I’ve been developing a software full-time focusing on elevating open source LLMs to understand large and diverse directories locally, using advanced RAG systems. Implemented with code and document understanding
The value for end-user being:
- load up any local LLM model and point it at any project directory
- When done indexing, talk with agent regarding your project
- Get answers grounded in reality (with citations).
**
fully air-gapped, 100% offline - zero data shared.**
My aim being letting users to utilize AI models securely on their local folders, using their own hardware - without sharing information.
I made this account to discuss ideas within the field, as well as document my development (no spam - promise). Preparing for alpha release very soon