How a feature gets built, end to end
Two independent starting points feed into one pipeline, converging on Generate Workflows and coming out the other end as a working, testable feature. AI is involved at every stage, not just the generation step.
Style GuideLight/Dark design tokens for the project.Project OverviewA doc template holding your tech stack per layer.
MockupsAI-generated screens, styled with your tokens.Project LayersAI proposes real layer names, scaffolds, and setup steps.
Generate WorkflowsTurns screens + docs into ordered, runnable layers.Workflow RunnerExecutes each layer, building and fixing until it compiles.
Testable FeatureA working piece of your app, ready to test.
Two kinds of “layer” — a note on terminology
The word “layer” is used for two different things in the app: a Project Layer (an architectural sub-project with a folder path, e.g. “Backend API”) and a Workflow Layer (one step inside a Workflow, e.g. “Data Model”). A workflow layer can optionally be linked to a Project Layer so it knows which folder to run in.