WEBVTT 00:00:00.000 --> 00:00:02.359 You stare at the screen. The Figma canvas is 00:00:02.359 --> 00:00:05.660 completely blank. Or maybe you just generated 00:00:05.660 --> 00:00:08.039 another landing page using an AI tool. Right. 00:00:08.199 --> 00:00:10.339 And it has that same generic purple gradient. 00:00:10.599 --> 00:00:12.599 You know the exact look I am talking about. It 00:00:12.599 --> 00:00:15.359 looks perfectly competent, but it feels, well, 00:00:15.460 --> 00:00:18.199 entirely soulless. It lacks any real perspective. 00:00:18.600 --> 00:00:21.480 Yeah, it's totally flat. Beat. Welcome to the 00:00:21.480 --> 00:00:23.460 Deep Dive. I am really glad you're joining us 00:00:23.460 --> 00:00:25.620 today. We are unpacking something that completely 00:00:25.620 --> 00:00:29.260 changes how we think about creativity and systems. 00:00:29.539 --> 00:00:31.519 We really are looking at a fundamental shift 00:00:31.519 --> 00:00:35.399 today. Our mission is decoding a methodology 00:00:35.399 --> 00:00:39.020 called Claude Code Design. We're exploring how 00:00:39.020 --> 00:00:42.079 you take abstract design rules, you know, the 00:00:42.079 --> 00:00:44.619 things we normally just call taste, and turn 00:00:44.619 --> 00:00:47.159 them into a repeatable plain text system. Which 00:00:47.159 --> 00:00:49.500 sounds impossible at first. It does. But the 00:00:49.500 --> 00:00:51.780 promise here is massive. Because once you build 00:00:51.780 --> 00:00:53.780 this underlying architecture, you never have 00:00:53.780 --> 00:00:57.920 to start from zero again. That concept of codifying 00:00:57.920 --> 00:01:00.039 taste is exactly what hooked me. So let's start 00:01:00.039 --> 00:01:01.979 by looking at the baseline problem we are all 00:01:01.979 --> 00:01:05.140 facing. Everyone recognizes current AI design. 00:01:05.420 --> 00:01:08.250 Oh, absolutely. It defaults to a highly specific, 00:01:08.609 --> 00:01:11.629 totally forgettable aesthetic. It is the digital 00:01:11.629 --> 00:01:15.069 equivalent of elevator music. You ask an AI to 00:01:15.069 --> 00:01:17.870 make a modern website, and it immediately outputs 00:01:17.870 --> 00:01:21.469 a massive, centered hero section. Always. Below 00:01:21.469 --> 00:01:24.530 that, you inevitably get three perfectly spaced 00:01:24.530 --> 00:01:27.150 feature cards. The buttons are always softly 00:01:27.150 --> 00:01:29.189 rounded. Right. And there's usually a subtle 00:01:29.189 --> 00:01:31.689 drop shadow on everything. It's clean, sure. 00:01:32.090 --> 00:01:35.109 But it has absolutely no point of view. It is 00:01:35.109 --> 00:01:38.549 technically proficient, but artistically vacant. 00:01:38.930 --> 00:01:41.349 And the root cause isn't a lack of computing 00:01:41.349 --> 00:01:43.950 power. The source material points directly to 00:01:43.950 --> 00:01:46.250 vague prompting. Yeah, the classic five -word 00:01:46.250 --> 00:01:48.269 prompt. We sit down and type, you know, make 00:01:48.269 --> 00:01:50.450 a modern fintech site. And that is simply not 00:01:50.450 --> 00:01:52.670 enough structural information. You really can't 00:01:52.670 --> 00:01:54.950 expect a machine to invent your brand's soul 00:01:54.950 --> 00:01:56.829 from a five -word prompt. It's just going to... 00:01:57.150 --> 00:01:59.469 guess. I still wrestle with prompt drift myself. 