Brown AI — the autonomous, 100% offline AI assistant for Windows and Android
Brown AI is the official 100% private, autonomous offline AI assistant and app for Windows and Android. It runs local LLMs through Ollama, automates your desktop with PowerShell, and holds natural voice conversations without a single byte leaving your machine.
Download Brown AI for Windows · Read the documentation · Read the engineering blog
Features
- Autonomous task execution — Brown plans and runs multi-step desktop work — files, apps, scripts — under the authorization boundary you choose.
- 100% offline local LLMs — Run Ollama and GGUF models on your own CPU or GPU with no network call and no telemetry.
- Native Windows & PowerShell automation — Structured, auditable shell operations instead of copy-pasted commands.
- Local neural voice — Kokoro text-to-speech and Whisper speech-to-text both execute on-device for zero-latency conversation.
- Local vector memory (RAG) — Documents and conversation history are embedded and searched in an encrypted local SQLite store.
- Hybrid cloud routing — Optionally connect Google Gemini, OpenAI, Claude, DeepSeek or Groq with your own API keys.
- Android companion — A native Android app with Qualcomm and Vulkan acceleration, paired to the desktop over local Wi-Fi.
Latest from the blog
- Why 100% Offline Local AI is the Future of Desktop Productivity & Privacy — Cloud AI services promise convenience at the steep cost of confidential desktop telemetry, recurring API bills, and unpredictable outages. Learn why the paradigm is shifting to sovereign on-device computing.
- Automating Windows & PowerShell with Local AI: A Deep Technical Guide — Connecting an AI model to an operating system shell can be dangerous without proper isolation. Here is how Brown safely translates conversational commands into structured PowerShell operations.
- Zero-Latency Voice AI: How Kokoro TTS and Whisper STT Work 100% Locally — Cloud voice assistants have noticeable lag and send raw audio over the web. Brown couples on-device Whisper speech recognition with Kokoro 82M neural voice synthesis for real-time natural dialogue.