You promised them something last Tuesday. Nod remembers what.

FAQ

Frequently asked questions

• Does Nod join my meeting as a bot?

  No. It listens through your Mac and that's it. No bot in the participant list, no "Nod has joined" moment, no link to share with the host. The people on your call don't see anything new.

• Does Nod get more useful the more I use it?

  Yes. Every call, meeting, and voice note you capture joins one searchable memory you can ask across, so the more you record, the more Nod can answer. A summary from six months ago isn't dead weight; it's still there when you ask "what did we decide about that?" Most meeting apps hand you a one-off summary and forget it. Your Nod memory keeps its value as it grows.

• Which languages does it work in?

  Fourteen for now: English, Mandarin, Hindi, Spanish, French, Arabic, Bengali, Portuguese, Russian, Urdu, German, Italian, Ukrainian, and Serbian. Whatever language the meeting is in, that's the language your notes come back in. More are on the way, prioritized by what customers ask for.

• Do you store audio recordings?

  No. Audio is held in memory only long enough to chunk it (~5 seconds) and transcribe it. The bytes are released the moment the transcript returns. No file, no waveform export, no cloud-stored audio. Only the transcript and your AI summary are saved. Details: Security § Audio recordings.

• Is transcription on-device, or does my audio go to the cloud?

  Capture and voice detection run on-device. Nod listens through your Mac and an on-device model cuts the audio into complete utterances. Transcription and summarization run in the cloud: each chunk is sent to Whisper through our EU proxy, transcribed, and the audio is discarded the moment the transcript returns. It's routed with Zero Data Retention and no model training, and audio is never stored (by us or the provider), but to be clear, it does leave your Mac for processing. We don't run on-device transcription today. Details: Security § How data flows.

• Where is my data stored?

  In Supabase Postgres in the EU (AWS eu-west-1, Ireland), encrypted at rest with AES-256. Per-user Row-Level Security on every table. See Security § Infrastructure.

• Do you train any models on my data?

  No. Neither we nor our subprocessors. Speech-to-text runs on Groq, whose terms prohibit training on inputs or outputs and which keeps no data by default. LLM and fallback calls go through OpenRouter with Zero Data Retention enabled and all "may train on request data" routes disabled, so the upstream providers (AWS Bedrock, Google Vertex, Azure OpenAI) don't train on your data either. Details: Security § Model training.

• Does it work with Zoom, Google Meet, and Microsoft Teams?

  Yeah, all three. Nod listens to your Mac's audio, so it doesn't care which app the call is in. Slack huddles, Discord, FaceTime, even a phone call you've got on speaker. All the same to it.

• How do you notify other participants that Nod is being used?

  Nod doesn't post a notice in the meeting chat. It can't, because it runs locally on your Mac and never integrates with Zoom, Google Meet, or Microsoft Teams. So letting other participants know is on you.

  Before your first recording, Nod shows a one-time disclaimer reminding you to get consent from everyone on the call. Recording laws vary by place, and some require all-party consent. See the Recording & Consent guide for a quick reference.