Skip to content

Build an Animated Landing Page with AI (5 Prompts)

Build an animated landing page with AI using one brand brief and 5 tested prompts: a smoke background, a 2,000-dot logo, dark mode and a 90+ speed score. No coding needed.

Mahnoor 5 min read 18-item prompt pack

Agencies quote $5,000 or more for an animated landing page like this one: smoke that follows your cursor, a logo made of 2,000 dots that scatter when you touch them, dark mode, and a 90+ speed score on mobile. You can build your own version with AI in a single sitting, without writing code yourself. The trick to a good AI landing page is not a magic prompt. It's one short brief and five prompts, one layer at a time, with a quick check after each one.

What you need

An AI chat that can write code (Claude, ChatGPT, Gemini or v0 all work), a browser, and about ten minutes to fill in your brand brief. Have your logo ready as an SVG file, plus the two or three fonts and colors you want to use. That's it: no installs, no framework, no build step. Every prompt asks for plain HTML, CSS and JavaScript, so the result runs anywhere and you own every line of it.

Why AI gives you the same purple page every time

Ask an AI to "build me a landing page" and you get a purple gradient, three boxes and the same layout as everyone else. The AI isn't bad at design. It's guessing, because you didn't tell it your brand, your colors, your fonts or what the page is for. Fix that once, in a brief, and every prompt after it builds on your decisions instead of its defaults.

Step 0: Fill in the brand brief

The brand brief is a short template: your brand name, a one-line description, your headline and buttons, your light and dark colors, your fonts, your logo and the text for each section. Fill it in once and paste it at the top of every prompt. That's what makes the page look like yours instead of everyone else's. The brief is free below.

Why not one giant prompt?

You can try the one-shot prompt below. It builds the whole page in one go, and it's a fair starting point. But when we tested it, about half the animations were missing: the smoke was a flat grey wash and the dots never formed the logo. One layer at a time works much better, because the AI can focus on one job and you can catch problems before they stack up.

Step 1: Foundation and design system

The first prompt sets up colors as CSS variables, fonts, a type scale, buttons, the sticky header, the footer and the dark mode switch. Nothing flashy yet. This is the base every later step reuses.

Check before step 2: flip the theme toggle, refresh the page, and make sure your choice is remembered with no white flash on load. Press Tab and check every link and button gets a visible outline.

Step 2: Hero, smoke, preview card and stats

Prompt 2 builds the whole top of the page: an animated smoke background made with a small WebGL shader, your headline, a product preview card and a stats strip. Move your mouse and the smoke follows; switch to dark mode and it changes color.

Check before step 3: make the card title very long on purpose. It should wrap or cut off cleanly, not break the card. If anything breaks, use the free fix-it prompt and tell the AI exactly what you see. Don't re-run the whole step.

Step 3: How it works and the dot logo

This is the step everyone comes for. Prompt 3 adds a "how it works" timeline and turns your logo into about 2,000 dots on a canvas. The dots fly in, scatter away from your cursor and spring back into shape.

The one mistake: if you get a blob instead of your logo, your SVG is too detailed. Use just the icon, one bold shape with no text, and clean it up on SVGOMG first. Then check that the dots stop moving when you stop moving, so the page doesn't burn CPU.

Step 4: Features, FAQ, sign-up and motion polish

Prompt 4 adds the feature cards with a cursor spotlight, an FAQ and the sign-up block. Then it goes through every animation on the page so it all feels like one system: the same speed and easing, nothing jumpy, smooth interactions.

Check before step 5: scroll the whole page once and make sure nothing pops in late or flickers, and that the FAQ opens with the keyboard.

Step 5: Audit, fix and go live

The last prompt turns the AI into a strict reviewer. It checks mobile layouts, color contrast, speed and SEO, lists every problem it finds and then fixes them. Run the live page through PageSpeed Insights, then deploy it for free with Netlify Drop or GitHub Pages.

Get the full prompt pack

The brand brief, the one-shot prompt and the fix-it prompt are free on this page. Sign up free to unlock the full 5-step build, the checklist for every step and links to every free tool used in the video. Built something? Share it with us.

New to building with AI? Browse all free lessons or follow a learning path from start to finish.

Prompt pack

Every prompt from this lesson

5-step build that matches the video, the full checklist for every step and 9 free tool links. Free with signup.

Brand brief (fill this in first)
BRAND BRIEF (paste this at the top of every prompt)

Brand name: [YOUR BRAND]
What it is, one line: [e.g. "Free tools that help freelancers get paid faster"]
Audience: [who it's for]
Headline (max 7 words): [e.g. "Get paid without the chasing."]
Subheadline (1–2 sentences): [...]
Primary button: [text] → [link]
Secondary button: [text] → [link]
Small line under the buttons: [e.g. "Free forever · no card"]

Colors:
- Light mode: background [#ffffff], text [#0a0a0a], accent [#...]
- Dark mode: background [#000000], text [#ededed], accent [#...]
(Leave accent empty for a pure black and white look.)

