The 3 Levels Of AI Design... Only A Few Reach Level 2

English

Transcript

00:00:00If you've been watching this channel for some time, you probably know that we've covered a lot
00:00:03of design workflows and tools. We've been testing all of them for months and we finally figured out
00:00:08why the same model can give you something that looks completely custom or something that immediately
00:00:13screams AI generated. It comes down to three levels. Level one is designing a single page and
00:00:19there's one thing most people skip that's the entire reason their output looks generic. Level
00:00:23two is where you stop designing pages and start designing systems and the workflow here is
00:00:28completely different. And level three is how we test designs against each other to find the version
00:00:34that actually works, which is the part we use on every real project now. So level one is about
00:00:39creating a good design for a single page. This is the level that most people teach because it's the
00:00:44foundation of every good design. We talked in our previous video about how Opus 4.7's design
00:00:50capability has gotten so much better and a lot of the AI slop we used to see is gone. Earlier when we
00:00:55used to give it a simple prompt like creating a landing page, it would just straight up take the
00:00:59purple and white theme and build everything around it. That specific pattern has gotten better. But
00:01:04just like any other AI model, this one also converges to safe patterns. And from all our testing and
00:01:09experimenting with it, we have found that it defaults to one particular style every time. So now whenever
00:01:15we see that style, it's a dead giveaway that the site came from Opus 4.7 and it's only a matter of time
00:01:21before it becomes the next AI slop. So we need other ways to make this website look better.
00:01:25Now this level mostly comes down to prompt engineering and how we specify the app because
00:01:30if you structure your prompt properly, you can just one shot the app entirely. The prompt should start
00:01:35with the intent of the website you are aiming to build, then mention the non-negotiables, like the
00:01:39exact things you want in the app and how you want the UI elements to look. After that, you specify the
00:01:44color system. Now here we use OKLCH, which is basically a measure of lightness, chroma and hue.
00:01:49Using OKLCH instead of the usual RGB or HSL is better because it represents colors the way the
00:01:55human eye actually perceives them, so it handles lightness and balance better. It also creates
00:01:59smoother gradients, unlike hex codes which can produce ones that look uneven. Now once you have
00:02:04set the color scheme, you also need to mention the contrast flows. Contrast is a very important
00:02:08factor of UI design because it actually creates a hierarchy that guides your eyes toward the things that
00:02:13matter. Without explicit contrast, the model treats every element as equally important,
00:02:18which makes it hard to form the visual hierarchy. And to make sure the website doesn't look like
00:02:22AI slop, you also have to control the typography from the prompt, so you define which fonts are banned
00:02:27because of AI slop and which ones to use in different areas of the design. Fonts like Inter and Geist
00:02:33have become AI slop giveaways because every agent reaches for them by default, so calling them out
00:02:38explicitly forces the model to look elsewhere. Then you define the layout and rhythm of the website,
00:02:42but first you need to know about symmetry and asymmetry. Symmetric layouts have components
00:02:47evenly placed on the grid with a balanced look, which is more suited toward professional and
00:02:51straight designs. But for a more artistic look, go for asymmetry because that gives you more room
00:02:56to experiment. It is especially good when you need to use negative space since that lets the design
00:03:01breathe. The kind of product you are building decides which one fits better. Then define all the sections
00:03:06that you want, the materials you'll use, and how the website should behave responsively. And the most important
00:03:11part is mentioning the anti-patterns. These are the hallmarks of AI slop like the simple centered CTA,
00:03:17Lucid icons, and gradients with glassmorphic design. So once you give this prompt to Claude Code or
00:03:22whichever agent you are using, it will analyze your app and go through the implementation details. Then it will
00:03:27build the app just like the prompt described, with asymmetry because of the artistic goal and proper use of negative space.
00:03:34So level 2 is about keeping the same design across every page of the site because most agent generated apps
00:03:40fall apart the moment you leave the landing page. Often when generating whole apps with agents, you might
00:03:45have encountered exactly this. The landing page is mostly pretty good, but when you go to the other pages,
00:03:50they don't follow the UI style as coherently as they should. The dashboard ends up with different
00:03:55button styles, different spacing, different typography, almost like the agent forgot it was building the
00:04:00same app. The other pages end up looking like they're not even a part of the same site and it gives away
00:04:05that the site was generated by an agent. Sometimes the design does hold on the auth pages, but then on
00:04:10the dashboard, the style breaks completely. So for that, you need to create two of the most important
00:04:15files, Claude.md and Design.md. These two files are what keep the design consistent across the whole site.
00:04:21In Claude.md, as we've talked about many times, you only put your project information, not the design.
00:04:27This is because the file stays loaded in the session all the time and design content there will just
00:04:32distract the agent when it's working on something else, but it's still the key file because it keeps
00:04:37the context of the project which informs good design. For the design itself, we need a separate file
00:04:42which mentions everything for the visual system, the layout, the colors, the typography and all the
00:04:47details we covered in level 1. The design.md should be the kind of file that any agent could pick up and
00:04:52immediately understand what the visual system is. And just like in the previous level, you need to
00:04:57define the color system in OKLCH here too. To create these two files, we gave Claude code a detailed
00:05:03prompt covering what each file needs and it generated both files for us. The Claude.md is short, just
00:05:09containing the project's details. The design.md is longer with each and every detail, including color codes,
00:05:15typography choices, and everything else. But that isn't the end of the design.md. We need to keep
00:05:20refining it over time. So we put a line at the start, telling the agent to add any new design value
00:05:25it finds to this file. That way, every session starts from a more refined version of the design
