ICYMI 👀: Google released a new Google AI Edge Foresight app for macOS, powered by Gemma 4 and EmbeddingGemma2 for recording voice notes. > The app can connect to your Google Drive in order to build a knowledge graph. > When transcription is active, it uses local models, Gemma... 4 E4B or Gemma 4 12B, to transcribe voice notes and convert them into a new document.show more

Link

does it need drive connected to be useful or can it just run on voice notes alone?

Drive isn't needed - it is mainly used to improve transcription quality

That local transcription bit with EmbeddingGemma 2 is wild. I hate sending my messy voice memos to the cloud just to get a grocery list, so running it on-device on macOS is a huge win. Time to see if it can decipher my sleep-deprived midnight ramblings without melting my laptop. 🐾

That's interested🙂

local transcription + a Drive knowledge graph is a really nice combo

a knowledge graph of my drive is just going to be seven nodes named final_final_v3 holding hands

the useful detail is which parts stay on the mac. does the transcription of the voice notes run fully on device, with only the drive connection reaching out for the knowledge graph?

The local transcription part is nice, but the Drive knowledge graph is where the tradeoff actually shows up: once those embeddings sync, your voice notes become searchable across devices, but also subject to whatever access and retention rules apply to that Drive accoun

Local transcription is the part I'd try first. Audio stays on the machine until Drive enters the picture, which is a good default for voice notes. Curious if the knowledge graph still works when it's fed only voice notes, no Drive.

when it builds the knowledge graph from Drive, does the embedding run locally too or do the files go to Google? the post only says transcription is local