00:01:59.650 --> 00:02:02.030 I'll sit there late at night tweaking an AI prompt 00:02:02.030 --> 00:02:04.989 for hours. Oh man, we've all been there. I'll 00:02:04.989 --> 00:02:07.329 adjust the wording, ask it to make things sleeker 00:02:07.329 --> 00:02:10.530 or more dynamic. My eyes will literally start 00:02:10.530 --> 00:02:13.250 to hurt from staring at the variations. And after 00:02:13.250 --> 00:02:16.110 all that manual labor, the final result still 00:02:16.110 --> 00:02:18.789 looks artificially generated. That frustration 00:02:18.789 --> 00:02:22.580 is universal right now. The fix. requires a completely 00:02:22.580 --> 00:02:25.219 different approach. We have to translate our 00:02:25.219 --> 00:02:28.520 visual preferences into strict, measurable rules. 00:02:28.639 --> 00:02:31.020 Right. We need to dictate the exact typography 00:02:31.020 --> 00:02:33.580 scales. We need to define the spacing tokens. 00:02:33.819 --> 00:02:36.479 We really need to establish strict boundaries 00:02:36.479 --> 00:02:39.139 for layout and colors. It reminds me of working 00:02:39.139 --> 00:02:42.379 with a master chef. If you walk into a Michelin 00:02:42.379 --> 00:02:44.680 starred kitchen and just tell the chef, cook 00:02:44.680 --> 00:02:46.900 me something good. They're going to play it safe. 00:02:47.060 --> 00:02:50.099 Exactly. They will probably fall back on a standard 00:02:50.099 --> 00:02:53.139 safe dish. They don't know your palate. But if 00:02:53.139 --> 00:02:55.439 you hand them your grandmother's specific family 00:02:55.439 --> 00:02:58.539 recipe, complete with exact ratios for spices 00:02:58.539 --> 00:03:01.400 and precise resting times, you are guiding their 00:03:01.400 --> 00:03:04.360 technical expertise. That chef analogy perfectly 00:03:04.360 --> 00:03:06.379 captures the dynamic. You aren't just handing 00:03:06.379 --> 00:03:08.340 them the ingredients. You are defining the technique. 00:03:08.860 --> 00:03:11.080 The source material contrasts this beautifully. 00:03:11.300 --> 00:03:13.620 The weak prompt just asks for a modern website. 00:03:13.960 --> 00:03:16.680 But the better prompt. acts like that detailed 00:03:16.680 --> 00:03:20.379 recipe. It demands a, quote, clean fintech landing 00:03:20.379 --> 00:03:23.560 page with sharp pricing cards, dark navy accents, 00:03:23.699 --> 00:03:27.280 and strict sans serif typography. That specific 00:03:27.280 --> 00:03:30.099 language bridges the gap. You're stripping away 00:03:30.099 --> 00:03:33.740 the AI's ability to guess. You dictate that headlines 00:03:33.740 --> 00:03:36.840 must be oversized. You force the layout to maintain 00:03:36.840 --> 00:03:39.560 massive open sections. You restrict the color 00:03:39.560 --> 00:03:42.650 palette to three specific hex codes. Right. It 00:03:42.650 --> 00:03:45.129 forces the output out of that generic purple 00:03:45.129 --> 00:03:47.129 gradient territory. Which brings up an interesting 00:03:47.129 --> 00:03:50.430 philosophical point about the tools we use. So 00:03:50.430 --> 00:03:53.289 is it that the AI lacks the capability to design 00:03:53.289 --> 00:03:55.789 well, or do we just lack the vocabulary to instruct 00:03:55.789 --> 00:03:59.009 it? It is entirely about our instructions. The 00:03:59.009 --> 00:04:01.750 AI has the capability. It just needs us to define 00:04:01.750 --> 00:04:03.740 the boundaries of taste. Right. We just need 00:04:03.740 --> 00:04:06.120 to translate visual taste into specific text 00:04:06.120 --> 