Claude Design in 16 Minutes: 5 Real Projects You Can Build Right Now

Claude Design is Anthropic’s new AI-native design tool that turns almost any creative output into editable code. Instead of painstakingly crafting every slide, screen, or animation by hand, you describe what you want and let the AI generate a working artifact that you can then tweak.

In this guide, we’ll walk through five concrete things you can build with Claude Design in just a few minutes: animated videos, pitch decks, landing pages, mobile apps, and even a full design system. Along the way, you’ll see where it shines, where it’s still rough, and how it fits into Anthropic’s bigger push to cover the entire “knowledge worker” stack.

Claude Design: An AI-Native Design Stack

Anthropic is clearly building more than a single tool. Claude Design sits alongside Claude Code for development and Claude Co-work for documents, with Claude itself able to manage tasks like email and calendar. The common idea: code is becoming the universal medium for knowledge work.

In Claude Design, videos, slides, websites, apps, and animations are all just HTML, CSS, and a bit of JavaScript under the hood. Because the output is code, you can both generate and refine it with AI, then step in to manually polish the final 10%.

This puts Anthropic in direct competition with tools like Figma, Google Workspace, Microsoft Office, and AI-native products such as Gamma or Replit’s design and coding experiences. Traditional tools assume humans craft everything manually; Claude Design assumes AI does the heavy lifting and humans refine.

Use Case 1: Animated Product Videos from Plain Text

Claude Design can turn a block of text into a fully animated promo-style video in a few minutes. For example, you can paste the text from a product announcement and simply ask it to “make an animated video” based on that content.

Once you submit your request, Claude Design asks a series of clarifying questions instead of guessing blindly. It will ask about:

• Video length (e.g., 30 seconds)
• Tone (e.g., playful and warm)
• What to emphasize (e.g., hero feature vs. overview – or let it decide)
• Visual direction (choose from preset styles)
• Where it will be used (e.g., social media)
• Motion style (e.g., subtle typographic transitions)
• Whether to show product UI mockups
• How the video should end

After that, it generates a fully animated sequence: value propositions, animated illustrations, and a clear call to action. Because the video is code, you can:

• Open a “tweaks” panel to change the color palette or layout
• Edit text and elements directly in the artifact
• Adjust motion and transitions

The main limitation right now is export: you can’t yet download a standard video file, so you may need to screen record playback. But the speed is impressive—work that might take a professional editor hours can appear in minutes.

Use Case 2: Smarter Slide Decks by Starting with Video

Claude Design can also generate slide decks, but how you prompt it matters a lot.

One approach is to first refine your slide copy in a text-based Claude chat: outline the deck, iterate on each slide’s text, and end up with a clean list of slide titles and bullet points. You can then paste that copy into Claude Design and ask it to “make a slide deck based on this copy.”

This works, but the result can feel static and a bit generic.

A Better Workflow: Video First, Deck Second

A more powerful approach is to ask Claude Design to create an animated video from your slide copy first. The video version tends to be more dynamic, with better pacing, motion, and visual storytelling.

Once you like the video, you can:

1. Duplicate the project.
2. Prompt Claude Design to “convert this video into an animated slide deck I can manually toggle through.”

The tool then restructures the animated sequence into a click-through deck, preserving much of the motion and visual interest from the video.

From there, you can manually polish details:

• Enter edit mode to remove or adjust unwanted text or labels.
• Use the tweaks panel to change color palettes (e.g., from warm to ink or clay).
• Prompt it to reorder sections (e.g., swap the order of topics) or adjust layouts.

This pattern—AI generates the first 90%, humans refine the last 10%—is quickly becoming the baseline expectation for modern presentation tools. If you’re interested in broader AI workflows for presentations and solo builders, you might also like this breakdown of AI tools that can actually make you money.

Use Case 3: Landing Pages, Interactive Graphics, and Playful Experiments

Claude Design isn’t limited to static marketing pages. It can recreate existing layouts, remix them, and even build interactive visual experiments.

Recreating and Remixing a Marketing Page

You can ask Claude Design to “recreate the Claude Co-work marketing page” with nothing more than that prompt. Even without pasting a URL, it can often find and approximate the page structure on its own.

The result won’t be pixel-perfect—videos might be missing, and some sections will be off—but it can get you roughly 80% of the way there: hero sections, feature rows, illustrations, and calls to action.

From there, you can get creative. For example, you might prompt it to turn the page into a tongue-in-cheek version about Anthropic “taking over all knowledge work.” Claude Design can then rewrite the copy in the same visual style, producing lines like:

• “Anthropic has acquired the concept of work. Please remain seated.”
• “We’ll take it from here.”
• “Locate the nearest hammock.”

