WEBVTT 00:00:00.000 --> 00:00:02.680 For as long as Silicon Valley has existed, there 00:00:02.680 --> 00:00:06.940 has been a hierarchy. You had the engineers at 00:00:06.940 --> 00:00:09.380 the top, the builders. Right, the people who 00:00:09.380 --> 00:00:10.859 actually spoke the language of the machines. 00:00:11.119 --> 00:00:12.820 And then way down at the bottom, you had the 00:00:12.820 --> 00:00:16.019 idea guys. The person at the party is like, I 00:00:16.019 --> 00:00:18.260 have a million dollar app idea. I just need a 00:00:18.260 --> 00:00:21.320 tech nerd to build it. And for, what, 30 years, 00:00:21.420 --> 00:00:24.140 that tech nerd held all of the cards? If you 00:00:24.140 --> 00:00:26.239 couldn't wrestle with syntax and semicolons, 00:00:26.239 --> 00:00:28.980 your idea was basically worthless. But I think 00:00:28.980 --> 00:00:31.059 we're looking at a source today that suggests 00:00:31.059 --> 00:00:34.179 that the whole hierarchy is flipping. The idea 00:00:34.179 --> 00:00:36.859 guy might just be about to have the last laugh. 00:00:37.100 --> 00:00:39.600 It's a massive shift. It's huge. We're talking 00:00:39.600 --> 00:00:43.039 about a lead scraper sauce with real -time government 00:00:43.039 --> 00:00:46.909 APIs. A dashboard that has AI summarization built 00:00:46.909 --> 00:00:49.229 right in. All built in roughly 15 minutes. Without 00:00:49.229 --> 00:00:51.750 writing a single line of manual code. It is, 00:00:51.869 --> 00:00:54.310 and this isn't an exaggeration, the death of 00:00:54.310 --> 00:00:56.469 the six -figure budget. Welcome back to the Deep 00:00:56.469 --> 00:00:59.270 Dive. Today we are unpacking a guide by Maxon 00:00:59.270 --> 00:01:02.869 called Build a Sauce in 15 Minutes. The 2026 00:01:02.869 --> 00:01:06.569 No -Code AI Blueprint. Quite a title. It is. 00:01:06.670 --> 00:01:08.549 I have to be honest, when I first saw that, I 00:01:08.549 --> 00:01:10.090 kind of rolled my eyes. It sounds like total 00:01:10.090 --> 00:01:12.530 clickbait. It does. But then I looked at the 00:01:12.530 --> 00:01:14.989 actual stack he's using, this new Google trio, 00:01:15.349 --> 00:01:18.409 Gemini, Stitch, and Anti -Gravity. And the methodology 00:01:18.409 --> 00:01:21.730 feels different. It's a system. It's not just 00:01:21.730 --> 00:01:24.530 make me an app. Exactly. We're not exploring 00:01:24.530 --> 00:01:26.530 a faster way to code. We're exploring the shift 00:01:26.530 --> 00:01:29.870 from being a coder to being a director of systems. 00:01:30.109 --> 00:01:32.939 That's the core theme. Yeah. It's no longer about 00:01:32.939 --> 00:01:36.099 knowing how to mix the concrete. It's about knowing 00:01:36.099 --> 00:01:38.319 exactly where the building is supposed to stand. 00:01:38.519 --> 00:01:41.620 That is a bold claim. I want to test it. I want 00:01:41.620 --> 00:01:43.400 to see if this is, you know, actual engineering 00:01:43.400 --> 00:01:46.640 or if it's just a toy. So how are we breaking 00:01:46.640 --> 00:01:49.200 this down? Okay, let's treat it like a real engineering 00:01:49.200 --> 00:01:51.560 sprint. First, we have to understand the tools, 00:01:51.760 --> 00:01:54.500 the employees, because they are not just chatbots. 00:01:54.680 --> 00:01:57.620 We need to look at Gemini, Stitch, and anti -gravity. 00:01:57.680 --> 00:02:00.340 Then we'll walk through the build itself. phase 00:02:00.340 --> 00:02:04.359 by phase. So strategy, then design, then the 00:02:04.359 --> 00:02:06.819 actual code generation. And the data. We have 00:02:06.819 --> 00:02:09.500 to talk about the data. Oh, crucially, we'll 00:02:09.500 --> 00:02:12.219 look at the government APIs and that secret sauce 00:02:12.219 --> 00:02:16.560 of AI summarization. And then we have to do a 00:02:16.560 --> 00:02:18.759 reality check. Good. Because I don't care what 00:02:18.759 --> 00:02:21.360 the title says, things always break. We need 00:02:21.360 --> 00:02:24.159 to talk about what happens when the AI hallucinates 00:02:24.159 --> 00:02:27.120 and just leaves you with a blank screen. I am 00:02:27.120 --> 00:02:30.069 glad you mentioned that. I'm skeptical that A 00:02:30.069 --> 00:02:32.949 director of systems can really fix a broken Python 00:02:32.949 --> 00:02:36.110 dependency without, you know, knowing Python. 