Getting started

First launch

When you open MeetingDebrief for the first time, the setup wizard guides you through the initial configuration. The wizard runs once and covers permissions, model selection, and model downloads.

System permissions

MeetingDebrief needs access to your microphone to record audio. On macOS, the app requests microphone permission through the standard system dialog. Grant this permission to proceed.

If you plan to capture system audio (audio from virtual meetings), MeetingDebrief also requires screen recording permission on macOS. The setup wizard prompts you for this separately.

On Windows, microphone access is granted through the system privacy settings. MeetingDebrief requests permission automatically during setup.

Choosing models

The setup wizard asks you to select two types of AI models:

• Transcription model: converts speech to text. Setup suggests Parakeet TDT v3, which covers 25 languages and also powers dictation. Whisper models are available too, including a multilingual model for 99+ languages. See Transcription.
• Summaries & chat: choose On this computer to download a model that generates summaries and chat answers, based on your hardware capabilities, or On a server you already run to use your own AI server. Nothing large is downloaded for the server option.

You can change these selections later in Settings.

Downloading models

After you select your models, the wizard downloads them along with the inference engine. This requires an internet connection and may take several minutes depending on your connection speed and the model sizes you selected.

Once downloaded, the models run entirely on your device. No internet connection is needed for recording, transcription, or summarization, unless you run summaries on a server elsewhere.

Re-running the setup wizard

If you need to reconfigure permissions or re-download models, open Settings → General and select Run setup again. This restarts the setup process without affecting your existing sessions or data.