Meeting-note software often begins with a convenient transcript and quietly turns the entire conversation into a third party's input.
For customer data, hiring discussions, product strategy, credentials, or other sensitive work, I want a different default: capture, transcription, synthesis, and retention on the Mac.
The complete local pipeline
A useful meeting workflow needs more than a microphone recording. It should:
- capture the user's microphone and the other side of the call
- preserve the tracks
- transcribe them locally
- segment the conversation
- produce a summary, decisions, and action items
- link important output back to source moments
- expose storage, retention, export, and deletion
The last two steps are easy to omit and are essential when generated notes need to be checked.
Pace meeting mode
Pace records two local audio tracks: the user's microphone and system or call audio. When the meeting ends, local transcription feeds a privacy-pinned local note model.
The resulting notes include structured sections and transcript-grounded action items. Meeting files remain ordinary local artifacts under Pace's application-support directory and follow a configurable retention policy. The current public FAQ describes a 30-day default.
Why the privacy pin matters
Pace can use a cloud or command-line planner for other capabilities when the user explicitly selects it. That choice does not allow meeting-note synthesis to leave the local path.
The meeting pipeline has its own stricter boundary. Users can still export content or invoke another external workflow themselves, but the default synthesis remains local.
What I would evaluate
- Does the app capture both sides reliably on the target Mac?
- Do transcription and synthesis work without a network fallback?
- Can an action item show its source quote or timestamp?
- Are files easy to locate, export, and delete?
- Is retention configurable and enforced?
- Are microphone and system-audio permissions explained clearly?
Local does not mean unlimited
Local transcription and synthesis consume memory, storage, and battery. Accuracy still depends on audio quality, speakers, language, and installed models.
A generated summary is a draft. Important decisions should be checked against the transcript rather than accepted because the model sounds certain.
The full meeting pipeline and current limitations are at https://heypace.app/on-device-meeting-notes-mac/.
Top comments (0)