00:02:36.330 --> 00:02:38.430 Fair enough. But let's start with that mindset, 00:02:38.610 --> 00:02:40.370 clarity of outcome. That's the phrase the author 00:02:40.370 --> 00:02:42.789 uses. It sounds great, but what does it actually 00:02:42.789 --> 00:02:45.770 mean to replace coding with clarity? It means 00:02:45.770 --> 00:02:48.490 that execution is now a commodity. It's cheap. 00:02:48.750 --> 00:02:51.409 In the past, you could have a fuzzy idea, hire 00:02:51.409 --> 00:02:53.050 a team, and they would figure it out. Right. 00:02:53.129 --> 00:02:55.610 Now, since the AI is doing the building, you 00:02:55.610 --> 00:02:58.110 have to be the architect. And Anna introduces 00:02:58.110 --> 00:03:02.030 this Google triad that basically replaces a whole 00:03:02.030 --> 00:03:04.349 software team. Okay, let's unpack the team then. 00:03:04.409 --> 00:03:07.389 Who are we hiring? First up is Gemini. And in 00:03:07.389 --> 00:03:09.330 this workflow, Gemini isn't just a chat bot. 00:03:09.449 --> 00:03:11.949 It's your product manager. It's the strategist. 00:03:12.050 --> 00:03:15.349 Then you have Stitch. That's your UI UX master, 00:03:15.610 --> 00:03:18.069 the designer. And finally, you have the heavy 00:03:18.069 --> 00:03:20.870 lifter, Antigravity. Antigravity is the one that 00:03:20.870 --> 00:03:23.590 really caught my eye. The guide calls it an agentic 00:03:23.590 --> 00:03:26.289 IDE. We should probably pause on that for a second. 00:03:26.469 --> 00:03:30.189 Yeah, that's jargon. Agentic IDE. So IDE is an 00:03:30.189 --> 00:03:32.250 integrated development environment. It's the 00:03:32.250 --> 00:03:37.669 text editor coders use. But agentic, how is that 00:03:37.669 --> 00:03:39.669 different from me just pasting a prompt into 00:03:39.669 --> 00:03:42.669 ChatGPT? That is the crucial distinction. If 00:03:42.669 --> 00:03:45.530 you use ChatGPT, it just gives you text. You 00:03:45.530 --> 00:03:48.050 have to copy the text, switch windows, paste 00:03:48.050 --> 00:03:51.080 it in a file, try to run it. see an error, copy 00:03:51.080 --> 00:03:54.139 the error. It's just this tedious loop. The copy 00:03:54.139 --> 00:03:57.300 -paste loop, yeah. An agentic IDE like anti -gravity 00:03:57.300 --> 00:03:59.840 has hands. It lives inside your computer's terminal. 00:04:00.139 --> 00:04:03.020 So it has permission to execute commands. Yes. 00:04:03.240 --> 00:04:06.080 It installs dependencies, runs browser tests. 00:04:06.280 --> 00:04:09.439 It's not suggesting code. It is actively manipulating 00:04:09.439 --> 00:04:11.500 your machine's environment. So it's less like 00:04:11.500 --> 00:04:14.259 a chatbot and more like a remote junior developer 00:04:14.259 --> 00:04:16.620 who has control of my keyboard. Precisely. It's 00:04:16.620 --> 00:04:18.899 a text -to -software converter, not a text -to 00:04:18.899 --> 00:04:21.410 -text converter. That sounds incredibly powerful, 00:04:21.529 --> 00:04:23.949 but also a little unnerving. It can be. Couldn't 00:04:23.949 --> 00:04:25.870 I accidentally tell it to, I don't know, delete 00:04:25.870 --> 00:04:29.149 my hard drive? Theoretically, yes. Which brings 00:04:29.149 --> 00:04:31.149 us right back to that director of systems role. 00:04:31.350 --> 00:04:34.290 Yeah. The human isn't obsolete. You're the pilot. 