00:05:30system than the one before it. But just letting Claude create the design.md isn't enough,
00:05:35because what it generates doesn't follow best practices properly. Google has open-sourced their
00:05:40template for the design.md file. The template also contains commands to cross-verify your design.md
00:05:46against it and flag any errors. So you can just prompt your agent to iterate using those commands
00:05:51to perfect the design.md. And this still isn't the end of level 2. To generate good enough designs at
00:05:56this level, you also need to audit them against existing design principles. For that, there are
00:06:00many open-source skills that do exactly this. You can use any of them, but we use VersaLab's skill,
00:06:06because instead of hard-coding all the principles inside the skill, it points to an external source
00:06:11that they're actively maintaining. So the principles stay up to date with current best practices instead
00:06:16of being frozen at whatever was state-of-the-art when the skill was first written. You install this skill
00:06:21in the project, run it, and your design comes out in way better shape than it was before. But before we
00:06:25move forwards, let's have a word by our sponsor. So I recently started using ZillysCloud and let me tell
00:06:30you why. Most RAG apps work fine with a handful of docs, but the moment you throw in real data,
00:06:35they start falling apart because the setup just wasn't designed to handle that kind of load.
00:06:40Milvus is the most starred open-source vector database on GitHub with over 44,000 stars,
00:06:46and it's built to handle that kind of load, but self-hosting means managing infrastructure yourself.
00:06:51So that's where ZillysCloud comes in, the fully managed version with the same API that's up to
00:06:5610 times faster, and you can set it up in minutes without changing a single line of code.
00:07:00So we ran a semantic search query on ZillysCloud, and the results are actually relevant because it
00:07:05understands meaning not just keywords, and the response time is almost instant even with a large
00:07:11dataset. We also ran a recommendation query given one article. It found the five most similar ones
00:07:16across the entire dataset ranked by similarity in under a second. And the dashboard tracks your cluster
00:07:21performance, storage usage, and data metrics including collection and entity counts in real time. No credit card
00:07:27needed, just click the link in the pinned comment and try ZillysCloud for free.
00:07:31So level three is about testing the design programmatically, the same way engineers verify
00:07:36code with TDD. Now we know you can't visually write tests the way you do with code. With code,
00:07:41there are clear inputs and outputs for everything. Design doesn't have that because it's more subjective
00:07:46and can't be quantified like code can, but just because it's subjective doesn't mean we can't write
00:07:51tests for it. The reason TDD works for code is that the test pins down what the behavior should be,
00:07:57and the implementation has to satisfy that pin. The same idea applies to design just with different
00:08:01kinds of pins. In the app we were building, the first step was the same as before, to create the
00:08:05clod.md and design.md files before even thinking about implementation. Now tests should always be
00:08:12written before the code, that way the implementation can actually be tested against them. If we write tests
00:08:17after the implementation, the agent slacks off. It just writes test cases that optimize toward the
00:08:22existing code, because that code is already in its context. Writing the test first forces the
00:08:27implementation to fit the test instead of the test fitting the implementation. So we use the design
00:08:32files as the source of truth for the tests because these files contain all the anti-patterns that we can
00:08:37programmatically verify against. Every anti-pattern in design.md becomes a test case. Every color rule,
00:08:44every spacing constraint, every typography choice gets a programmatic check. We gave Claude Code a
00:08:49detailed prompt to write the test cases specifying every section it should focus on. Also if you are
00:08:54enjoying our content, consider pressing the high button because it helps us create more content like
00:08:59this and reach out to more people. With your prompt, it will write all the test cases for the design
00:09:04of the app. It writes multiple types of tests. There are the static tests, which directly check for the
00:09:09anti-patterns we mentioned in the prompt. Then there's the visual testing, which basically uses
00:09:14play right underneath and runs regression testing to make the site incrementally better. It will also
00:09:19write test cases for other components and helper functions like scan and report. Now these tests
00:09:24check for the static anti-patterns, but design testing needs something else. For that, there is
00:09:28another tool called Visly Test, which is basically a CLI that conducts TDD for UI. The way it works is
00:09:34that it runs local TDD, where you can check the design as the code changes. So you can monitor the
00:09:39diffs yourself instead of relying on the agent's self-monitoring. You also get a better diff with
00:09:44metadata and other details, which makes review faster. Without that metadata, you are just comparing two
00:09:49screenshots side-by-side and hoping you spot the difference. With it, Visly tells you exactly which
00:09:54pixels changed and by how much. To use it, first install the CLI by running the install command from the
00:10:00docs. Once it's set up and initialized, it's ready to go. Now just open Claude code and tell it to use
00:10:05TDD and implement whichever part of the UI you want using the Visly CLI as the testing medium. When
00:10:10you run the Visly TDD command, a local server starts and monitors the screenshot changes. To send the
00:10:16screenshots, Claude basically writes separate tests with the name Visly. These tests use playwright
00:10:21screenshotting mechanisms to push the images to the viewer on the server. From there, you can approve or deny the
00:10:27design and view diffs comparing it to the previous version. Each rejected diff becomes feedback the
00:10:32agent uses to adjust the next pass. Over a few iterations, the design converges to what you actually
00:10:37want instead of what the agent thinks you want. Now the prompts used here can be found in AI Labs Pro
00:10:43for this video and for all our previous videos from where you can download and use it for your own
00:10:47projects. If you've found value in what we do and want to support the channel, this is the best way to do it. The
00:10:52links in the description. That brings us to the end of this video. If you'd like to support the channel
00:10:57and help us keep making videos like this, you can do so by using the super thanks button below. As
00:11:02always, thank you for watching and I'll see you in the next one.