00:04:08.379 rules. Exactly. So if the fix is giving the AI 00:04:08.379 --> 00:04:11.460 a highly specific technical recipe, how do we 00:04:11.460 --> 00:04:13.419 actually write that document? That leads us straight 00:04:13.419 --> 00:04:15.460 into the mechanics of text -based design. Two 00:04:15.460 --> 00:04:18.240 sec silence. This is where the methodology gets 00:04:18.240 --> 00:04:20.850 incredibly practical. We are looking at a real 00:04:20.850 --> 00:04:23.029 -world repository called Awesome Designer AI. 00:04:23.329 --> 00:04:26.850 It's a GitHub tool that basically acts as a library 00:04:26.850 --> 00:04:29.449 of plain text design systems. That's the part 00:04:29.449 --> 00:04:32.089 that feels so counterintuitive to me. High -end, 00:04:32.110 --> 00:04:34.769 nuanced visual taste things that usually require 00:04:34.769 --> 00:04:37.649 a senior art director to understand can actually 00:04:37.649 --> 00:04:40.569 live inside a simple markdown text file. Yeah, 00:04:40.610 --> 00:04:42.689 we're talking about a specific file usually named 00:04:42.689 --> 00:04:46.089 design .md. This mundane little text document 00:04:46.089 --> 00:04:49.680 becomes Claude's ultimate design c***. Wow. It 00:04:49.680 --> 00:04:52.120 means your brand's visual identity is no longer 00:04:52.120 --> 00:04:54.899 locked inside proprietary software. It isn't 00:04:54.899 --> 00:04:57.480 trapped in a messy Canva template or, you know, 00:04:57.480 --> 00:05:00.680 a complex Figma file. It lives as raw code. I 00:05:00.680 --> 00:05:02.839 was reading through the awesome designer AI examples 00:05:02.839 --> 00:05:04.779 in the source, and it is fascinating how they 00:05:04.779 --> 00:05:07.860 break down major brands. Stripe is a great example 00:05:07.860 --> 00:05:10.139 to unpack here. Oh, Stripe is perfect. Stripe 00:05:10.139 --> 00:05:12.240 is widely considered the gold standard for clean 00:05:12.240 --> 00:05:14.620 sauce design. Stripe's design language is entirely 00:05:14.620 --> 00:05:17.600 focused on projecting trust and technical competence. 00:05:18.120 --> 00:05:20.680 Their plain text system doesn't just say, look 00:05:20.680 --> 00:05:23.720 trustworthy. It defines specific, measurable 00:05:23.720 --> 00:05:26.879 rules. It dictates crisp pricing layouts. Right. 00:05:27.240 --> 00:05:30.360 It enforces rigid rules around using massive 00:05:30.360 --> 00:05:33.019 amounts of white space to let the content breathe. 00:05:33.360 --> 00:05:35.319 Then you look at Spotify's rules, which are fundamentally 00:05:35.319 --> 00:05:39.379 opposed to Stripes. Spotify's text file dictates 00:05:39.379 --> 00:05:42.399 deep, rich, dark themes. Completely different 00:05:42.399 --> 00:05:45.899 vibe. It forces the AI to use bold, vibrant neon 00:05:45.899 --> 00:05:48.639 accents that pop against that dark background. 00:05:48.860 --> 00:05:52.139 The rules actively encourage heavy visual contrast. 00:05:52.620 --> 00:05:55.180 You also have SpaceX in that repository. Their 00:05:55.180 --> 00:05:58.319 brand is about scale. and drama so their text 00:05:58.319 --> 00:06:02.379 rules demand cinematic layouts the markdown dictates 00:06:02.379 --> 00:06:04.319 sweeping product sections that stretch across 00:06:04.319 --> 00:06:07.600 the entire viewport an apple system is predictably 00:06:07.600 --> 00:06:10.339 minimal the text file aggressively enforces a 00:06:10.339 --> 00:06:13.069 product first hierarchy The typography rules 00:06:13.069 --> 00:06:15.269 are incredibly rigid, stripping away anything 00:06:15.269 --> 00:06:17.670 that distracts from the core device being showcased. 