Zuki — My First Dev Project: A Personal AI Operating System
Hey everyone 👋 I wanted to share my first real development project — Zuki, a personal AI operating system I've been building in Python over the past few weeks. The idea started simple: I wanted an AI assistant that actually knows me, runs locally, and can help me with real business work — not just answer questions in a chat window. It grew into something bigger than I expected. What it does right now: - Multi-provider LLM routing (Gemini, Claude, GPT) with automatic fallback - Voice in/out via Whisper STT and pyttsx3/Piper TTS - Business analysis workspace — on-site Gastro analysis with PDF reports, competitor mapping, and a structured client interview flow - Tenant isolation — separate data contexts for personal use and client work (DSGVO-aware) - Cloud memory via Vercel KV with offline outbox for resilience - Cyberpunk terminal UI with a React/WebSocket web UI planned next The architecture is built around 5 workspaces (Core, Broker, Business Services, OS Layer, Cloud Memory), each with clear boundaries and routing — heavily inspired by the workspace patterns I learned from this community. What's next: A full web UI (React + Tailwind + WebSocket bridge), a live news/market broker workspace, and eventually a Linux migration to run it on dedicated hardware with a Live2D avatar panel. This is my first project and I'm still learning — the codebase went through a major restructuring just this week to get it GitHub-ready. I'd love feedback on the architecture, the workspace pattern, or anything that looks off to a more experienced eye. 🔗 https://github.com/ehwohl/Zuki Excited to hear what you think!