00:04:34.769 --> 00:04:37.730 If you give vague instructions to a human developer, 00:04:37.910 --> 00:04:40.329 they'll ask you questions. If you give vague 00:04:40.329 --> 00:04:43.439 instructions to an agent, it just guesses. And 00:04:43.439 --> 00:04:46.540 AI guesses can be very confident and very wrong. 00:04:46.759 --> 00:04:50.100 Exactly. So the human's job becomes high context 00:04:50.100 --> 00:04:52.319 direction and setting guardrails. Which is a 00:04:52.319 --> 00:04:56.199 perfect segue to phase one, strategy. The source 00:04:56.199 --> 00:04:59.139 material makes a huge deal about the product 00:04:59.139 --> 00:05:02.139 requirements document, the PRD. Hallucinated 00:05:02.139 --> 00:05:04.920 code is the term he uses. If you skip the PRD, 00:05:04.980 --> 00:05:06.759 you're basically pouring concrete without any 00:05:06.759 --> 00:05:09.180 blueprints. You'll get a wall, but it might be 00:05:09.180 --> 00:05:11.300 blocking the front door. So the process starts 00:05:11.300 --> 00:05:13.180 with Gemini. Yep. You don't start by coding. 00:05:13.300 --> 00:05:15.500 You start by writing a really structured prompt 00:05:15.500 --> 00:05:18.040 to Gemini to define the whole product. And we're 00:05:18.040 --> 00:05:20.019 not just talking, I want a scraper. Yeah. The 00:05:20.019 --> 00:05:23.060 guide list specifics, user personas. Who's this 00:05:23.060 --> 00:05:25.660 actually for? The happy path, which is the main 00:05:25.660 --> 00:05:29.019 user flow. And interestingly, adherence to material 00:05:29.019 --> 00:05:32.639 design three. Why is that specific design standard 00:05:32.639 --> 00:05:35.259 so important here? Why not just say, make it 00:05:35.259 --> 00:05:39.029 look cool? Because cool is subjective. And AI 00:05:39.029 --> 00:05:41.990 really struggles with subjectivity. Right. Material 00:05:41.990 --> 00:05:44.709 Design 3 is Google's design language. It has 00:05:44.709 --> 00:05:48.129 strict rules, spacing, typography, color. By 00:05:48.129 --> 00:05:50.889 forcing the AI to use that standard. You're putting 00:05:50.889 --> 00:05:53.050 bumpers on the bowling lane. You're constraining 00:05:53.050 --> 00:05:55.250 it, which makes the output better. Exactly. You 00:05:55.250 --> 00:05:57.290 reduce the variables. Now, there was a pro tip 00:05:57.290 --> 00:05:59.529 in this phase that I liked. The author suggests 00:05:59.529 --> 00:06:02.250 asking Gemini for API recommendations at the 00:06:02.250 --> 00:06:04.629 very start. Yes. This is the director of systems 00:06:04.629 --> 00:06:07.529 mindset in action. Instead of figuring out later 00:06:07.529 --> 00:06:10.350 how to get the data, you ask the strategist Gemini 00:06:10.350 --> 00:06:13.029 to research federal databases and API docs right 00:06:13.029 --> 00:06:15.399 up front. So you walk out of phase one not with 00:06:15.399 --> 00:06:17.839 code, but with this massive detailed document. 00:06:18.040 --> 00:06:20.040 A document that describes exactly what success 00:06:20.040 --> 00:06:23.240 looks like. So is the role of the human obsolete? 00:06:23.660 --> 00:06:26.680 No. The human becomes the pilot. Vague instructions 00:06:26.680 --> 00:06:29.300 equal vague software. Okay, we have our blueprint. 00:06:29.560 --> 00:06:33.240 Phase two is design. We move from Gemini to Stitch. 00:06:33.579 --> 00:06:36.720 Now, I've seen AI image generators. They make 00:06:36.720 --> 00:06:40.079 beautiful, surreal art. That's not what we want 00:06:40.079 --> 00:06:42.360 for a software interface, is it? No. And that's 00:06:42.360 --> 00:06:44.480 the trap. You do not want pretty art. You want 00:06:44.480 --> 00:06:47.160 functional wireframes. The guide talks about 00:06:47.160 --> 00:06:50.240 using blueprint mode in Stitch. So it's low fidelity. 