Google’s experimental Mac app transcribes meetings entirely on-device
Google has launched Google AI Edge Foresight, an experimental note-taking app that transcribes meetings and audio entirely on your device without using cloud services. The app addresses growing privacy concerns by keeping sensitive meeting information completely local, offering a privacy-first alternative to competitors like Granola and Wispr Flow.
The free app currently runs only on Mac computers with Apple Silicon and uses Google's on-device EmbeddingGemma 2 model for transcription and processing. Features include automatic meeting summarisation, the ability to transform shorthand notes into polished versions, and a built-in assistant for asking questions about recorded content—with all data remaining on the user's computer.
- Google launches free offline AI note-taking app for Mac
- Transcribes meetings locally without sending data to cloud
- Includes meeting summaries and AI assistant features
New here? Start with this
Meeting transcription apps record conversations and convert them to text, creating searchable records that professionals use for reference and accountability. Many existing services upload these recordings to company servers, which worries users who want to keep sensitive business conversations confidential.
On-device transcription processes audio directly on your computer rather than sending it elsewhere, so nothing leaves your machine. This approach has been less common than cloud-based alternatives, partly because the technology was not powerful enough to work efficiently on personal computers until recently.
Granola and Wispr Flow are established services in the meeting transcription market, both using cloud processing. Growing interest in privacy-first alternatives suggests there is demand from users who prefer their sensitive conversations to remain completely local.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Privacy advocates support on-device transcription because it eliminates data transmission to cloud servers, removing the risk of centralised breaches or exploitative data practices. By keeping meeting transcripts entirely local with no external dependency, users gain genuine control over sensitive information—this directly addresses legitimate concerns about corporations accessing and exploiting confidential meeting content. For privacy-conscious professionals, the on-device approach represents principled protection where sensitive data never enters any company's infrastructure.
The case against
Cloud services with robust encryption and transparent privacy policies offer superior privacy without sacrificing functionality, reliability, or accessibility. On-device processing creates unnecessary practical limitations—restricted device storage, inability to sync recordings across multiple devices, potentially reduced transcription quality, and exclusive access only for those who can afford Apple Silicon Macs—making it a privacy solution available only to the privileged few. Users reasonably expect both privacy and usability; forcing them to choose between these should not be the path to widespread adoption of privacy-protecting tools.
AI Business Companies Research Science Software Technology
Read the full article at the source →
Originally published by The Verge as “Google’s AI note-taking app transcribes your meetings completely offline”.