00:06:17.689 --> 00:06:20.370 And Claude Ries' design .md file internalizes 00:06:20.370 --> 00:06:22.230 these strict constraints and then actually builds 00:06:22.230 --> 00:06:24.750 functional assets. It's wild. It writes the code 00:06:24.750 --> 00:06:27.189 for fully responsive pricing pages. It generates 00:06:27.189 --> 00:06:30.350 perfectly formatted HTML slide decks. It builds 00:06:30.350 --> 00:06:32.509 interactive website sections that perfectly match 00:06:32.509 --> 00:06:35.110 the Codify brand. I do have to push back a little 00:06:35.110 --> 00:06:37.550 bit on the mechanics here, though. I am slightly 00:06:37.550 --> 00:06:40.209 skeptical that a plain text document can truly 00:06:40.209 --> 00:06:43.970 capture an abstract vibe. Take SpaceX's cinematic 00:06:43.970 --> 00:06:47.230 feel. Okay, yeah. How does Claude know what cinematic 00:06:47.230 --> 00:06:50.129 actually means just from reading text without 00:06:50.129 --> 00:06:53.370 any visual aids? Claude already has broad training 00:06:53.370 --> 00:06:56.189 on those concepts. The text just sets the strict 00:06:56.189 --> 00:06:58.589 boundaries and constraints for it to apply that 00:06:58.589 --> 00:07:01.629 training. Got it. Text defines boundaries and 00:07:01.629 --> 00:07:04.310 Claude fills in the creative execution. That 00:07:04.310 --> 00:07:07.110 is the perfect way to summarize it. Knowing these 00:07:07.110 --> 00:07:09.810 pre -made templates exist in a GitHub repository 00:07:09.810 --> 00:07:13.649 is fantastic. But relying on someone else's templates 00:07:13.649 --> 00:07:15.889 isn't enough if you want to build a unique brand. 00:07:16.089 --> 00:07:18.910 You have to build a custom system for your specific 00:07:18.910 --> 00:07:21.329 workflow. Building your own system is where the 00:07:21.329 --> 00:07:23.889 real leverage is. The source outlines a very 00:07:23.889 --> 00:07:26.829 precise five -step framework to codify your own 00:07:26.829 --> 00:07:29.779 work. Step one is choosing the format. Right. 00:07:30.060 --> 00:07:31.959 You have to be strategic here. You must pick 00:07:31.959 --> 00:07:34.120 the task with the highest return. You shouldn't 00:07:34.120 --> 00:07:36.060 pick the most exciting task or the one you enjoy 00:07:36.060 --> 00:07:38.339 doing the most. You pick the bottleneck. Exactly. 00:07:38.540 --> 00:07:40.860 If you're running a YouTube channel, your biggest 00:07:40.860 --> 00:07:43.920 bottleneck might be creating HTML slide decks 00:07:43.920 --> 00:07:46.779 for your educational videos. So you codify that. 00:07:47.019 --> 00:07:49.160 Yeah. If you run a creative agency doing client 00:07:49.160 --> 00:07:52.740 work, your highest return task is probably generating 00:07:52.740 --> 00:07:55.680 pristine client proposals. You build the system 00:07:55.680 --> 00:07:58.610 around the task you repeat constantly. Once the 00:07:58.610 --> 00:08:01.449 format is chosen, step two is adding your tools 00:08:01.449 --> 00:08:03.959 and integrations. Crucial step. Claude shouldn't 00:08:03.959 --> 00:08:06.420 be forced to guess what your brand looks like. 00:08:06.519 --> 00:08:08.819 You have to feed it context from your actual 00:08:08.819 --> 00:08:11.319 business operations. You plug it into your existing 00:08:11.319 --> 00:08:14.120 ecosystem. You connect it to your GitHub repositories. 