Updated
Updated · Android Police · Aug 29
Author Uses 3GB LocalRAG! Model to Fix Obsidian Note Chaos on Mobile
Updated
Updated · Android Police · Aug 29

Author Uses 3GB LocalRAG! Model to Fix Obsidian Note Chaos on Mobile

1 articles · Updated · Android Police · Aug 29

Summary

  • A 3GB on-device model in LocalRAG! let the author index an Obsidian vault on a phone and recover three notes about the same book setting written over a year.
  • That solved a gap Obsidian search could not: full-text search works only if the user knows what to look for, while the vault had grown into hundreds of loosely tagged, poorly linked notes.
  • LocalRAG! reads Markdown files directly from the device, supports 25 file formats, and performs extraction, indexing, search and answer generation offline, with tap-through citations to the source passages.
  • The author said the app does not auto-organize the vault; it surfaces likely connections, while links still have to be reviewed and added manually inside Obsidian.
  • LocalRAG! offers five free questions a day, with Standard at $4.99 a month and Pro at $9.99, while the author also pointed to open-source AnythingLLM as a free on-device alternative.

Insights

What hidden privacy risks lurk within mobile AI apps that promise completely offline access to your most sensitive digital vaults?
Could relying on an offline AI to search your personal notes actually destroy your ability to remember your own ideas?
As personal knowledge vaults grow beyond human memory, will AI-driven hybrid search completely replace traditional folders and tags forever?