Meeting app detection

One of the most persistent frustrations of digital collaboration is remembering to start a recording. You enter a Zoom or Google Meet room, exchange greetings with attendees, and immediately dive into rapid technical brainstorming or client requirements.

Twenty minutes into the call, someone asks a complex clarifying question, and you realize with dismay that nobody took notes or triggered recording.

Most AI notetaker services attempt to solve this problem by connecting to your cloud Google or Outlook calendar and dispatching an automated meeting bot into every scheduled calendar invitation. But calendar bots arrive uninvited to private 1-on-1s, trigger confusion in external client meetings, and get blocked by corporate security firewalls.

Oyma takes a smarter, native approach. Implemented in src/main/meeting/meeting-detector.ts, Oyma monitors native macOS audio activity to detect when conferencing applications start a call, prompting you to record with a single click right when it matters.

Native audio session detection on macOS

Instead of scraping your web accounts or relying on third-party calendar webhooks, Oyma leverages native macOS CoreAudio monitoring APIs directly on your Mac.

The detection engine operates quietly in the background:

1. Audio pipeline monitoring: When an application initializes an active microphone input stream or begins transmitting bidirectional VoIP audio, macOS CoreAudio dispatches low-level notifications.
2. Process signature inspection: The detector matches the originating application signature against recognized conferencing platforms.
3. Subtle desktop banner: If a matching session is identified and you are not currently recording, Oyma displays an unobtrusive desktop notification banner in the top-right corner of your screen:

   Zoom meeting detected.
   [Record Meeting]   [Dismiss]

4. Immediate one-click capture: Clicking the prompt starts our zero-bot meeting recording pipeline instantly, capturing both your microphone and call participants without delay.

Because detection runs locally on your Mac hardware, it functions whether the meeting was pre-scheduled on your calendar or launched spontaneously as an unscheduled Slack huddle.

Comprehensive platform support

The meeting detector recognizes eight major conferencing ecosystems across both dedicated desktop applications and modern web browsers:

• Zoom: Detects meetings launched via the native macOS Zoom client.
• Google Meet: Recognizes Meet sessions active within Google Chrome, Brave, Safari, and Arc.
• Microsoft Teams: Detects calls in both classic Teams and the modern Teams 2.0 architecture.
• Cisco Webex: Full support for enterprise Webex desktop client audio sessions.
• Slack Huddles: Identifies when you drop into a spontaneous team huddle or channel call.
• Discord: Detects voice channel activity and direct voice calls.
• Around & Whereby: Native awareness for browser-based lightweight collaborative video calls.

Whether you are participating in a global company town hall or a 2-person pair programming call, Oyma ensures you never miss critical context.

Automatic calendar event matching

When you choose to record an identified call, Oyma can automatically link the recording to your active schedule via macOS Calendar integration (src/main/calendar/applescript-event-bridge.ts):

• Meeting title extraction: Matches the current timestamp against your local calendar events to pre-fill the document title with the scheduled meeting name.
• Attendee roster resolution: Automatically populates YAML frontmatter with invited participant names and email addresses.
• Meeting links and agendas: Extracts calendar description notes and embeds them directly into your new meeting document.

When the call concludes, your local AI summaries and live transcription are already linked to the right project, the right attendees, and the right date.

Respecting your privacy and attention

Many commercial monitoring tools feel intrusive, constantly checking screen pixels or nagging users with modal dialogs. Oyma is designed to be calm, respectful, and strictly opt-in:

• Zero keystroke monitoring: The detector inspects only audio stream state flags; it never logs keystrokes, tracks open browser tabs, or inspects unrelated window titles.
• No phantom background drain: The detector uses native event callbacks rather than aggressive polling loops, consuming less than 0.1% CPU while idle.
• Strictly manual recording authority: Detection never begins recording automatically without your explicit consent. You remain in absolute control of what gets captured.
• Per-app alert customization: If you frequently use Discord for gaming or Slack for casual catchups where recording is unnecessary, you can disable detection for specific applications while keeping alerts active for client calls in Zoom or Google Meet.
• Snooze and quiet modes: When you need uninterrupted focus during rapid-fire internal status checks, you can snooze detection alerts for one hour, for the rest of the day, or activate quiet mode directly from the macOS menu bar.

Combined with our zero-bot meeting recording architecture and live on-device transcription, intelligent meeting detection ensures your meetings land in your vault reliably without administrative overhead.