00:08:14.279 --> 00:08:16.500 You link it to your Gmail to understand your 00:08:16.500 --> 00:08:18.959 communication style. You feed it massive amounts 00:08:18.959 --> 00:08:21.439 of meeting transcripts. Whoa, imagine generating 00:08:21.439 --> 00:08:23.819 hundreds of perfectly branded slide decks just 00:08:23.819 --> 00:08:26.720 from meeting transcripts. You finish a one -hour 00:08:26.720 --> 00:08:29.500 Zoom call, and the system automatically outputs 00:08:29.500 --> 00:08:32.100 the client presentation. That is the reality 00:08:32.100 --> 00:08:34.750 of this. workflow. But to get that level of accuracy, 00:08:35.049 --> 00:08:37.190 the source highlights a critical tool for this 00:08:37.190 --> 00:08:39.389 step called FireCrawl. Let's define that quickly 00:08:39.389 --> 00:08:42.750 for the audience. FireCrawl is a tool that copies 00:08:42.750 --> 00:08:46.149 website data into text for AI. It acts as a bridge. 00:08:46.309 --> 00:08:49.549 It crawls your existing live website. It pulls 00:08:49.549 --> 00:08:52.470 down your exact logo files. It maps out your 00:08:52.470 --> 00:08:55.389 entire DOM structure. Wow. Yeah. It extracts 00:08:55.389 --> 00:08:58.210 your precise brand colors and topography scales. 00:08:58.350 --> 00:09:00.549 Yeah. It basically turns your live website into 00:09:00.549 --> 00:09:03.159 data Claude can read. Which leads us directly 00:09:03.159 --> 00:09:06.279 into step three, providing references. Tools 00:09:06.279 --> 00:09:09.080 like Firecrawl give the AI factual information. 00:09:09.960 --> 00:09:12.539 But references give the AI taste. Right, because 00:09:12.539 --> 00:09:14.600 information isn't enough to make something beautiful. 00:09:14.799 --> 00:09:17.179 You have to show Claude what your version of 00:09:17.179 --> 00:09:19.240 excellent looks like. How do you practically 00:09:19.240 --> 00:09:22.100 do that within a text -based workflow? The source 00:09:22.100 --> 00:09:24.419 mentions using mid -journey. Yeah, you can generate 00:09:24.419 --> 00:09:26.539 conceptual images in mid -journey that capture 00:09:26.539 --> 00:09:28.940 your desired aesthetic. You feed those images 00:09:28.940 --> 00:09:32.100 into Claude. Claude's vision model analyzes the 00:09:32.100 --> 00:09:34.500 flat image. It maps out the spatial relationships, 00:09:34.860 --> 00:09:37.899 extracts the implicit color hierarchy, and translates 00:09:37.899 --> 00:09:39.960 those visual cues into... to structural code 00:09:39.960 --> 00:09:42.840 rules. You can also use your own historical assets. 00:09:43.240 --> 00:09:45.860 You feed it your best performing Canva templates 00:09:45.860 --> 00:09:49.399 or past agency designs. You are giving it a visual 00:09:49.399 --> 00:09:52.139 target to aim for before it writes a single line 00:09:52.139 --> 00:09:55.500 of code. Which brings us to step four, generating 00:09:55.500 --> 00:09:58.779 the first version. The crucial mindset shift 00:09:58.779 --> 00:10:01.179 here is that you are only aiming for an 80 % 00:10:01.179 --> 00:10:04.240 draft. Because expecting perfection on the first 00:10:04.240 --> 00:10:06.759 output will just lead to frustration. Exactly. 