ExtraBrain supports local NVIDIA Parakeet transcription for on-device workflows, on-device AI where compatible hardware supports it, and optional Deepgram when you choose cloud transcription.
Use local transcription when you want the strictest default data-flow posture.
Use on-device AI on compatible hardware when model requests should stay local.
Use Deepgram when cloud transcription quality, language support, or workflow needs make sense for the session.
Transcript text can stay local with on-device AI or be sent to the configured model provider when you request external AI help.
Turn spoken instructions, clarifications, and constraints into structured notes before you answer.
Keep decisions, risks, owners, blockers, and follow-ups connected to the live conversation.
Choose local transcription when audio should remain on the device, or optional Deepgram when cloud transcription fits the session.
Confirm whether audio is transcribed on your device, uploaded to a transcription service, or processed by a meeting bot vendor.
Local transcription can reduce cloud dependency, but model downloads, app updates, external providers, and optional cloud transcription still require network access.
A transcript can be local while the summary is cloud-generated if you send it to an external model provider. Pick both paths deliberately.
Short answers for people and crawlers comparing ExtraBrain with other live AI assistants.
Yes. ExtraBrain supports local NVIDIA Parakeet transcription for workflows where transcription should stay on the Mac.
Yes. Deepgram is optional and should be used only when its cloud transcription path fits your session and privacy requirements.
Transcript text can stay local with local transcription and on-device AI, but it can be sent to external providers when you choose provider-backed requests.
No. Local transcription only covers the speech-to-text path. A fully local posture also requires on-device AI on compatible hardware and no external provider requests.