Description

Try Zilliz Cloud claim $100 in credits with a business email: https://tinyurl.com/zilliz-AI-Labs Milvus on GitHub (44k+ stars): https://github.com/milvus-io/milvus AI design from Claude is everywhere, but our AI design workflow shows why most outputs still scream AI-generated. This AI design process covers three levels, from a single page to a full AI design system, plus how we test AI web design and design with AI on every project. Community with All Resources 📦: http://ailabspro.io Video Code: V61 We're building The Roundup — a daily newsletter covering the AI stories that actually matter. Join now at: https://www.theroundup.so/ Vizzly: https://github.com/vizzly-testing/cli Web Design Guidelines Skill: https://github.com/vercel-labs/agent-skills/tree/main/skills/web-design-guidelines We break the full pipeline into three levels that take you from one-shot landing pages to programmatically tested interfaces. If you have been frustrated with claude ai website design that looks fine on the surface and falls apart the moment you leave the landing page, this is the system we now use on every real build. Level one is the foundation of ai web design at the single-page level. We walk through why Opus 4.7's design has improved a lot but still converges to one recognizable default style, and exactly how to structure your prompt to escape it. That means specifying intent and non-negotiables, defining your color system in OKLCH instead of RGB or HSL for better perceptual lightness and smoother gradients, controlling contrast flows so the model builds real visual hierarchy, banning AI-slop fonts like Inter and Geist, choosing between symmetry and asymmetry based on whether you are shipping a B2B dashboard or a creative portfolio, and explicitly listing anti-patterns like the centered CTA with three feature cards, Lucide icons, and glassmorphism with frosted backgrounds. This is the part of claude ai design that most tutorials skip, and it is the entire reason most outputs read as vibe code instead of intentional work. Level two is where website design ai breaks down for almost everyone. The landing page looks fine, but the dashboard, auth pages, and internal screens drift into a completely different visual language. We fix this with two files: a lean claude.md that only holds project context, and a separate design.md that carries the full visual system, color tokens in OKLCH, typography stack, layout rhythm, and anti-patterns. We use Google's open-sourced design.md template along with their cross-verification commands to refine the file over time, and we run Vercel Lab's open-source design audit skill against the output because it points to an externally maintained source of design principles instead of hardcoding rules that go stale. This is what turns a single ai design tool prompt into a proper ai system design that actually holds across a real product. Level three is where claude ai web design becomes testable. We treat design the way engineers treat code, writing tests before implementation so the agent has something to satisfy instead of optimizing toward code that is already in its context. Every anti-pattern in design.md becomes a programmatic check. We cover static tests that catch the obvious slop, visual regression that uses Playwright underneath, and Vizzly Test, a CLI that runs local TDD for UI with proper diffs and metadata so you can see exactly which pixels changed instead of squinting at two screenshots side by side. Each rejected diff becomes feedback the agent uses on the next pass, and the design converges to what you actually want rather than what the agent thinks you want. This is the full workflow we use on every real project now, and it works whether you are doing web design with ai inside Claude Code, Cursor, or any other agent. Whether you came from Figma mockups, Google Stitch outputs, or an ai website builder, the same principles apply the moment you move into ai ui design at production scale. The prompts and the claude.md and design.md files we use in this video are available inside AI Labs Pro, linked below, along with every prompt and template we build for the channel. If you want the full claude design tutorial pipeline that turns design ai into something you can actually ship, that is where it lives. 00:00 Intro 00:37 Level 1 03:34 Level 2 06:25 Sponsor 07:31 Level 3 Hashtags #ai #claudecode #claudedesign #figma #cursor #claudedesigntutorial #googlestitch #aiwebsitebuilder #vibecode #aidesign

Community Posts

No posts yet. Be the first to write about this video!

Write about this video