Google Introduces Foresight, Offline AI Notes for Mac

Google unveils AI Edge Foresight, a Mac meeting notes application that operates offline thanks to Gemma models running on the device. The tool combines transcription, a document base, and an assistant to provide real-time responses, as the note-taking market becomes increasingly crowded and a public version via Gemini is being discussed.

An Already Active Market with Recent Launches and Acquisitions

The meeting note-taking sector is described as competitive and saturated. In recent months, Wispr has launched its own tool, as has Calendly. Last month, Superhuman, formerly Grammarly, acquired Fathom, a YC-backed note-taker. These initiatives precede the arrival of Google's application.

Taking Notes and Querying Documents During the Meeting

AI Edge Foresight offers a split-screen view that allows users to take shorthand notes while the AI generates a summary in parallel. Users can continue to write manually during the transcription. The application allows users to consult the transcription or interact with an assistant powered by Gemma 4 for answers. It accepts imports from PDF, Google Docs, Microsoft Office formats, plain text, Markdown, and web bookmarks to build a knowledge base, and can respond in real-time if a point refers to this data.

Local Execution on Mac with EmbeddingGemma 2 and Apple Silicon Optimization

Designed for macOS, the application works entirely offline thanks to EmbeddingGemma 2 running on the device, a model with 740 million parameters, to capture meeting notes across various applications, including in-person meetings. The FAQ specifies optimization for Apple Silicon and confirms the fully local mode. This functionality allows users to work on their notes while on the go or without internet access.

Between Past Experimentation and Future Developments via Gemini

In early April, Google had already launched an AI dictation tool based on local models, which remained experimental and did not see widespread adoption. The team behind this project now unveils AI Edge Foresight. This note-taking system could showcase the performance of the Gemma model range in offline mode. Google is also considering launching a version aimed at the general public, accessible via video call applications through its Gemini apps, this time relying on cloud models.