First-Time Setup

SafeRag SE guides you through setup when you first open the app. Here's what to expect.

Welcome Screen

When you launch SafeRag SE for the first time, you are greeted with a welcome screen that introduces the app and its core principles: private, local AI that never sends your data to external servers. From here, you proceed into the setup steps.

Choose Your AI Provider

The first setup step asks you to select an AI provider. SafeRag SE supports three options:

  • llama.cpp (Built-In) — The recommended option for most users. Uses open-source models that run directly on your Mac with no external dependencies.
  • Apple Foundation Models — Available on macOS 26 (Tahoe) and later. Uses Apple's on-device AI models with no separate download required.
  • Ollama — If you already have Ollama installed on your Mac, SafeRag SE can connect to it and use any models you have downloaded there.
Running macOS 26+?
If you are on macOS 26 or later, Apple Foundation Models is the easiest option -- no download needed. Just select it and you are ready to chat.

Download an AI Model (llama.cpp)

If you chose llama.cpp as your provider, the next step is to download an AI model. SafeRag SE presents a list of available models you can browse.

For Macs with 8 GB of RAM, we recommend starting with a smaller model in the 3B to 8B parameter range. Larger models deliver better responses but require more memory and storage.

The download progress is shown in the app. Depending on the model size and your internet speed, this may take several minutes. Models are stored locally on your Mac and can be managed later in Settings.

Set Up Encryption (Optional)

SafeRag SE offers AES-GCM encryption to protect your chat messages at rest. During onboarding, you can choose to set a passphrase that will be used to encrypt your conversations.

When encryption is enabled, your messages are encrypted before being written to the local database. Only someone with the correct passphrase can decrypt and read them.

Encryption is optional and can be enabled or disabled later in Settings.

Remember Your Passphrase
If you forget your encryption passphrase, encrypted messages cannot be recovered. There is no reset or recovery mechanism. Store your passphrase somewhere safe.

You're Ready

Once you have selected a provider (and downloaded a model if using llama.cpp), onboarding is complete. SafeRag SE takes you directly to the Chat Interface where you can create your first session and start having private AI conversations.