00:06:50.540 --> 00:06:53.019 Very low fidelity. And it strictly follows those 00:06:53.019 --> 00:06:55.160 material design rules we just established. And 00:06:55.160 --> 00:06:59.319 you have to set the aspect ratio to web, 16 .9. 00:06:59.579 --> 00:07:01.500 Right. And here's the step that saves you a ton 00:07:01.500 --> 00:07:03.800 of money. Yeah. You can click through the generated 00:07:03.800 --> 00:07:06.779 screens. So Stitch creates editable screens like 00:07:06.779 --> 00:07:10.360 login, dashboard, a settings page. Yep. And you 00:07:10.360 --> 00:07:12.800 test the logic visually. If a filter button is 00:07:12.800 --> 00:07:15.819 in a weird spot, you tell the AI to move it now 00:07:15.819 --> 00:07:18.060 before a single line of code exists. Because 00:07:18.060 --> 00:07:20.319 moving a pixel in a design tool is basically 00:07:20.319 --> 00:07:23.180 free. Moving a component in a React code base 00:07:23.180 --> 00:07:26.139 can break 20 other things. Exactly. It creates 00:07:26.139 --> 00:07:28.600 a visual boundary for the AI. When we get to 00:07:28.600 --> 00:07:31.220 the coding phase, the AI isn't guessing what 00:07:31.220 --> 00:07:33.060 the dashboard should look like. It's just translating 00:07:33.060 --> 00:07:36.660 the picture into code. So why is visual prototyping 00:07:36.660 --> 00:07:39.480 so much cheaper than coding? It creates a visual 00:07:39.480 --> 00:07:42.500 boundary for the AI, which prevents logic errors 00:07:42.500 --> 00:07:44.879 later on. Which brings us to the main event, 00:07:45.060 --> 00:07:48.879 phase three, the flint end build. This is where 00:07:48.879 --> 00:07:51.500 we bring in anti -gravity. The author calls this 00:07:51.500 --> 00:07:54.360 the moment of wonder. It really is. You take 00:07:54.360 --> 00:07:56.300 that folder, you export it from Stitcher Design, 00:07:56.579 --> 00:07:58.620 and you just drop it into anti -gravity. And 00:07:58.620 --> 00:08:00.980 then you paste in that big PRD from step one. 00:08:01.100 --> 00:08:03.180 Right. And anti -gravity acts as your senior 00:08:03.180 --> 00:08:05.420 front -end engineer. It reads the design, it 00:08:05.420 --> 00:08:07.160 reads the requirements, and it just starts working. 00:08:07.300 --> 00:08:09.920 It creates the React app, it sets up the routing. 00:08:10.060 --> 00:08:12.740 So login goes to the login page, dashboard to 00:08:12.740 --> 00:08:14.319 the dashboard. It installs the dependencies. 00:08:15.550 --> 00:08:17.370 This is that agentic nature we talked about. 00:08:17.410 --> 00:08:19.350 It's literally running NPM install. It's setting 00:08:19.350 --> 00:08:21.670 up the whole file structure. I have to admit, 00:08:21.769 --> 00:08:23.889 this is where I usually get skeptical. I've played 00:08:23.889 --> 00:08:25.930 with these tools, and I still wrestle with prompt 00:08:25.930 --> 00:08:28.629 drift myself. Sometimes the AI just forgets the 00:08:28.629 --> 00:08:30.829 design system halfway through. That's a very 00:08:30.829 --> 00:08:33.590 real vulnerability. The guide actually mentions 00:08:33.590 --> 00:08:36.190 this. You have to be explicit. You have to keep 00:08:36.190 --> 00:08:38.970 reminding it. Material Design 3, MUI standards, 00:08:39.370 --> 00:08:42.269 you are the guardrails. Okay. But the amazing 00:08:42.269 --> 00:08:45.159 part is, once it builds that first version, It 00:08:45.159 --> 00:08:47.139 launches a local server. You could open your 00:08:47.139 --> 00:08:49.299 browser and see the app running on your machine. 00:08:49.500 --> 00:08:51.120 But what happens when you open it and it's just 00:08:51.120 --> 00:08:54.179 a blank page or there's a huge error message? 