Skip to main content
Post-processing uses an LLM to clean up, reformat, or restructure your transcriptions after they’re generated.

​
How It Works

  1. You speak and Handless transcribes your audio
  2. The raw transcription is sent to an LLM with your chosen prompt
  3. The LLM returns a cleaned-up version
  4. The polished text is pasted into your app

​
Providers

Configure a post-processing provider in Settings → Post-Processing:
ProviderNotes
OpenAIGPT-4o, GPT-4o-mini, and other models.
AnthropicClaude Sonnet, Haiku, and other models.
GroqFast inference. Llama, Mixtral, and other models.
CerebrasUltra-fast inference.
GeminiGoogle’s Gemini models.
Z.AIxAI’s Grok models.
OpenRouterAccess many models via one API key.
Apple IntelligenceOn-device, Apple Silicon only. No API key needed.
CustomAny OpenAI-compatible endpoint.

​
Built-in Prompts

Three pre-built prompt levels to choose from:
  • Mild — Correct Transcript — fixes typos, punctuation, and obvious errors. Keeps your original wording.
  • Medium — Improve Fluency — smooths out filler words and awkward phrasing while preserving meaning.
  • Aggressive — Restructure & Format — rewrites for clarity, adds formatting (paragraphs, lists), and restructures as needed.

​
Custom Prompts

Create your own prompts for specific use cases:
  1. Go to Settings → Post-Processing
  2. Click Add Prompt
  3. Write your instructions (e.g. “Format as bullet points” or “Translate to Spanish”)
  4. Save and select it as your active prompt

​
Shortcut Binding

Link specific prompts to keyboard shortcuts for fast switching:
  1. Go to Settings → Shortcuts
  2. Add a new shortcut
  3. Assign a post-processing prompt to it
This lets you use different polish levels without changing settings. For example, Fn for raw transcription and Option + Shift + Space for polished output.