I built a spec driven development platform for AI coding agents
Over the last year, I’ve been using AI coding tools almost every day. They’re incredible at writing code, but often were terrible at maintaining the context of the product. Every new session felt like starting over.
So I built BuildBoard, a local-first Mac app that creates a structured workflow from before the AI starts coding to finished.
The flow looks like: product idea, to requirements, to visual mocks, to implementation plan, then building with an AI coding agent. Everything stays in plain text files, so it works with Git and doesn’t require another cloud account.
I have used my own app to create a couple other ideas, prototype or improve existing projects. It allows me to track, and build on an idea. It helps me establish a shared understanding with my agents and align on the expectations. Also it’s helpful to visualize before building. I found it easier to adjust UI/UX before implementation but right in the planning phase.
I’d love feedback on whether this matches your workflow, what’s missing, and how you’re handling long-term context with tools like Claude Code or Codex.