ReWeaver AI

ReWeaver AI develops tools to detect and reconcile differences between design systems and AI-generated code. Its About page describes checks against shared rules, with people choosing whether to correct, suppress, or postpone findings. The company says it supports deterministic changes in both code and Figma, recording decisions in git. These capabilities are first-party descriptions rather than independently verified results.

The company’s GitHub overview extends this approach to software governance, describing rule-based and syntax-tree-based detection alongside human review of proposed changes. It also offers a public AI Coding Survival Kit with prompts, rules, and skills for coding assistants. The overview describes the product as early alpha and says the underlying engine is not public, so the available toolkit should be distinguished from the full product.

At AI Engineer World’s Fair 2026, Jonathan Gordon presented “The Design-Code Roundtrip That Isn't,” demonstrating ReWeaver with generated code and a Figma canvas. The recording describes checks for design quality, performance, tokens, and accessibility. Gordon reported that repeated model-only iterations lost fidelity while deterministic guardrails helped preserve it, with human judgment still necessary. The demonstration frames accumulated design-code drift as a continuing maintenance problem.