00:08:54.679 --> 00:08:57.279 Because that definitely happens. It does. And 00:08:57.279 --> 00:08:59.659 in the old days, you Google the error code and 00:08:59.659 --> 00:09:02.480 scroll through Stack Overflow for an hour. Now, 00:09:02.720 --> 00:09:06.080 you just screenshot the error. You feed the image 00:09:06.080 --> 00:09:07.919 back to anti -gravity and you say, fix this. 00:09:08.159 --> 00:09:10.659 And it iterates. What happens when that first 00:09:10.659 --> 00:09:13.440 version is blank or broken? You screenshot the 00:09:13.440 --> 00:09:16.149 error. feed it back to the agent, and it iterates. 00:09:16.309 --> 00:09:18.190 Okay, so we have a front end. It looks like an 00:09:18.190 --> 00:09:20.230 app, but for now it's just a Hollywood set, right? 00:09:20.330 --> 00:09:22.309 It's a facade. Yeah, if you refresh the page, 00:09:22.429 --> 00:09:24.690 everything disappears. We need to talk about 00:09:24.690 --> 00:09:27.389 phases four and five refining and the back end. 00:09:27.509 --> 00:09:29.629 Refining is that tight loop you mentioned. Exactly. 00:09:30.009 --> 00:09:33.409 Ask, review, refine. Add a sidebar for project 00:09:33.409 --> 00:09:37.149 details. The agent codes it, you check it. But 00:09:37.149 --> 00:09:40.149 the back end, that's the brain of the operation. 00:09:40.700 --> 00:09:43.019 The guide recommends Supabase here, which is 00:09:43.019 --> 00:09:46.460 basically a wrapper for Postgres. Why use that 00:09:46.460 --> 00:09:49.000 instead of just say local files for a simple 00:09:49.000 --> 00:09:51.000 project? Because if you want this to be a real 00:09:51.000 --> 00:09:53.860 product, it needs persistence. It has to remember 00:09:53.860 --> 00:09:57.360 users. It needs to store data securely. Using 00:09:57.360 --> 00:09:59.440 Supabase gets you a production -grade database 00:09:59.440 --> 00:10:01.820 from day one. And there's a security note here 00:10:01.820 --> 00:10:04.820 that I think is critical. The AI knows to put 00:10:04.820 --> 00:10:08.639 API keys in a .env file. Yes. This is a classic 00:10:08.639 --> 00:10:10.779 rookie mistake, to hard -code your secret keys 00:10:10.779 --> 00:10:12.659 into the code itself where anyone can see them. 00:10:12.779 --> 00:10:15.019 Right. The AI is smart enough to abstract those 00:10:15.019 --> 00:10:16.799 into an environment file, which keeps the app 00:10:16.799 --> 00:10:19.000 secure. So the workflow is interesting. AnyGravity 00:10:19.000 --> 00:10:22.190 generates the SQL query, the database code. And 00:10:22.190 --> 00:10:24.789 you, the human, copy that into Supabase to run 00:10:24.789 --> 00:10:26.769 it. Right. It'll create tables for user profiles, 00:10:26.950 --> 00:10:29.070 for saved opportunities, whatever you need. Then 00:10:29.070 --> 00:10:31.450 anti -gravity builds a Python server using Flask 00:10:31.450 --> 00:10:33.870 or FastAPI to handle the logic. So now you have 00:10:33.870 --> 00:10:36.149 a front -end talking to a back -end, which talks 00:10:36.149 --> 00:10:38.590 to a database. That is a full stack application. 00:10:39.029 --> 00:10:41.769 So why use Supabase instead of just local files? 00:10:41.950 --> 00:10:44.149 Can make it a real product that persists data 00:10:44.149 --> 00:10:46.490 and remembers users. It's the classic stack, 00:10:46.710 --> 00:10:49.409 just built in minutes. Okay, now let's get to 00:10:49.409 --> 00:10:52.000 the good stuff. Phase six. The secret sauce. 00:10:52.799 --> 00:10:55.980 We're building a lead scraper. We need data. 00:10:56.179 --> 00:10:58.740 We have to connect this beautiful shell to the 00:10:58.740 --> 00:11:01.960 real world. The guide details three specific 00:11:01.960 --> 00:11:05.820 APIs. First, the GovCon API for scraping federal 00:11:05.820 --> 00:11:08.960 contracts. Okay. Second, USA Spending, which 00:11:08.960 --> 00:11:11.100 is a huge public database of government expenditure. 