Sigma AI secures $42M to launch CodeCanvas, a multimodal design-to-code platform

AI · 5 min read

CodeCanvas accepts input across modalities—hand sketches, layered Figma files, and natural-language descriptions—and produces clean, componentized code with tests and storybook entries. Sigma AI emphasized maintainable output, avoiding brittle, one-off code by mapping to design-system components when available.

The Series B, led by Horizon Capital, will expand engineering teams and integrations for enterprise workflows. CodeCanvas includes a verification step where designers can validate generated components in an interactive preview before committing to a repository.

Early clients appreciated the time savings in scaffold creation while noting the need for organizational governance to decide when auto-generated components become first-class. Sigma plans to add audit trails for generated code and contracts that lock behavioral expectations for regression testing.