SafeRag SE is a private, local-first AI chat app for macOS that lets you have intelligent conversations with your documents — all without sending a single byte to the cloud
SafeRag SE is a private, local-first AI chat app for macOS that lets you have intelligent conversations with your documents — all without sending a single byte to the cloud.
Chat with Your Documents
Import PDFs, Word documents, text files, web pages, and more. SafeRag SE uses Retrieval-Augmented Generation (RAG) to search your documents and provide accurate, sourced answers. Ask questions,
get summaries, compare documents, and extract key information — all powered by AI that runs entirely on your Mac.
Completely Private
Your data never leaves your device. SafeRag SE runs local language models using llama.cpp, optimized for Apple Silicon. No API keys, no cloud services, no subscriptions. Your documents stay on
your Mac where they belong.
Apple Intelligence Ready
On macOS 26 and later, SafeRag SE integrates with Apple Foundation Models for seamless on-device AI. Choose between local llama.cpp models, Apple Intelligence, or connect to a local Ollama server
— the choice is yours.
Built for macOS
SafeRag SE is a native SwiftUI app designed for Mac. It follows Apple's Human Interface Guidelines, supports light and dark mode, and feels right at home on your desktop.
Key Features
- Import PDFs, DOCX, TXT, HTML, EPUB, and web pages
- RAG-powered document Q&A with source attribution
- Download and manage GGUF models directly in the app
- Watch folders for automatic document import
- Multi-session chat with history
- Optional password encryption for your conversations
- Prompt templates for common tasks (summarize, explain, compare, find)
- Adjustable font sizes and appearance settings
- Keyboard shortcuts for power users
Supported Document Formats
PDF, DOCX, TXT, RTF, HTML, EPUB, Markdown, CSV, and ZIP archives containing supported files. Simply drag and drop or use the document manager to build your knowledge base.
Requirements
- Apple Silicon Mac (M1 or later)
- macOS 14.0 Sonoma or later
- At least 8 GB RAM recommended for local model inference
No Subscriptions, No Hidden Costs
| # | Name | Price | Currency | Region |
|---|---|---|---|---|
| 1 | Item | ¥ 0 | CNY | 中国 |
