The Three Modes

Vocoding operates in three modes, accessible via the mode selector in the header bar or with dedicated hotkeys.

1. Transcribe Mode

What it does: Voice to clean text (no LLM processing)

Hotkey: Cmd + Shift + R (macOS) / Ctrl + Shift + R (Windows)

Use when:

• You want clean dictation
• LLM is not configured
• You need your exact words, cleaned up

2. Optimize Mode

What it does: Voice to text to LLM optimization to structured prompt

Hotkey: Cmd + Shift + O (macOS) / Ctrl + Shift + O (Windows)

Use when:

• You want structured, optimized output from your voice input
• Writing code documentation, emails, blog posts, or any professional content
• You need an AI agent to refine and format your spoken ideas

3. Chat Mode

What it does: Conversational AI interaction with your selected agent

Access: Use the main hotkey (Option + Space / Alt + Space) while in the Chat view, or type directly in the composer.

Use when:

• Planning features
• Brainstorming architecture
• Multi-turn conversations with context

Switching Modes

You can switch between modes at any time using:

• The mode selector in the header bar (Transcribe | Optimize | Chat)
• The dedicated hotkeys listed above
• The mode-specific global hotkeys work from any application

How Modes Interact with Profiles

Each mode processes your input differently:

• Transcribe: Bypasses the LLM entirely. Output is your cleaned-up spoken words.
• Optimize: Sends your transcription through the selected optimization profile and agent for structured output.
• Chat: Full conversational mode where the agent maintains context across multiple messages.