Google Releases AI Edge Foresight for Offline Meeting Notes on macOS
Google has launched AI Edge Foresight, a free experimental macOS application that processes online and offline meeting notes and performs real-time Q&A completely offline. Powered by the new EmbeddingGemma 2 model, the app captures microphone audio and user notes to automatically summarize meetings on Apple Silicon Macs. This release showcases Google's push to bring powerful multimodal AI capabilities directly to user hardware without cloud dependencies, ensuring complete privacy for sensitive meeting data. It highlights the growing movement toward local Retrieval-Augmented Generation (RAG) and performant on-device processing on personal computers. The application uses EmbeddingGemma 2, an open-weights multimodal embedding model with 740 million parameters built on the Gemma architecture and released under an Apache 2.0 license. All audio processing, text indexing, and document querying happen locally on the Mac, keeping user files and audio entirely on-device.
## BACKGROUND
Multimodal embedding models convert unstructured data such as text, images, and audio into numerical vector representations to enable semantic search and Retrieval-Augmented Generation (RAG). Traditionally, executing these models required cloud servers, but efficient architectural optimizations now allow high-quality embedding models to run locally on consumer chips like Apple Silicon.