Framer AI AgentsDesign and publish professional sites with AI
OctoVaultAI
Your private AI paperwork vault. Locally.
14 followers
Your private AI paperwork vault. Locally.
14 followers
OctoVault turns your local documents into a private knowledge graph, so you can ask questions, trace facts to their source, and keep every byte on your device.
Upvoted! cool idea!
Do the documents stay local? Would it work with a google drive folder?
Thanks @tleyden
Yes — 100% local. Documents + extracted facts live in a SQLCipher
vault on your machine; the AI runs locally via Ollama. Only outbound
connections at runtime are to your own localhost (verifiable with
Little Snitch — there's a network-tab screenshot on the landing page
proving zero outbound calls during normal use).
On Google Drive: if you have Drive for Desktop syncing files to a
local folder, you can drag-import from there — we never touch the
gDrive API directly. Talking to Google's API would break the threat
model OctoVault exists to solve.
"Watch this folder for new files" is a reasonable roadmap ask if
that's what you're after — drop me a note at support@octovault.ai
with your workflow.
@sunil28071987 Ok thanks, that makes sense. Let's say I have a google drive folder sync'd to my desktop, would that be "imported" and then left in-place, and it would just give me an AI that would allow me to query for facts? Basically I am wondering if it needs to re-structure my google drive folder, or if it just imports it without touching it?
And yeah, as a user I would expect any changes to be sync'd into OctoVault - all CRUD ops basically.
@tleyden Great Question!
Restructuring: No, we never touch the source folder. Import is read-only — we read the bytes, extract facts into the encrypted vault, your gDrive folder stays exactly as it is.
Auto-sync CRUD: Not today. Import is one-shot — drag in, OctoVault extracts, then it's a static snapshot. Edit/delete in gDrive doesn't propagate.
Watch-a-folder with full auto-sync is exactly the right roadmap ask.
The product questions we'd want to get right:
Updates: re-extract when the file changes → straightforward
Deletes: mirror the folder, or treat extracted facts as user-owned and keep them even if the source disappears? Probably an opt-in
Conflicts: synced file content changes after you manually edited an extracted fact — what wins?
Mind if I reach out at support@ when we're ready to prototype this?
You're describing exactly the workflow I want to design for.