SafeRag Support

1. First Launch Setup

When you first launch SafeRag:

1. Create your first user account (becomes admin automatically)
2. Download your first AI model (SafeRag recommends Qwen3.5 4B on 8GB Macs and Gemma 4 E4B on 16GB or more)
3. Start chatting!

2. Downloading AI Models

SafeRag makes model management easy:

• Go to Settings → Models
• Click "Download New Model"
• Choose from available models (Qwen3.5, Gemma 4, Phi-4, Mistral Small 3.2, Qwen3, Granite 4.0, DeepSeek-R1, etc.)
• Wait for download to complete (can be several GB)
• Model appears in chat mode selector

SafeRag won't start - what should I check?

• Verify macOS 14.0+ (About This Mac)
• Quit SafeRag completely (Cmd+Q) and reopen it — this restarts its bundled AI runtime
• Ensure at least 5GB free disk space
• If you also run a separate Ollama installation, quit it to avoid port conflicts
• Check Console app for error messages

How do I upload documents for RAG?

To use RAG (Retrieval-Augmented Generation):

1. Click the Documents tab in the sidebar
2. Click "Upload Document" button, or drag files onto the Documents area
3. Select PDF, Word (.docx), TXT, Markdown, HTML, or image files (PNG, JPG, GIF, BMP, WebP — read with OCR)
4. Wait for processing (creates vector embeddings)
5. Document appears in your library
6. Enable RAG mode in chat to query your documents

AI responses are slow - how can I speed them up?

• Use smaller models: Qwen3.5 4B or Gemma 4 E4B instead of a 24B or 30B model
• Upgrade RAM: 16GB+ helps significantly
• Apple Silicon: M-series chips provide excellent AI performance
• Close other apps: Free up system resources
• Reduce context length: Shorter conversations process faster

How do I add more users?

Multi-user setup (requires Pro):

1. Log in as admin user
2. Go to Settings → Users
3. Click "Add User"
4. Enter username, email, password
5. Assign role (Admin or User)
6. New user can now log in

Where is my data stored?

All SafeRag data is stored locally:

• Location: ~/Library/Application Support/io.halldor.SafeRag/
• Database: data/chat.db (SQLite)
• Vector DB: data/rag/ (LanceDB)
• Backups: data/backups/
• Audit logs: audit/

How do I backup my data?

SafeRag Pro creates encrypted backups (AES-256-GCM):

1. Open Settings → Backup
2. Turn on automatic backups and choose daily or weekly
3. Backups run while SafeRag is open and are saved to ~/Library/Application Support/io.halldor.SafeRag/data/backups
4. Backups contain accounts and chat history, not your document library — keep your original document files
5. To restore from a backup, contact support@corixa.io

Can I export my data?

Yes. Admins can export data:

• Use Export All Data (admin only)
• Includes accounts (without password hashes), chat sessions and messages, and the list of documents
• Document files themselves are not included — keep your originals

What's the difference between Free and Pro?

New installs start with a 7-day trial that includes every Pro feature.

Free Version (after the trial):

• 1 user account
• Up to 5 documents
• Up to 3 chat sessions
• All chat modes
• Model selection
• Session management

Pro Version ($29.99 one-time):

• Up to 5 user accounts (admin and user roles)
• Unlimited documents and chat sessions
• Watch folder automation
• Audit logging
• Encrypted backups
• Retention settings

How do I upgrade to Pro?

• Buy a license ($29.99 one-time, up to 3 Macs) — see Pricing
• LemonSqueezy emails your license key right after purchase
• Enter the license key in Settings → License (needs an internet connection)
• Pro features unlock immediately; the license is re-checked online at least every 30 days
• Lost your key? Contact support@corixa.io

Ollama Connection Error

Symptoms: "Cannot connect to Ollama" or red status indicator

Note: SafeRag includes a bundled Ollama - no separate installation needed.

Solutions:

1. Quit SafeRag completely (Cmd+Q) and restart it
2. If you have a separate Ollama installed, quit it to avoid port conflicts
3. Restart your Mac if the issue persists

Database Locked Error

Symptoms: "Database is locked" when trying to save

Solutions:

1. Quit SafeRag completely
2. Check for multiple SafeRag processes: ps aux | grep SafeRag
3. Kill duplicate processes if found
4. Restart SafeRag
5. If it persists, contact support

RAG Not Finding My Documents

Symptoms: RAG mode enabled but documents not referenced

Solutions:

1. Verify documents are fully processed (green checkmark)
2. Ensure RAG mode is enabled in chat (toggle in chat input)
3. Try more specific questions related to document content
4. Check embedding model is downloaded (Settings → Models)
5. Re-upload document if processing failed