What’s the feature?
A toggleable mode (or secondary shortcut) where instead of pasting cleaned-up dictation verbatim, FreeFlow passes the transcript through a summarization/compression step first — outputting a tight, intent-preserving
version of what was said.
Why it matters
Wispr Flow has a version of this. When you’re dictating directly into an AI tool (Claude, ChatGPT, Cursor, etc.), you naturally speak in a rambling, conversational way — filler, repetition, false starts. The current cleanup does a great job fixing grammar and filler words, but the output is still the full length of what you said. For AI prompts specifically, a concise prompt performs better than a verbose one.
Proposed behavior
• Normal mode: cleanup as-is (current behavior, untouched)
• Prompt mode: cleanup → condense → paste a tight 1–3 sentence version that preserves full intent
• Could be a settings toggle, a separate hotkey combo, or context-triggered (e.g., when active app is a browser pointing to claude.ai, chatgpt.com, etc.)
Workaround I’m currently using
Custom system prompt tweak — works partially but applies globally and doesn’t intelligently compress, just cleans.
What’s the feature?
A toggleable mode (or secondary shortcut) where instead of pasting cleaned-up dictation verbatim, FreeFlow passes the transcript through a summarization/compression step first — outputting a tight, intent-preserving
version of what was said.
Why it matters
Wispr Flow has a version of this. When you’re dictating directly into an AI tool (Claude, ChatGPT, Cursor, etc.), you naturally speak in a rambling, conversational way — filler, repetition, false starts. The current cleanup does a great job fixing grammar and filler words, but the output is still the full length of what you said. For AI prompts specifically, a concise prompt performs better than a verbose one.
Proposed behavior
Workaround I’m currently using
Custom system prompt tweak — works partially but applies globally and doesn’t intelligently compress, just cleans.