00:11:11.379 --> 00:11:13.899 Right. And third, a news API to get articles 00:11:13.899 --> 00:11:15.720 about companies. And the integration process 00:11:15.720 --> 00:11:19.460 is just giving the documentation to the AI. That's 00:11:19.460 --> 00:11:21.419 the director of systems skill again. You don't 00:11:21.419 --> 00:11:23.279 need to know how to write the request. You copy 00:11:23.279 --> 00:11:25.879 the URL for the API documentation, paste it in 00:11:25.879 --> 00:11:28.100 the prompt, and say, integrate this. When a user 00:11:28.100 --> 00:11:31.200 searches for construction, query this API. And 00:11:31.200 --> 00:11:33.639 store the results in Supabase. Exactly. But just 00:11:33.639 --> 00:11:35.519 listing contracts is kind of boring. Everyone 00:11:35.519 --> 00:11:37.799 does that. The guide mentions a secret sauce 00:11:37.799 --> 00:11:40.720 around AI summarization. This is where the value 00:11:40.720 --> 00:11:43.899 add explodes. Federal contracts are so dense. 00:11:43.960 --> 00:11:47.860 There are 20, 50, 100 pages of bureaucratic legalese. 00:11:48.299 --> 00:11:51.000 Unreadable. Completely. Yeah. The secret sauce 00:11:51.000 --> 00:11:53.820 is plugging in a summarization layer. The app 00:11:53.820 --> 00:11:56.480 doesn't just fetch the contract. It feeds the 00:11:56.480 --> 00:11:59.259 text to an AI model, which turns it into a three 00:11:59.259 --> 00:12:02.600 -sentence summary. So the user doesn't see solicitation 00:12:02.600 --> 00:12:07.299 number 594 -B. No. They see Department of Defense 00:12:07.299 --> 00:12:10.220 is looking for AI cybersecurity vendors. Budget 00:12:10.220 --> 00:12:13.649 $5 million. due next Friday. You're transforming 00:12:13.649 --> 00:12:16.950 raw data into insight. That is what people pay 00:12:16.950 --> 00:12:18.970 for. So what is the ultimate value add here? 00:12:19.409 --> 00:12:21.950 Transforming raw data into insight via the AI 00:12:21.950 --> 00:12:24.350 summarization layer. So we've built the strategy, 00:12:24.529 --> 00:12:27.210 the design, the code, the database, the APIs. 00:12:27.710 --> 00:12:30.870 Now for the moment of truth, the test. The guide 00:12:30.870 --> 00:12:32.950 calls it the intelligence command center test. 00:12:33.129 --> 00:12:34.590 Right. You have to walk through the user journey. 00:12:34.690 --> 00:12:36.690 You search for artificial intelligence defense 00:12:36.690 --> 00:12:38.990 contracts. You see live results pop up. You click 00:12:38.990 --> 00:12:40.909 save on one of them. And when you go to your 00:12:40.909 --> 00:12:43.330 save tab, it should be there. Which proves the 00:12:43.330 --> 00:12:46.070 database connection works. Then you search for 00:12:46.070 --> 00:12:49.889 a company, say SpaceX. The news API fires, pulling 00:12:49.889 --> 00:12:52.330 recent articles. And if all those chains hold 00:12:52.330 --> 00:12:55.389 together. Search to API, to backend, to database, 00:12:55.610 --> 00:12:58.549 to frontend. You have a shipping product. It 00:12:58.549 --> 00:13:01.230 sounds incredible, but we did promise a reality 00:13:01.230 --> 00:13:04.970 check. We did. Segment G in our outline is literally 00:13:04.970 --> 00:13:07.230 the honest reality check. I want to slow down 00:13:07.230 --> 00:13:09.830 here because if I go home and try this, it is 00:13:09.830 --> 00:13:12.350 not going to be flawless. It won't be. Not at 00:13:12.350 --> 00:13:15.070 all. And Anne is very honest about this. The 00:13:15.070 --> 00:13:18.149 friction is real. Syntax breaks. Buttons look 00:13:18.149 --> 00:13:21.730 off. APIs refuse to connect because of some minor 00:13:21.730 --> 00:13:24.340 version mismatch. The guide mentions you might 00:13:24.340 --> 00:13:26.779 have to adjust spacing for the third time or 00:13:26.779 --> 00:13:29.360 guide the system through a logic error. It requires 00:13:29.360 --> 00:13:31.659 patience. It requires a willingness to iterate. 