gives you the current market value from a database 00:09:30.080 --> 00:09:33.720 like Discogs. The technology here is AI vision. 00:09:34.200 --> 00:09:35.980 We're talking tools like Google Cloud Vision 00:09:35.980 --> 00:09:38.889 or GPT -4 Vision. They're trained to read that 00:09:38.889 --> 00:09:41.649 complex, sometimes faded text on an album cover 00:09:41.649 --> 00:09:45.190 with incredible accuracy. Whoa. Just, I mean, 00:09:45.509 --> 00:09:48.649 stop and imagine scaling that accuracy to a billion 00:09:48.649 --> 00:09:51.110 queries a day for every niche collector looking 00:09:51.110 --> 00:09:53.769 for that hidden gem. The speed and precision 00:09:53.769 --> 00:09:56.470 required to turn a phone photo into a verified 00:09:56.470 --> 00:09:59.129 price in milliseconds. That's a phenomenal feat. 00:09:59.330 --> 00:10:01.049 And that speed is the whole value proposition. 00:10:01.669 --> 00:10:04.049 A regular Google search is just too slow, too 00:10:04.049 --> 00:10:06.350 clunky for that moment. This app can pay for 00:10:06.350 --> 00:10:09.629 itself with just one good find. So if we synthesize 00:10:09.629 --> 00:10:11.750 all five of these businesses, from the vanity 00:10:11.750 --> 00:10:14.610 videos all the way to the collector tools, what's 00:10:14.610 --> 00:10:17.429 the one single defining characteristic that ensures 00:10:17.429 --> 00:10:20.129 they're profitable? They all offer a near instantaneous 00:10:20.129 --> 00:10:22.870 personalized answer to a problem that all the 00:10:22.870 --> 00:10:25.990 existing general tools just solve poorly or slowly 00:10:25.990 --> 00:10:28.129 or not at all. And that really brings us back 00:10:28.129 --> 00:10:30.169 to the core idea we pulled from the source material. 00:10:30.490 --> 00:10:33.330 These profitable AI Saws models are so effective 00:10:33.330 --> 00:10:35.960 because they take a simple human problem. Boredom, 00:10:36.179 --> 00:10:39.139 memory loss, fear. Exactly. Fear of commitment, 00:10:39.379 --> 00:10:43.399 curiosity. And they use this advanced AI to provide 00:10:43.399 --> 00:10:47.259 an immediate, valuable, and super -specific solution. 00:10:47.399 --> 00:10:49.980 And that's the massive opportunity right now. 00:10:49.980 --> 00:10:52.100 It's not about inventing the next GPT -4. That 00:10:52.100 --> 00:10:55.120 takes billions of dollars. Right. The real economic 00:10:55.120 --> 00:10:57.620 opportunity is in inventing that simple, focused 00:10:57.620 --> 00:11:00.200 interface, that elegant door that lets normal 00:11:00.200 --> 00:11:03.059 people apply all that existing AI power to their 00:11:03.059 --> 00:11:05.799 own specific needs. We've only covered the first 00:11:05.799 --> 00:11:08.340 five examples here. They show how that narrow 00:11:08.340 --> 00:11:11.120 focus creates high returns and how the tech is 00:11:11.120 --> 00:11:13.639 really accessible now. But part two of this research 00:11:13.639 --> 00:11:15.539 is where it gets even more interesting. It actually 00:11:15.539 --> 00:11:18.259 reveals the single top earner on the whole list. 00:11:18.419 --> 00:11:21.600 An app making a confirmed $300 ,000 a month. 00:11:21.789 --> 00:11:24.370 And it targets a totally different kind of anxiety. 00:11:25.250 --> 00:11:28.049 And maybe more importantly, part two introduces 00:11:28.049 --> 00:11:31.090 the master framework. Which is the checklist 00:11:31.090 --> 00:11:33.730 these founders use to test if an idea is profitable 00:11:33.730 --> 00:11:35.990 before they even write a single line of code. 00:11:36.230 --> 00:11:38.549 Yeah. I mean, we've seen six -figure success 00:11:38.549 --> 00:11:40.769 built around vinyl collectors and sermon notes. 00:11:41.070 --> 00:11:43.090 The common thread is just finding that quiet 00:11:43.090 --> 00:11:45.350 corner of the market. So here's the final thought 00:11:45.350 --> 00:11:49.200 to chew on. If focusing on these quiet, specific 00:11:49.200 --> 00:11:51.379 communities can yield these kinds of results, 00:11:52.299 --> 00:11:54.799 what small, overlooked community problem could 00:11:54.799 --> 00:11:58.620 you solve with AI vision or generation or transcription? 00:11:58.820 --> 00:12:01.600 The real opportunity is in those spaces the big 00:12:01.600 --> 00:12:04.220 companies are intentionally ignoring. We appreciate 00:12:04.220 --> 00:12:06.600 you joining us for this deep dive into these 00:12:06.600 --> 00:12:09.799 shockingly simple and highly profitable AI business 00:12:09.799 --> 00:12:11.879 models. We'll catch you next time. Happy building.