Introducing Trace: Private, Local Meeting Transcripts for macOS

Trace is a specialized utility designed for those who need meeting transcripts but refuse to compromise on privacy. Unlike cloud-based services, Trace ensures your data never leaves your hardware.

"Meeting transcripts that never leave your Mac."

🚀 Core Functionality

Trace monitors both your system audio and microphone, performing transcription locally on your device. The result is a clean Markdown file featuring "flagged" highlights inserted exactly where they occurred during the conversation.

System Requirements:

macOS 14 or newer
Apple Silicon (M-series chips)
System Audio Recording permissions

🛠 The User Experience

Trace operates as a lightweight overlay. With a single keystroke, it appears over any active application—whether you are using ~~Zoom, Teams, Meet, or Slack~~ any communication tool.

The "Pill" Interface

The app manifests as a small "pill" that can expand into a live view of the last 120 seconds of conversation, interleaved with your manual flags.

Quick Command Reference:

Action	Shortcut	Description
Record	`⌘ R`	Start/Stop the session
Flag	`⌘ K`	Mark a "Key Moment" inline
Recap	`⌘ P`	View the summary/transcript
Hide	`⌘ H`	Collapse the pill interface

Data Flow Architecture

📂 Local-First Storage

There are no accounts to create and no servers to trust. Every session is stored directly in your Finder as plain text and audio.

Example Directory Structure:

~/Application Support/Trace/2026-04-16-sync-with-alex/
├── mic.wav        # 1.2 MB
├── system.wav     # 1.5 MB
├── transcript.md  # 4 KB
├── transcript.json # 12 KB
└── meta.json      # 1 KB

Because these are standard files, you can easily move them into an Obsidian vault, track changes via git, or simply delete them.

🧠 Intelligence & Integration

AI-Ready Output

Trace provides "contextual markdown" that is optimized for LLMs. You can copy the transcript and paste it into:

ChatGPT
Claude
Notion
Obsidian

Calendar Awareness

By connecting your Mac's local calendar or linking a Google account, Trace becomes "meeting-aware":

It automatically assigns the Event Title to the recording.
It triggers a subtle reminder $\approx 1$ minute before the start time.

Speaker Identification

The software distinguishes between different voices, assigning a unique identifier to each participant. Example: Speaker 2 [00:10]: "Yeah, the build succeeds and the tests pass."

🔒 Privacy & Performance

The efficiency of Trace is driven by the power of Apple Silicon. We can define the privacy model as: $\text{Privacy} = \text{Local Processing} - (\text{Cloud Servers} + \text{Telemetry})$

The speed is equally impressive: $\text{Processing Time} \approx \text{Seconds (not minutes)}$

Comparison with Alternatives

Trace is built for users who ~~want a bot in their call~~ hate having an AI bot join their meeting.

Feature	Trace	Traditional Bots (e.g., Otter)	Cloud Notebooks (e.g., Granola)
On-Device	✅ Yes	❌ No	❌ No
No-Bot Presence	✅ Yes	❌ No	✅ Yes
Offline Mode	✅ Yes	❌ No	❌ No
Account Required	❌ No	✅ Yes	✅ Yes

🏁 Summary: What You Get

A professionally formatted Markdown transcript.
Inline "Key Moments" for rapid scanning.
Total ownership of audio and text data.
Zero telemetry and zero cloud dependency.

Available now on the Mac App Store.