00:10:07.399 --> 00:10:10.480 The goal of step four is purely structural. You 00:10:10.480 --> 00:10:12.700 want the layout to be logically sound. You want 00:10:12.700 --> 00:10:15.120 the typography to match your brand rules. You 00:10:15.120 --> 00:10:17.480 want the primary colors placed correctly. You're 00:10:17.480 --> 00:10:19.440 just trying to escape the terror of the blank 00:10:19.440 --> 00:10:22.740 page. A flawed 80 % draft gives you something 00:10:22.740 --> 00:10:25.700 tangible to react to. Reacting to a draft is 00:10:25.700 --> 00:10:27.960 always faster than creating from scratch. That's 00:10:27.960 --> 00:10:30.600 why step five is so important. Step five is refine 00:10:30.600 --> 00:10:33.919 and save as a skill. This step is where the actual 00:10:33.919 --> 00:10:36.480 magic of codification happens. The feedback loop. 00:10:36.600 --> 00:10:38.580 But you can't use traditional client feedback 00:10:38.580 --> 00:10:41.059 here. You can't just tell Claude, make it pop 00:10:41.059 --> 00:10:44.059 or make it feel more dynamic. No, vague feedback 00:10:44.059 --> 00:10:47.580 breaks the system completely. Your feedback must 00:10:47.580 --> 00:10:50.100 be hyper -specific. You have to speak in structural 00:10:50.100 --> 00:10:52.519 rules. Instead of saying, it looks too busy, 00:10:52.639 --> 00:10:55.500 you tell Claude, limit the layout to three hex 00:10:55.500 --> 00:10:58.190 codes. Instead of saying make it smoother, you 00:10:58.190 --> 00:11:01.350 command it to use slower motion on the CSS transitions. 00:11:01.870 --> 00:11:03.809 You're literally tweaking the machine's parameters. 00:11:04.490 --> 00:11:07.669 Reduce padding by 20 pixels. Increase the contrast 00:11:07.669 --> 00:11:10.730 ratio on the subheaders. Like stacking Lego blocks 00:11:10.730 --> 00:11:13.870 of data. Yes. And once you dial in those parameters 00:11:13.870 --> 00:11:16.450 and the output looks perfect, you save that entire 00:11:16.450 --> 00:11:19.490 prompt chain as a reusable workflow. It becomes 00:11:19.490 --> 00:11:22.389 a permanent skill in your library. Beat. I am 00:11:22.389 --> 00:11:24.409 curious about step two again, though. Why is 00:11:24.409 --> 00:11:26.389 Firecrawl necessary if you can just attach a 00:11:26.389 --> 00:11:28.990 screenshot of a website? Well, screenshots are 00:11:28.990 --> 00:11:31.289 totally flat. Firecrawl pulls the exact text 00:11:31.289 --> 00:11:34.950 codes, typography, and HTML stature. Ah, Firecrawl 00:11:34.950 --> 00:11:37.470 pulls exact code and colors, not just a flat 00:11:37.470 --> 00:11:40.129 image. Exactly. That underlying data makes the 00:11:40.129 --> 00:11:43.009 whole system robust. Sponsor. All right, let's 00:11:43.009 --> 00:11:45.710 zoom out now. We've mapped out this exact step 00:11:45.710 --> 00:11:48.190 -by -step framework for codifying taste. What 00:11:48.190 --> 00:11:50.620 does this mean for the industry at large? We 00:11:50.620 --> 00:11:53.179 are witnessing a fundamental paradigm shift. 00:11:53.559 --> 00:11:56.940 The source material refers to this as the code 00:11:56.940 --> 00:12:00.940 first design future. Code first design. That 00:12:00.940 --> 00:12:03.360 phrase carries a lot of weight. It changes the 00:12:03.360 --> 00:12:06.230 entire objective of creative work. Under this 00:12:06.230 --> 00:12:08.690 model, you aren't actually making a design. You 00:12:08.690 --> 00:12:10.629 are building the machine that makes the design. 