Can I vibecode Bluedot?

Build a deliberately narrow personal substitute for Bluedot, not a full clone.
Use exactly this stack: Python 3.12 + FastAPI + whisper.cpp + SQLite to wire it all together; do not offer alternative stacks or languages.
The core loop is: Build a private meeting recorder pipeline that accepts an audio file, transcribes it, creates structured notes, and exports Markdown. Do not aim to be a general platform; focus on deep, reliable implementation of this specific core loop.
Ensure the first run works locally with a single documented command (e.g., `docker compose up --build`).
Store all user data locally by default, avoiding any external SaaS dependencies for core functionality, and make data export straightforward via a dedicated API endpoint or script in standard formats like CSV or JSON.
Put all secrets, API keys, and database credentials in a `.env` file, ship a `.env.example`, and implement pre-commit hooks or CI checks to ensure credentials are never accidentally committed to version control.
Implement the smallest polished interface that completes the core loop end to end. Use realistic sample data that is clearly labelled and easy to delete.
Store every transcription job, generated notes, and metadata in the local SQLite database for auditability and durability.
Design all side effects to be idempotent and resilient. Support retrying transcription or re-generating notes for an audio file without issues.
Build in rate limiting for uploads, execution cancellation for long transcriptions, and a mechanism for secret rotation.
Ship a complete Docker Compose setup with health checks, comprehensive test suites, database backup scripts, and clear documentation.
Design a minimal but polished web UI with clear states for empty, loading, validation, success, and recoverable errors, ensuring accessible keyboard navigation, focus states, and sensible contrast.
Implement strict input validation using modern libraries, enforce safe data handling, and gracefully handle external API outages without crashing the application. Never log secrets or private file contents unnecessarily.
Write focused unit tests for the data model and the most important workflow, along with at least one robust end-to-end smoke test that proves the core loop works.
Create an exhaustive README detailing the setup process, system architecture, permission models, data storage locations, backup/restore steps, and limitations.
Strictly avoid adding multi-tenant user accounts, billing systems, telemetry, tracking analytics, ads, or third-party accounts.
Deliberately exclude these paid-product advantages: cross-call team analytics; meeting-bot auto-join; live multi-speaker accuracy.
Do not fake integrations, network effects, proprietary data, model quality, compliance, or security claims. Where an external API is optional, keep the app useful without it and explain the degraded mode.
Finish by running the test suite, building the application, and outputting the exact commands used to verify the system. Fix errors rather than merely describing them.