Troubleshooting
Solutions for common issues with SafeRag SE.
App Won't Start
If SafeRag SE fails to launch or quits unexpectedly:
- Verify your Mac meets the minimum requirements: macOS 15.0+, Apple Silicon, and at least 5 GB of free disk space
- Try restarting your Mac and launching the app again
- Open Console.app (in Applications > Utilities) and filter for "SafeRag" to check for error messages
- If the app was recently updated, try deleting and reinstalling from the Mac App Store
Model Download Failed
If a model download fails or gets stuck:
- Check your internet connection — model downloads require a stable connection
- Verify you have enough free disk space. Models can range from 2 GB to 20 GB or more depending on the model size.
- Try downloading a smaller model first (e.g., a 3B model) to confirm downloads work on your network
- Delete any partial downloads in Settings > Models and retry the download
- If using a VPN or proxy, try disabling it temporarily
AI Responses Are Slow
If the AI takes too long to generate responses:
- Use a smaller model — On 8 GB Macs, stick to 3B–8B parameter models. Larger models require more RAM and processing time.
- Close other apps — Resource-heavy applications compete for GPU and memory. Close apps you don't need while using SafeRag SE.
- Shorter conversations — Long conversation histories require more processing. Start a new session if responses become noticeably slower.
- Try Apple FM — On macOS 26+, Apple Foundation Models may provide faster responses for some tasks since they are optimized for your hardware.
RAG Not Finding Documents
If RAG mode is not returning relevant information from your documents:
- Check that your documents are fully processed — look for a green checkmark next to each document in the document list
- Make sure RAG mode is enabled in your current chat session
- Try asking more specific questions that relate directly to content in your documents
- If a document shows a processing error, delete it and re-upload to trigger reprocessing
- Verify the document format is supported (PDF, TXT, MD)
Encryption Issues
Common encryption-related problems and solutions:
- Forgot passphrase — Encrypted messages cannot be recovered without the correct passphrase. There is no reset mechanism.
- Disable for future messages — You can turn off encryption in Settings. This only affects new messages; existing encrypted messages remain encrypted.
- Before changing settings — Export any unencrypted messages you want to keep before modifying encryption settings.
Ollama Connection Issues
If SafeRag SE cannot connect to Ollama:
- Verify Ollama is running — open Terminal and run
ollama listto confirm - Check that Ollama is listening on its default port (11434)
- Restart Ollama and try reconnecting from SafeRag SE
- SafeRag SE does not require Ollama — if you're having persistent connection issues, switch to llama.cpp in Settings as an alternative
Watch Folder Not Working
If documents in your watch folder are not being detected:
- Grant folder access — SafeRag SE runs in a sandbox and needs explicit permission to access folders. When prompted, click Allow to grant access.
- Verify the folder path — Make sure the configured path still exists and is accessible
- Check file types — Only supported file types (PDF, TXT, MD) are processed. Other file types are ignored.
- Try removing and re-adding the watch folder in Settings
Known Issues
- First model download can be slow depending on your internet connection and model size
- Large PDF files (100+ pages) may take several minutes to process for RAG indexing
- Very long conversations (100+ messages) may gradually slow down response generation
Best Practices
- Keep SafeRag SE updated through the Mac App Store for the latest fixes and improvements
- Use a model size appropriate for your Mac's RAM — see AI Models for recommendations
- Enable encryption if you store sensitive information in your conversations
- Periodically clear old chat sessions to keep the app running smoothly
- Start new sessions for new topics rather than continuing very long conversations
Still Need Help?
If you're experiencing an issue not covered here, we're happy to help:
- Visit the Support Page for additional resources
- Email us at support@corixa.io with a description of your issue
When contacting support, include your macOS version, Mac model, the AI provider and model you're using, and a description of the steps that led to the issue.