00:13:31.980 --> 00:13:34.720 You aren't typing the code, but you are debugging 00:13:34.720 --> 00:13:36.740 the logic. You have to look at the screen and 00:13:36.740 --> 00:13:38.940 say, that's wrong, and then explain why it's 00:13:38.940 --> 00:13:40.879 wrong to the agent. But let's look at the trade 00:13:40.879 --> 00:13:42.799 -off, even with that friction, with those errors. 00:13:43.100 --> 00:13:45.899 It is still 10 to 20 times faster than the old 00:13:45.899 --> 00:13:48.240 way. The traditional method. Right. With the 00:13:48.240 --> 00:13:50.080 old way, you're talking about a team of three 00:13:50.080 --> 00:13:53.419 to five developers. A burn rate of $150 ,000. 00:13:53.840 --> 00:13:56.299 Three to six months before you even see a working 00:13:56.299 --> 00:13:59.200 product. And here, you're doing it alone. Potentially 00:13:59.200 --> 00:14:01.820 in a weekend, for the cost of a few API subscriptions. 00:14:02.340 --> 00:14:05.620 That is the death of the developer aspect. Not 00:14:05.620 --> 00:14:07.980 that code is going away, but the cost of code 00:14:07.980 --> 00:14:10.820 is collapsing to near zero. The barrier is no 00:14:10.820 --> 00:14:14.179 longer capital. It's patience. So what is the 00:14:14.179 --> 00:14:16.740 only barrier left? Patience and the willingness 00:14:16.740 --> 00:14:20.000 to iterate through small errors. So if we step 00:14:20.000 --> 00:14:22.500 back, look at the big picture. What does this 00:14:22.500 --> 00:14:24.860 mean for the person listening right now? The 00:14:24.860 --> 00:14:27.860 person curious about the future of work? I think 00:14:27.860 --> 00:14:30.019 it means we're moving from a world of, can I 00:14:30.019 --> 00:14:33.100 code this? To a world of, do I know what I want 00:14:33.100 --> 00:14:35.659 to build? The skill set of the future is clarity. 00:14:35.980 --> 00:14:39.210 Clarity is the new coding. Precisely. The flow 00:14:39.210 --> 00:14:42.149 we just described, strategy with Gemini, design 00:14:42.149 --> 00:14:45.190 with Stitch, build with anti -gravity, it's a 00:14:45.190 --> 00:14:47.169 workflow that rewards the person who can think 00:14:47.169 --> 00:14:49.529 clearly about systems. If you can define the 00:14:49.529 --> 00:14:52.129 PRD, if you can visualize the outcome. You can 00:14:52.129 --> 00:14:54.769 build tools that used to require a venture -backed 00:14:54.769 --> 00:14:57.549 startup. It's incredibly empowering, but also 00:14:57.549 --> 00:15:00.009 a little daunting. The responsibility is entirely 00:15:00.009 --> 00:15:03.529 on you. There's no tech guy to blame if the vision 00:15:03.529 --> 00:15:06.330 is blurry. That's the tradeoff. You have incident 00:15:06.330 --> 00:15:08.350 leverage, but you also have total accountability. 00:15:08.750 --> 00:15:10.929 So here's a challenge for our listeners. Max 00:15:10.929 --> 00:15:14.230 Ann suggests trying to define a PRD this weekend, 00:15:14.350 --> 00:15:17.470 even if you don't build the app. Just sit down 00:15:17.470 --> 00:15:20.690 with Gemini and practice the skill of defining 00:15:20.690 --> 00:15:23.570 a product with that kind of rigor. I think that's 00:15:23.570 --> 00:15:26.769 a fantastic exercise. Yeah. Because even if you 00:15:26.769 --> 00:15:29.149 never run the code, just learning to think in 00:15:29.149 --> 00:15:32.570 happy paths and user personas and API integrations 00:15:32.570 --> 00:15:35.409 makes you a better thinker, period. And who knows? 00:15:35.429 --> 00:15:36.950 You might just accidentally build the next big 00:15:36.950 --> 00:15:38.870 thing in your spare time. Stranger things have 00:15:38.870 --> 00:15:41.730 happened. This has been a fascinating deep dive. 00:15:41.850 --> 00:15:44.529 A huge thanks to Max Ann for the source material. 00:15:44.929 --> 00:15:48.190 And remember, clarity is your superpower. Thanks 00:15:48.190 --> 00:15:49.350 for listening. See you next time.