ReWeaver AI Philosophy

Catch AI-generated UI inconsistencies before they ship.

ReWeaver AI is the reconciliation layer that keeps design and code in sync in both directions, deterministically, and with every fix on record.

What We Do

AI writes code faster than anyone can verify it matches the design.

Every AI coding tool (Cursor, Copilot, Claude Code, v0) generates an approximation of what the design actually specifies. Most of the time it’s close. Close enough to ship, and close enough that nobody catches the gap until it’s already in production.

ReWeaver reads your design system and your codebase against the same rule set, surfaces exactly where they disagree, and lets a human decide what happens next: fix it, suppress it, or defer it. Every decision is recorded in git. Nothing is fixed silently, and nothing is ignored silently either.

Drift is what happens when AI-generated code and the design it’s supposed to match quietly stop agreeing with each other. A hardcoded color that bypasses a token, a component that no longer matches its Figma source, an accessibility pattern silently dropped. Nobody decides this should happen. It just accumulates, one AI-generated change at a time, until the design system and the product are lying to each other.

Why This, Not Something Else

Plenty of tools can tell you design and code disagree.

That part of the problem is getting solved industry-wide, and we’d rather say so than pretend otherwise. What’s still unsolved almost everywhere is what happens after detection.

What’s Common Now

Detect the mismatch, on one side or both

A growing set of tools can flag that a Figma value and a code value don’t match, or that a component has gone stale. That’s real, useful, and no longer rare.

What ReWeaver Does

Fix it in both directions, including inside Figma itself

A code-side fix rewrites the hardcoded value to the correct token. A Figma-side fix mutates the design file itself. Both are deterministic, exact match or refuse, never a near-miss, and every fix is on record in git, not applied silently.

Founder Story

Jonathan Gordon

Jonathan Gordon

Jonathan Gordon

CEO / Founder

CEO / Founder

I started my career as a software developer, then became a software designer after seeing how much pain a design that didn’t account for real users could cause. What I ran into on the other side was a different version of the same problem: designers spec one thing, developers build another, and by the time it ships, it doesn’t match what was designed. I closed that gap for years through documentation and process, and it never fully closed. Drift always found a way back in.

I started my career as a software developer, then became a software designer after seeing how much pain a design that didn’t account for real users could cause. What I ran into on the other side was a different version of the same problem: designers spec one thing, developers build another, and by the time it ships, it doesn’t match what was designed. I closed that gap for years through documentation and process, and it never fully closed. Drift always found a way back in.

When AI coding tools arrived, I thought this was finally the way to solve it: a system that could understand both design intent and code output at once. Instead, most of the industry is using AI to generate faster, not to verify what gets generated. The gap between design and code isn’t closing. It’s widening, faster than any team can manually track.

When AI coding tools arrived, I thought this was finally the way to solve it: a system that could understand both design intent and code output at once. Instead, most of the industry is using AI to generate faster, not to verify what gets generated. The gap between design and code isn’t closing. It’s widening, faster than any team can manually track.

ReWeaver is the mission that came out of that: put AI to work verifying quality, not just producing output, and keep a human genuinely in control of every fix, not just nominally “in the loop.”

ReWeaver is the mission that came out of that: put AI to work verifying quality, not just producing output, and keep a human genuinely in control of every fix, not just nominally “in the loop.”

Leadership

The team

Roger Ruttimann

Roger Ruttimann

Roger Ruttimann

Engineering Leadership

Engineering Leadership

25+ years building and scaling data-driven systems, including Senior Director at Salesforce and CTO / VP Engineering roles across multiple startups. Focused on the unglamorous parts AI doesn’t solve on its own: scalability, reliability, and security.

25+ years building and scaling data-driven systems, including Senior Director at Salesforce and CTO / VP Engineering roles across multiple startups. Focused on the unglamorous parts AI doesn’t solve on its own: scalability, reliability, and security.

LinkedIn ↗

Derek Hoiem

Derek Hoiem

Derek Hoiem

Product Leadership

Product Leadership

Three decades in tech, starting on Microsoft’s first user interface group. Early research contributed to the Windows Start Button; later led UX for Visual Studio. Has since coached C-suite executives and product teams on strategy and design.

Three decades in tech, starting on Microsoft’s first user interface group. Early research contributed to the Windows Start Button; later led UX for Visual Studio. Has since coached C-suite executives and product teams on strategy and design.

LinkedIn ↗

Macall Gordon

Macall Gordon

Macall Gordon

Thought Leadership / Research

Thought Leadership / Research

Researcher and author, and self-described Chief of Getting Things Done. BS from Stanford, MA in applied psychology.

Researcher and author, and self-described Chief of Getting Things Done. BS from Stanford, MA in applied psychology.

Platform

Runs on Mac today. Windows support is available in beta, with some rough edges while it matures.

Want to see it catch a real drift?

Get on the beta waitlist. Catch a real drift in your first session.

Join the Beta