Realtime Register Docusense
AbandonedA knowledge base search engine that uses large language models to provide accurate and relevant answers to user queries based on the knowledge base of Realtime Register.
AI
On this page
Overview
An LLM-powered / RAG-based documentation search engine
Challenge
It used to be quite hard to find certain information in the documentation, since it was limited by Helpscout search capabilities. This meant having to know at least some specifics about what you are looking for.
Approach
The project leverages Ollama for running the LLM and ChromaDB for vector embeddings, providing a scalable and efficient solution for knowledge base search.
Outcome
Searching through documentation is a breeze now, you can just ask your local LLM certain questions and it will answer based on available documentation.
Highlights
- Not associated with Realtime Register directly, personal side-project
- Scraped all relevant documentation pages from Realtime Register
- Used Ollama to power the LLM-based search engine
- Used the all-MiniLM-L6-v2 sentence transformer model for vector embeddings
- Used ChromaDB for an easy-to-use local vector database