BluePainter
Design and code, one surface.
The problem
Design tools export code nobody keeps. Code-first teams lose the visual layer. BluePainter is a workflow bet rather than an LLM wrapper: AST-preserving bidirectional sync and team-configurable design policy inside existing repositories.
Key decisions
Generated code that replaces a developer’s file gets reverted.
Regenerate components from the canvas, or write back into the existing AST.
AST-preserving write-back. Canvas edits change the nodes they touch and nothing else.
Much harder to build. It is the only version developers would keep.
The idea is big enough to waste a year on.
Build the product, or build the thing that tests the product.
A validation prototype that opens straight into the canvas, with an interview guide, in-app feedback, facilitator mode and session export, and a go/no-go after eight sessions.
No revenue yet. Also no year wasted.
The work
How it was built
- 01DiscoverCalls and walk-throughs with the people who would use it
- 02CompileTranscripts, files and constraints in one requirements brief
- 03SpecifyA detailed, sourced spec from Perplexity, reviewed and locked
- 04BuildDesigned and built with the latest AI models, in the production stack
- 05TestUser testing on every preview, with real users on real tasks
- 06IterateFindings turned into the next round of changes, worst problems first
- 07ShipTests, security review and release, then back to testing and iterating↻ Back to Test
Where AI was used
The sync is deterministic by design. AI assists authoring and the distribution site’s guides; the position is explicitly “not an LLM wrapper”.
What’s next
Go/no-go after the validation sessions, then the VS Code extension to the Marketplace and signed desktop builds.