Launching today
Skales
Local AI agent that actually does things on your computer 🦎
13 followers
Local AI agent that actually does things on your computer 🦎
13 followers
Skales is a local AI agent for Windows and macOS that acts, not just chats. It manages your calendar, sends emails, automates browser tasks, runs research - and asks before anything destructive. A desktop buddy stays on your screen so you never leave your workflow. Works with any LLM (OpenRouter, Claude, Gemini, Ollama) or fully local. No Docker, no terminal - install in 30 seconds. Your keys, your data, your machine. Built solo from Vienna by a marketing manager who vibecoded it all with AI.
Hey Product Hunt! I'm Mario from Vienna, Austria — senior marketing manager with a decade of hands-on web experience. Vibecoded Skales entirely with Claude and Gemini.
Skales is a local AI agent that actually does things on your computer - not just answers questions. It handles your calendar, sends emails, controls your browser, runs multi-step research tasks, and manages files. The difference: it asks before it acts.
The desktop buddy is just the interface - a persistent AI on your screen you can talk to without switching windows. Think of it as the front door to a full agent system running underneath.
🛡️ Approve or cancel any action before it executes
📱 Control it from your phone via Telegram
🧠 Bi-temporal memory - remembers context across sessions
🤖 Group Chat - 5 different AIs debating one topic
♿ Screen reader and keyboard accessible from day one
My 6yr old built a Snake game with its code builder in 12 minutes. He named the app too.
What's next:
🔌 Custom OpenAI-compatible endpoints (llama.cpp, LM Studio, vLLM)
🎨 Buddy skins - choose your own desktop companion
🌅 Morning briefing - emails, calendar, weather before you ask
🔋 Eco Mode - reduce background token usage
🗣️ Local voice input/output (Whisper + Piper, fully offline)
What would you want a local AI agent to handle for you?
Quick update:
Already got a YouTube comment from someone downloading Skales and asking about LM Studio support - custom OpenAI-compatible endpoints are confirmed for the next update.
Keep the feedback coming! 🦎