Sanctum

Sanctum is a desktop application designed for users who prioritize data privacy and control in the age of artificial intelligence. It serves as a private, local gateway to the world of open-source Large Language Models (LLMs). Unlike cloud-based AI services that process your data on remote servers, Sanctum runs entirely on your own device (Windows or Mac, with Linux support coming soon). This local-first approach guarantees that all your conversations, documents, and data are encrypted and remain securely within your personal 'Sanctum,' never being shared or exposed to the internet unless you choose to.

The core mission of Sanctum is to empower individuals and developers to harness AI's potential without compromising on security. It simplifies the often complex process of setting up and running powerful open-source models from hubs like Hugging Face. With an intuitive interface, users can easily download, manage, and interact with a vast library of GGUF models, including popular ones like Llama 3, Mistral, and Phi-3, directly within the application.

How to use Sanctum

Getting started with Sanctum is designed to be straightforward and user-friendly:

1. Download and Install: Visit the official Sanctum website and download the application for your operating system (Windows 10+ or macOS for Intel/Apple Silicon).
2. Explore and Download Models: Once installed, browse the integrated Hugging Face library. Sanctum helps you check if your system's hardware is compatible with a specific model. Select a model and download it directly to your device.
3. Start Chatting: With a model downloaded, you can begin chatting immediately. Since the processing is local, no internet connection is required after the initial model download.
4. Chat with Files (RAG): Drag and drop your local files (PDFs, DOCX, code files, etc.) into the chat. You can then ask questions, summarize content, and extract information from your documents in a completely private environment.
5. Transcribe Audio: Use the audio-to-text feature by downloading a Whisper model. Upload an audio file (like a meeting recording or a voice memo) to get a local, private transcription.
6. For Developers (Local API): Activate the local API server to interact with the running LLM programmatically from your other applications or development projects, all on your local network.

Core Features of Sanctum

• Local AI Processing: Runs full-featured open-source LLMs on your device, ensuring 100% privacy and offline capability.
• Hugging Face Integration: Directly access, check compatibility, and download thousands of GGUF models from Hugging Face.
• Private Document Chat (RAG): Securely chat with, summarize, and ask questions about your local documents, including PDFs, DOCX, PPTX, and various code files.
• Local Audio Transcription: Transcribe audio files privately on your device using Whisper models, with features like real-time synced playback.
• Sanctum Vault: An encrypted local database that securely stores all your chat history. It's protected by a password and a 24-word recovery phrase.
• Local API Server: Powered by llama.cpp, it allows developers to run a local server with any GGUF model, customizable for ports and parameters.
• Advanced Model Customization: Fine-tune LLM performance with custom prompt templates, hardware optimization settings, and inference parameters.
• Cross-Platform Support: Available for Windows and macOS (both Intel and Apple Silicon), with a Linux version in development.

Use Cases for Sanctum

Sanctum is ideal for any task that involves sensitive information:

• Confidential Data Analysis: Analyze private financial data, personal health records, or proprietary business documents without risk of data leaks.
• Secure Content Creation: Draft sensitive emails, write reports, or brainstorm strategic ideas in a secure environment.
• Private Code Development: Collaborate on code, debug, and get programming assistance without exposing your source code to third-party services.
• Research and Study: Interact with research papers, textbooks, and notes in PDF format to summarize key points and deepen understanding.
• Meeting and Call Analysis: Transcribe and review call notes or meeting recordings to extract action items and insights privately.

Advantages of Sanctum

The primary advantage of Sanctum is its unwavering commitment to privacy.

• Unmatched Privacy: With on-device processing and encryption, your data never leaves your computer.
• Full Data Control: You own your data and conversations, stored securely in the local Sanctum Vault.
• Offline Functionality: Work with your AI assistant anywhere, anytime, without needing an internet connection.
• Cost-Effective: Avoid recurring subscription fees for cloud-based AI services by leveraging your own hardware.
• Access to Open-Source Innovation: Stay at the forefront of AI by easily running the latest and greatest open-source models.
• User-Friendly: It removes the technical barriers to running local LLMs, making them accessible to a broader audience.

Pricing and Plans

Sanctum operates on a freemium model, providing both a free and a paid tier to cater to different user needs.

• Base Plan (Free): This plan includes all the essential features for private chatting with locally run LLMs. It's perfect for users getting started with private AI.
• Sanctum Pro (Paid): This subscription plan unlocks advanced features for power users, such as secure and private PDF chatting, advanced search capabilities within your encrypted chat history (Sanctum Vault), and exclusive early access to new features and model updates. Specific pricing details are available on the official website or within the application.