Chat Interface

Learn how to use SafeRag's chat interface, manage sessions, and make the most of different chat modes.

Interface Overview

The SafeRag interface is divided into several key areas:

📸
Screenshot: Main Interface Overview
Capture the full SafeRag window showing: the left sidebar with session list, the main chat area with a conversation, and the toolbar at the top. Annotate with numbers or labels for each area if possible.
main-interface-overview.png

Key Areas

  • Sidebar (Left) - Session list, new session button, navigation to other sections
  • Toolbar (Top) - Model picker, thinking mode toggle, font controls
  • Chat Area (Center) - Message history and input field
  • Status Bar - Connection status, trial information

Managing Sessions

Sessions are separate conversations that keep your chat history organized. Each session maintains its own context and settings.

Creating a New Session

Click the + button in the sidebar or use Cmd + N to create a new session.

📸
Screenshot: New Session Dialog
Capture the new session dialog showing: session name field, mode selection (General/RAG/Spark), model picker dropdown, and thinking mode toggle.
new-session-dialog.png

When creating a session, you can configure:

  • Session Name - A descriptive name for the conversation
  • Mode - General, RAG, or Spark mode
  • Model - Which AI model to use
  • Thinking Mode - Enable to see the AI's reasoning process

Switching Sessions

Click any session in the sidebar to switch to it. Your last selected session is automatically restored when you reopen SafeRag.

Renaming Sessions

Right-click a session and select Rename, or double-click the session name in the sidebar.

Deleting Sessions

Right-click a session and select Delete. You can also select multiple sessions using checkboxes for bulk deletion.

⚠ Deletion is Permanent
Deleted sessions cannot be recovered. Make sure you don't need the conversation history before deleting.

Chat Modes

SafeRag offers three chat modes, each optimized for different use cases:

💬

General Mode

Standard AI conversations for everyday tasks like writing, coding, brainstorming, and questions.

📃

RAG Mode

Chat with your uploaded documents. The AI searches your documents and uses relevant content in responses.

✨

Spark Mode

Creative mode with higher temperature for brainstorming, creative writing, and exploratory conversations.

Changing Modes

You set the mode when creating a session. To use a different mode, create a new session with your preferred mode.

Sending Messages

Type your message in the input field at the bottom of the chat area:

Type Your Message

Enter your question or request in the text field. Press Enter to send, or Shift + Enter for a new line.

Wait for Response

The AI processes your message locally and streams the response in real-time. You'll see the text appear as it's generated.

Continue the Conversation

The AI remembers the conversation context, so you can ask follow-up questions naturally.

📸
Screenshot: Chat Conversation
Capture an active chat showing a few message exchanges. Include both user messages and AI responses to demonstrate the conversation flow.
chat-conversation-example.png

Stopping Generation

If the AI is generating a long response, you can click the Stop button (or press Esc) to halt generation immediately.

Thinking Mode

Thinking Mode reveals the AI's reasoning process before it gives an answer. This is useful for:

  • Understanding how the AI arrives at conclusions
  • Debugging complex problems
  • Learning from the AI's approach

Enabling Thinking Mode

Toggle the Thinking button in the toolbar, or enable it when creating a new session.

📸
Screenshot: Thinking Mode
Capture a response with thinking mode enabled, showing the "thinking" section (usually in a different style/color) above the main response.
thinking-mode-response.png
💡 Model Support
Thinking Mode works best with models that support extended thinking (like DeepSeek R1). Some models may not produce visible thinking content.

Toolbar Features

Model Picker

Click the model name in the toolbar to switch between your installed models. The current model name is always visible.

📸
Screenshot: Model Picker
Capture the model picker dropdown showing a list of installed models with the current selection highlighted.
model-picker-dropdown.png

Font Size Controls

Adjust the chat text size using the A- and A+ buttons in the toolbar, or use:

  • Cmd + + to increase font size
  • Cmd + - to decrease font size

Scroll to Bottom

When viewing older messages, click the Scroll to Bottom button (or press Cmd + End) to jump to the latest message.

Keyboard Shortcuts

SafeRag supports these keyboard shortcuts for efficient navigation:

Shortcut Action
Cmd + N New session
Cmd + D Go to Documents
Cmd + M Go to Models
Cmd + , Open Settings
Cmd + + Increase font size
Cmd + - Decrease font size
Enter Send message
Shift + Enter New line in message
Esc Stop generation

Tips for Better Conversations

💡 Be Specific
The more specific your questions, the better the AI can help. Include relevant context and be clear about what you need.
💡 Use RAG for Documents
If you're asking about specific documents, use RAG Mode. The AI will search your uploaded documents and include relevant sections in its response.
💡 Choose the Right Model
Larger models give better quality answers but respond slower. For quick tasks, a smaller model may be more practical.

Next Steps

📃 Upload Documents 🤖 Manage Models