Project 430: a local-first personal assistant
A work-in-progress Windows tray assistant for notes, reminders, Home Assistant controls, automations, and maybe AI if the funding works out.
Project 430 is my new work-in-progress personal assistant app.
The idea is simple: a small Windows tray app that keeps the boring daily stuff close instead of spreading it across five tabs, three mobile apps, and whatever cloud dashboard is currently asking for attention.
What works now
The current version is an Electron + React + TypeScript desktop app with a local SQLite database.
Right now it handles:
- Local notes with tags and pinned ordering
- Reminders with desktop notifications
- Daily recurring reminders
- Home Assistant setup, entity sync, and light/switch toggles
- Time-based automation rules
- Execution logs for automation runs
- Windows installer packaging with smoke checks and versioned release artifacts
It is not finished. The core loop exists, but I still consider it an MVP: the UX needs more polish, the automation flow needs more real-world use, and the app needs time in my own daily routine before I call it stable.
Why I built it
I wanted something smaller and more personal than a full productivity suite.
Not a chatbot window pretending to be a life manager. Not a giant SaaS board. Just a local desktop assistant that can remember things, remind me at the right time, and eventually help control parts of my setup.
The Windows tray matters because that is where this kind of tool should live: always nearby, but not in the way.
The AI part
I do want to add AI to Project 430.
The useful version would not be “AI for the sake of AI.” It would be things like:
- Natural-language commands for notes, reminders, and automations
- Smarter reminder creation from messy input
- Summaries or cleanup for old notes
- Suggestions for useful automations based on repeated actions
- A more assistant-like command flow without losing local control
The catch is cost. Good AI features need model/API budget, testing time, and careful design so the app does not become slow, expensive, or creepy. So the plan is: build the local-first base properly, then add AI if enough funding is found to support it.
Current status
Project 430 is now listed on the projects page as in development:
I will keep shaping it as I use it. If the local assistant part becomes solid enough, the AI layer is the next big step.