It’s a playful way to explore brand voice, satire, or alternative campaigns while keeping the same design language.

Interactive Visuals: Globes and Particle Effects

Because everything is code, you can also ask for interactive data-style graphics. For example, you might prompt:

• “Design an interactive dark theme graphic showing how culture flows between cities, with a rotating globe.”

Claude Design can then generate a 3D-style spinning globe with arcs between cities and controls to toggle visual elements. The first version may have rough geography or odd shapes, but you can refine it with follow-up prompts.

Another fun example is a large editable text box where certain words trigger particle effects. A prompt like:

• “Create a large editable text box. For words like fire, smoke, metal, wind, electric, ice, and stars, render visual and particle effects that match the word.”

produces a playful canvas where typing “I’m on fire” triggers flames, “seeing a lot of smoke” adds smoke effects, “hope for some wind” animates gusts, and “electric” lights up with lightning. It’s the kind of thing you can experiment with yourself or even build as a creative toy.

If you’re interested in using AI to build more interactive systems and tools, check out this practical guide to building your own local AI agents in under half an hour.

Use Case 4: Mobile App UI Prototypes

Claude Design can also generate mobile app screens, including basic interactivity. For example, you can ask it to “create a beautiful mobile fitness app,” then answer its clarifying questions:

• Primary use: strength training
• Target user: busy professional, not a serious athlete
• Visual style: energetic
• Screens: home screen plus an in-progress workout
• Variations: single or multiple
• Interactivity: let it decide, or specify clickable navigation

After a few minutes, you’ll get a tappable prototype. A typical output might include:

• A warm, welcoming home screen with your next workout (e.g., “Push day”) and stats.
• A workout-in-progress screen with exercises like bench press, adjustable weight controls, and a button to log sets.

Claude Design even appears to “playtest” its own work: it generates screenshots, checks for layout issues like overlapping text, and then fixes them automatically where possible.

The designs aren’t flawless—there can still be spacing issues or unclear UI labels—but for a first pass, it’s remarkably fast. You get a concrete starting point to refine instead of a blank screen.

Use Case 5: Auto-Generated Design Systems

Claude Design can attempt something much more ambitious: generating a design system from an existing UI kit.

For example, you can upload a Figma UI kit—such as one using an “Apple Liquid Glass” aesthetic—and prompt Claude Design to “create the Apple Liquid Glass design system.” The tool will then work through the kit and propose system-level definitions for:

• Typography
• Color tokens
• Spacing scales
• Iconography
• Motion and easing curves
• Radii and corner treatments
• Shadows and elevation
• Documentation (e.g., a README)

This process is heavy on tokens and can consume a lot of context, taking around 10 minutes and using most of your available tokens for that session. It’s best suited if you already have a UI kit and genuinely need a systematized version.

Claude Design will then step through each section and ask you to confirm whether it “looks good.” If you’re not a designer, some of these decisions may be hard to evaluate, but you can still accept, reject, or adjust as needed.

Once the system is in place, you can attach it to new projects. For example, you can regenerate the earlier fitness app using the new Liquid Glass design system. The result will adopt the glassmorphic look—blurred panels, glowing highlights, layered translucency—but may also surface layout issues if the underlying system isn’t fully tuned.

In practice, the auto-generated design system version of the app looked worse than the simpler, default version. That’s a good reminder: design systems are powerful, but they still need thoughtful human oversight.

What Claude Design Means for Knowledge Work

Across all these examples, a few themes stand out:

1. Anthropic is targeting the full knowledge stack. With Claude Code for development, Claude Design for visual work, and Co-work for documents, Anthropic is edging into territory traditionally owned by Figma, Google Workspace, Microsoft Office, and a wave of AI startups. It’s not yet clear whether Anthropic wants to be a platform for others or a direct competitor to many of them.

2. The new default is “AI does 90%, humans do 10%.” Whether you’re making slides, landing pages, videos, or app UIs, the pattern is the same: describe what you want, let AI generate a strong first draft, then step in to tweak layouts, copy, and details. Tools that don’t support this workflow will start to feel slow and outdated.

3. It’s an incredible time to be a solo builder. AI is now good enough to act as an entry-level designer and engineer. If you have product instincts or ideas, you can move from concept to working artifact in an afternoon, even without deep design or coding skills.

Claude Design is still early and buggy in places, and the integration story between Design, Co-work, and Code isn’t fully clear yet. But it’s already a surprisingly capable playground for building videos, decks, sites, apps, and systems—fast.

If you’re curious about where tools like this are heading and how they intersect with other cutting-edge models, you may also enjoy our weekly coverage in pieces like “AI Weekly: Claude Mythos Shockwave, New Open-Source Coding Beast, and the Best Video Model You Can Actually Use.”