00:12:10.889 --> 00:12:13.110 Right. You spend your energy defining the rules 00:12:13.110 --> 00:12:15.929 once, and then you let Claude execute the layout 00:12:15.929 --> 00:12:18.649 generation over and over again at near zero marginal 00:12:18.649 --> 00:12:21.450 cost. Let's break down the impact on different 00:12:21.450 --> 00:12:24.169 professions. For independent creators and startup 00:12:24.169 --> 00:12:26.409 founders, this shift is incredibly empowering. 00:12:27.009 --> 00:12:29.409 Historically, custom design has always been a 00:12:29.409 --> 00:12:32.350 massive bottleneck. It's expensive, it's slow, 00:12:32.529 --> 00:12:35.299 and it requires specialized talent. With a codified 00:12:35.299 --> 00:12:37.840 system, founders can launch new products infinitely 00:12:37.840 --> 00:12:41.039 faster. They maintain impossibly clean, consistent 00:12:41.039 --> 00:12:43.580 visuals across all their platforms. And they 00:12:43.580 --> 00:12:45.820 save thousands of dollars on routine, one -off 00:12:45.820 --> 00:12:48.259 design tasks. What about developers? This seems 00:12:48.259 --> 00:12:50.259 like a massive structural change for engineering 00:12:50.259 --> 00:12:53.200 teams. Developers no longer need to spend years 00:12:53.200 --> 00:12:56.240 mastering the intricacies of Figma to build beautiful 00:12:56.240 --> 00:12:59.879 products. But they do have to acquire a new hybrid 00:12:59.879 --> 00:13:03.360 skill. The source describes it as coding plus 00:13:03.360 --> 00:13:06.679 taste. Coding plus taste. They need to understand 00:13:06.679 --> 00:13:09.480 the underlying principles of visual hierarchy. 00:13:09.860 --> 00:13:12.679 They have to learn how to articulate design critique 00:13:12.679 --> 00:13:15.419 technically. A developer can't just look at a 00:13:15.419 --> 00:13:18.220 generated layout and say, make it look better. 00:13:18.340 --> 00:13:21.039 Right. They need the vocabulary to say, use stronger 00:13:21.039 --> 00:13:24.320 typographic hierarchy here or increase the negative 00:13:24.320 --> 00:13:26.679 space around the primary call to action. They 00:13:26.679 --> 00:13:29.179 act as the art director, guiding the creative 00:13:29.179 --> 00:13:31.679 direction, while Claude acts as the production 00:13:31.679 --> 00:13:34.580 designer, writing the actual code. And what is 00:13:34.580 --> 00:13:37.240 the impact on larger established teams? This 00:13:37.240 --> 00:13:39.220 might be the most practical application of all. 00:13:39.440 --> 00:13:41.659 Think about traditional corporate brand guidelines. 00:13:42.039 --> 00:13:44.899 Usually a company spends $50 ,000 on a rebrand. 00:13:45.059 --> 00:13:47.220 And they get a beautifully designed PDF that 00:13:47.220 --> 00:13:49.419 sits completely dead on a shared hard drive. 00:13:49.539 --> 00:13:52.000 Exactly. Nobody ever actually follows the PDF 00:13:52.000 --> 00:13:54.039 rules when they are rushing to finish a sales 00:13:54.039 --> 00:13:56.679 deck on a Friday afternoon. That is the reality 00:13:56.679 --> 00:13:59.679 of corporate design. But Clawed Code takes those 00:13:59.679 --> 00:14:03.159 dead PDF guidelines and turns them into living, 00:14:03.259 --> 00:14:06.539 usable engines. You transform a static reference 00:14:06.539 --> 00:14:09.100 document into an active system that automatically 00:14:09.100 --> 00:14:11.600 generates your daily deliverables. It builds 00:14:11.600 --> 00:14:13.700 your internal dashboards, your client proposals, 00:14:14.019 --> 00:14:16.899 your quarterly reports. It enforces strict brand 00:14:16.899 --> 00:14:20.000 consistency automatically without requiring a 00:14:20.000 --> 00:14:22.139 senior designer to review every single slide. 