DonApo

From Project Scaffold to a Voice-Intake Contract

The first implementation separated receiving audio, transcription, confirmation, and blueprint generation into explicit states.

DonApo began with a Next.js and TypeScript foundation, a Prisma data layer, environment validation, a health endpoint, and automated tests. The initial scaffold commit on 30 May 2026 established the application boundary without pretending that a complete product already existed.

The same day, the repository added a persisted voice-intake lifecycle. A VoiceInput could represent owner audio or manual text, track upload and transcription state, preserve review and confirmation timestamps, and remain linked to a project without automatically generating a blueprint.

That separation became an early design principle. Receiving audio was not transcription. Transcription was not confirmation. Confirmation was not blueprint generation. Each transition had its own state and validation, creating the basis for later human review and no-fallback behavior.

← Back to DonApo