Create, customize, and chat with your own AI assistants powered by local language models via Ollama.
⚡ Create Assistants in Seconds: Quickly set up helpful assistants just by describing what you need — like a proofreader, email helper, or markdown editor. The app handles the rest.
🎨 Real-time Ollama Models Listing: Models are pulled in real-time from the Ollama website, so you always see the latest available options — including custom variants created by the community.
📂 Use Your Own Files for Context: You can connect assistants to your local files for smarter, more relevant responses. Just head to the assistant’s Advanced settings and choose a folder.
💬 Keep Conversations Organized: Each assistant maintains its own conversation history with automatic title generation based on your first message. Switch between different conversations or start fresh anytime while keeping your chat history organized and searchable.
🔒 Local-Only by Design: No servers, no cloud. Everything runs on your device and stays private.
⌨️ Dynamic Shortcut Assignment: Use a simple keyboard shortcut to bring up your assistant over any app — super handy when multitasking.
🛠 Easy Setup: A step-by-step guide helps you install everything you need and choose your models, no tech background required.
Head to the releases page and download the latest version for your operating system.
In this video, I create a Proofreader assistant using Task mode, which is designed to respond only to the latest user message—perfect for focused, one-shot tasks like grammar correction or rewriting.
After generating the initial assistant, I also adjust the system instructions to better tailor its behavior to my specific needs. This allows the assistant to follow more precise guidelines, ensuring more accurate and helpful responses.
- ✅ Multiple assistant support with saved conversations
- ✅ Default assistants: Proofreader, Email Assistant, Markdown Helper, Multimodal (Gemma 3)
- ✅ Ollama model management with stream + cancellation support
- ✅ Modern UI powered by Shadcn UI & Shadcn Chat
- ✅ Image upload for vision models (Base64 support)
- ✅ AI-powered assistant behavior + instruction auto-generation
- ✅ Live model listing with metadata and variants via Ollama scraping
- ✅ Dynamic keyboard shortcut support (Windows & Linux)
- ✅ Streamed markdown rendering with code block highlighting
- ✅ Welcome/setup onboarding screen
- ✅ Contextual search over local markdown files using Langchain
- ✅ Planned: Multiple simultaneous chat sessions per assistant
- 🔜 Planned: Support for ChatGPT and other cloud model backends
This project wouldn’t be possible without:
- Electron
- Shadcn UI
- Tailwind CSS
- react-markdown
- react-syntax-highlighter
- Shadcn Chat
- shortcut-recorder-hook
npm inpm run dev# For Windows
npm run build:win
# For macOS
npm run build:mac
# For Linux
npm run build:linux