A lightweight, browser-native, local-first AI terminal and chat interface running entirely client-side.
JAMES is an advanced, client-side AI chat terminal designed to run machine learning models directly inside the browser by leveraging modern web APIs like WebGPU, WebAssembly, and ONNX Runtime Web. By executing tasks entirely on the client side, JAMES ensures complete user privacy, zero server-side model hosting costs, and ultra-low latency inference.
- 100% Client-Side Inference: Runs locally in your browser using WebGPU and ONNX Runtime Web.
- Worker-Based Architecture: Utilizes dedicated web workers to handle model generation asynchronously, ensuring the main UI thread remains responsive at all times (includes a responsive stop-generation button).
- Fast SmallTalk Handler: Features a dedicated, non-LLM
SmallTalkHandlerfor lightning-fast responses to common greetings and basic queries. - Sliding Window Memory: Efficient conversation history management to maintain context without overloading token limits or browser memory.
- Web Search Integration: Built-in web search tool integration designed to augment local knowledge safely and cleanly.
- File Upload Support: Allows users to upload and query documents directly within the chat interface.
- Progressive Web App (PWA): Fully installable as a PWA with manifest support for seamless desktop and mobile experiences.
- Terminal Aesthetic: Immersive retro-futuristic terminal boot sequence and sleek command-line interface design.
- SEO & Performance Optimized: Lightweight architecture optimized for fast initial load times and smooth rendering.
- Core: Vanilla JavaScript / HTML5 / CSS3
- AI / ML Runtime: ONNX Runtime Web, WebGPU, WebAssembly (Wasm)
- Architecture: Web Workers for background inference processing
- Deployment: Render (
chatbotjames.onrender.com)
To run JAMES locally on your machine, follow these simple steps:
A modern web browser with WebGPU support enabled (e.g., Google Chrome, Microsoft Edge, or a WebGPU-compatible browser) and a local development server (such as Python's HTTP server or VS Code's Live Server).
- Clone the repository:
git clone [https://github.com/tripping-alien/chatbotjames.git](https://github.com/tripping-alien/chatbotjames.git) cd chatbotjames