PrivAI is a privacy-focused writing tool that brings the power of Large Language Models (LLMs) directly to your browser. Unlike other AI tools that send your data to the cloud, PrivAI runs a Llama-3.2-3B model entirely on your device using WebGPU.
This means zero latency (after load), zero data leakage, and zero API costs.
- 100% Local & Private: Your text never leaves your computer. All processing is done client-side.
- Humanize Text: Transform robotic, AI-generated content into natural, engaging, and human-sounding text.
- Text to AI Polish: The reverse operation—turn casual notes or rough drafts into professional, structured, and grammatically perfect content.
- Content Analysis: Analyze text for "AI-isms", readability scores, and get actionable suggestions for improvement.
- Real-time Diff: Visual highlighting of changes to see exactly what the AI modified.
- Frontend: React 19, Vite, TypeScript
- Styling: Tailwind CSS v4, Lucide React (Icons)
- AI Engine: @mlc-ai/web-llm (WebGPU)
- Model: Llama-3.2-3B-Instruct-q4f16_1-MLC
- Visualization: Recharts (Analysis charts), Diff (Text comparison)
- Node.js (v18 or higher)
- A browser with WebGPU support (Chrome 113+, Edge 113+, or Firefox Nightly).
-
Clone the repository:
git clone https://github.com/HenryLok0/PrivAI.git cd PrivAI -
Install dependencies:
npm install
-
Start the development server:
npm run dev
-
Open your browser at
http://localhost:3000/. Note: The first time you run an AI task, the model (~2GB) will be downloaded and cached in your browser.
Contributions are welcome! Please feel free to submit a Pull Request.
This project is open source and available under the MIT License.