00:14:22.759 --> 00:14:24.740 But I do have to push back on behalf of professional 00:14:24.740 --> 00:14:28.159 designers here. If the machine handles all the 00:14:28.159 --> 00:14:30.500 execution and the guidelines are enforced automatically, 00:14:30.840 --> 00:14:33.659 does human taste actually matter anymore? Yes, 00:14:33.740 --> 00:14:36.059 the creative work is still there. You still need 00:14:36.059 --> 00:14:38.960 judgment and taste, but now taste becomes a system. 00:14:39.179 --> 00:14:42.139 So where does Figma fit into a world where Claude 00:14:42.139 --> 00:14:44.340 does all the heavy lifting? Figma is simply the 00:14:44.340 --> 00:14:46.519 final polished step. It is definitely not the 00:14:46.519 --> 00:14:48.919 starting line anymore. Exactly. Use Claude for 00:14:48.919 --> 00:14:51.620 the fast draft, Figma for the pixel perfect finish. 00:14:51.779 --> 00:14:54.240 That hybrid approach is how the best teams will 00:14:54.240 --> 00:14:57.789 operate moving forward. Two sec silence. This 00:14:57.789 --> 00:15:00.289 has been a thoroughly fascinating shift in perspective. 00:15:00.649 --> 00:15:04.309 The core truth we are uncovering is that good 00:15:04.309 --> 00:15:07.190 design doesn't have to be this ethereal, abstract 00:15:07.190 --> 00:15:10.409 concept trapped inside a creative director's 00:15:10.409 --> 00:15:12.669 head. No, it doesn't. It doesn't have to be stuck 00:15:12.669 --> 00:15:15.409 in a massive, disorganized Figma file full of 00:15:15.409 --> 00:15:17.610 disconnected layers. It is so liberating when 00:15:17.610 --> 00:15:20.250 you realize those visual rules can be extracted. 00:15:20.830 --> 00:15:22.769 They can be codified. They can be written down 00:15:22.769 --> 00:15:25.129 in plain text. They can be turned into a highly 00:15:25.129 --> 00:15:27.850 repeatable system that scales your specific visual 00:15:27.850 --> 00:15:30.850 identity. A landing page isn't just a page anymore. 00:15:31.090 --> 00:15:34.470 It is a localized output of a much larger system. 00:15:34.750 --> 00:15:37.210 That systemic thinking is the future. It leaves 00:15:37.210 --> 00:15:39.850 me with one final, slightly provocative thought 00:15:39.850 --> 00:15:42.110 to consider. Let's hear it. If our current modern 00:15:42.110 --> 00:15:45.169 visual taste can be perfectly codified into Markdown 00:15:45.169 --> 00:15:47.409 and perfectly replicated by Claude at scale, 00:15:47.610 --> 00:15:49.789 what happens to the value of current design trends? 00:15:50.429 --> 00:15:52.529 Will the most valuable design skill of tomorrow 00:15:52.529 --> 00:15:55.549 simply be having a wildly unique, undefinable 00:15:55.549 --> 00:15:57.909 perspective that an AI hasn't been trained on 00:15:57.909 --> 00:16:01.269 yet? That is a profound question to end on. It 00:16:01.269 --> 00:16:03.529 really challenges us to find the human element 00:16:03.529 --> 00:16:06.070 in our work. I encourage you to look at the very 00:16:06.070 --> 00:16:08.289 next document or presentation you create this 00:16:08.289 --> 00:16:10.970 week. Ask yourself, am I just doing a one -off 00:16:10.970 --> 00:16:13.529 chore or can I turn this into a system? Thank 00:16:13.529 --> 00:16:14.990 you for joining us on this deep dive.