Picture-based tools

  1. 1Design a picture
  2. 2Write the spec
  3. 3Rebuild it in code
  4. 4Review the differences
  5. 5Fix what drifted

And again, every time the design changes.

Plastic

=

Design it, and it’s built.

The canvas renders real HTML and CSS in a browser engine, so there’s no picture to rebuild and nothing to drift.

From first frame to shipped page

Five steps, one file.

  1. Step 1

    Start from anywhere

    Drop a frame from a preset, bring in a Figma file, or have an agent recreate a screen from your codebase with its own elements, fonts and icons.

    • Frame presets for phone, desktop, slides and social
    • .fig files arrive as HTML and CSS
    • Screens rebuilt from your code, faithfully
  2. Step 2

    Lay it out for real

    Wrap a selection in a stack and Plastic infers the direction and gap. It’s real flex and grid, so it behaves on the canvas the way it will in the browser.

    • Fit, Fixed, Fill and Relative sizing
    • Drag padding and gaps where you see them
    • Snapping, guides and measuring
  3. Step 3

    Style it once

    Tune type, colour and effects in the inspector, backed by your tokens. Add hover and focus states and breakpoints on the same element, not in copies of it.

    • Tokens as CSS variables, shown by value
    • Hover, focus, active and breakpoints
    • Components and variants that stay in sync
  4. Step 4

    Get help when you want it

    Ask Claude Code, Codex, Cursor or Grok to change the design beside the canvas. Hand them a layer, a screenshot or your comments, and watch it happen.

    • Every agent edit can be undone
    • Frames lock while an agent works
    • Agents see the canvas as you do
  5. Step 5

    Ship the same file

    Open the code panel: it’s already HTML and CSS. Send a frame link, copy it as React or Tailwind, export it, or commit the folder.

    • Frame links with full styling context
    • PNG, JPG, WebP, SVG, PDF and HTML
    • Plain files, ready for Git

Under the hood

A Plastic file is a folder you can read.

One HTML file per page, a stylesheet and your tokens. Open it in any editor, commit it to Git, or keep working on it with other tools. It saves as you go and syncs to every open editor.

  • Semantic elements: forms, inputs, buttons, tables and headings
  • Your stylesheet and tokens, with the cascade intact
  • Agents read and edit it over MCP

Same craft, without the translation.

Everything you expect from a design tool, built on the platform your work ends up on.

Picture-based toolsPlastic
What you saveA proprietary file of vector layersSemantic HTML, CSS and tokens
LayoutApproximations of flexboxReal flex and grid, rendered by the browser
Hand-offSpecs, redlines and inspect modeThe source itself, plus frame links
Responsive and statesDuplicated frames per breakpointMedia queries and :hover/:focus on one element
AI agentsRead screenshots and guess the codeRead and edit the design directly
OwnershipLocked inside an appPlain files in folders, ready for Git

One file for everyone who touches the interface.

  • Designers

    Frames, layers, a pen tool and an inspector, and the confidence that what you draw is exactly what ships.

  • Design engineers

    Move between canvas and code without a translation layer. A change on one side shows up on the other.

  • Engineers

    Start from semantic HTML, a real stylesheet and tokens instead of a screenshot. Diff it, review it, commit it.

  • Teams with agents

    Give agents the actual design source, and watch their edits land on the canvas while they work.

Skip the rebuild.

Plastic is invite-only for now. Leave your email and we’ll send your invite when it’s your turn.

You’re on the list.

We send invites in small waves. When yours is ready, it arrives at the address you gave us with a sign-up link.