Fonts (Google Fonts or Fontshare):
- Headings: [e.g. Geist / Satoshi / Manrope]
- Body: [same or another]
- Code and labels: [e.g. Geist Mono / JetBrains Mono]

Logo: [paste your logo SVG code here, or write "use a simple bold 5-point star"]

Section content:
- 3 proof points: [e.g. "100% free", "New tool every week", "Built for freelancers"]
- 3 "how it works" steps: [title + one line each]
- 3–6 feature cards: [title + one line each]
- 4–6 FAQ: [question + answer]
- Closing call to action: [headline + one line]
One-shot landing page prompt
You are a senior front-end engineer and product designer. Build a modern, animated,
high-converting landing page from my BRAND BRIEF below.

TECH
- 3 files: index.html, style.css, script.js. Vanilla HTML/CSS/JS only, no frameworks,
  no libraries, no build step. Load fonts from Google Fonts (or the source I named).
- Give me each file complete, in its own code block. No placeholders like "...rest here".

DESIGN DIRECTION
- Minimal, high contrast, lots of whitespace, thin 1px borders, subtle grid feel.
- All colors as CSS variables on :root (background, foreground, muted text, surface,
  border, border-strong, inverse). Dark mode overrides the same variables under
  html.dark. Theme toggle in the header; remember the choice in localStorage;
  default to the visitor's system setting.
- Big, tight headline (slightly negative letter-spacing), muted body text, max width ~1150px.

SECTIONS (in this order)
1. Sticky header: logo + name, 3 nav links, theme toggle, primary button. Blur behind it.
2. Hero (2 columns on desktop, stacked on mobile):
   - Left: small pill badge "New → [something]", headline, subheadline, 2 buttons, small line.
   - Background: an animated monochrome "smoke" made with a small WebGL fragment shader
     (fractal noise), brightest behind the headline, drifting gently toward the cursor.
     Colors read from the CSS variables so it works in both themes. Cap at 30fps, pause when
     off-screen or the tab is hidden, CSS radial-gradient fallback if WebGL is unavailable.
   - Right: a "product preview" card showing what the product looks like (window-style
     top bar, a title, a highlighted sample block, a list of 3 items with a blurred lock
     effect, and a call to action line). Soft, large shadow.
3. Proof strip: 3 items in a row separated by vertical dividers (stacked on mobile).
4. How it works: left = 3 numbered steps in a vertical timeline; a vertical line draws
   from top to bottom when the section scrolls into view. Right = my logo rebuilt as
   ~2000 small dots on a <canvas>: dots fly in on first view, scatter away from the cursor
   (or a tap on mobile) and spring back into the logo shape. Sample the dot positions by
   drawing the logo SVG path on an offscreen canvas. Show the static SVG if JS fails.
5. Feature cards: grid of cards, each with a soft glow that follows the cursor along the border.
6. FAQ: accordion using <details>/<summary>, a "+" icon that rotates to "×" when open.
7. Closing CTA: a big rounded block in the inverse colors with headline, line, email input + button.
8. Footer: logo, small links, copyright.

MOTION
- Hero text fades up with a small stagger. Other sections reveal on scroll (one shared
  IntersectionObserver, fade + 12px rise, staggered children).
- Do NOT animate the main headline in (it must paint instantly for performance).
- Respect prefers-reduced-motion: no shader movement (one still frame), no particles
  (static logo), no reveals.

QUALITY BAR
- Fully responsive from 360px to 1440px, no horizontal scroll.
- Accessible: semantic landmarks, one h1, visible focus rings, text contrast WCAG AA in both
  themes, decorative canvases aria-hidden, buttons are real <a> or <button> elements.
- Fast: no images required, start the particle canvas only when it is near the viewport.

BRAND BRIEF:
[paste your filled brief here]
Fix-it prompt
Something is wrong in the page you built.
What I see: [describe exactly, e.g. "the dots don't move on mobile", "dark mode text is invisible"]
What I expected: [...]
Browser + device: [...]
Console errors (F12 → Console): [paste them, or write "none"]

Find the cause first and explain it in 2 lines. Then give me only the changed code,
with the file name and where it goes. Don't rewrite parts that already work.

15 more items in the full pack

Free. Sign in once and every prompt pack on Morow Global unlocks.

Sign in free to unlock

Already a member? Sign in

FAQ

Do I need to know how to code?

No. You paste a brief and five prompts into an AI chat like Claude or ChatGPT, then check the result in your browser after each step. Knowing a little HTML helps when you fix things, but it is not required.

Which AI tool works best for this?

Any strong chat model works: Claude, ChatGPT, Gemini or v0. In the video we used Claude because it shows a live preview. Keep all five steps in the same chat so the AI keeps your files in context.

Why does my logo turn into a blob instead of dots?

Your SVG is too detailed. Use only the icon, one bold shape with no text, and clean it up on SVGOMG before step 3. Thin lines and small text do not survive the dot sampling.

Why not use one big prompt for the whole page?

You can, and the one-shot prompt on this page does exactly that. But in our test about half the animations were missing. Building one layer at a time lets the AI focus and lets you catch bugs before they stack up.

How do I put the page online for free?

Drag your project folder into Netlify Drop, or push it to GitHub Pages. Both are free. Then run the live link through PageSpeed Insights on mobile and aim for 90+.

#landing page #prompts #claude #vibe coding #web animation