Skip to content

Introduction

Retrieval-Augmented Generation (RAG) is a technique for improving the output of an LLM (Large Language Model) by referencing a domain specific knowledge base before generating a response. On-premise and private cloud PrimisAI customers have access to a customized RAG system to allow organizational knowledge including documentation, code style guidelines, and proprietary IPs to be used by the Magnus AI model to provide the most relevant results for each customer. For example, customers with an appropriate license using Vivado Design Suite may load the Xilinx IP catalog into the RAG, allowing Magnus to use those modules when generating code for designers.

The RAG system in Magnus currently supports a single database for all users of a Magnus server instance. A system administrator at the customer's site should configure the RAG with appropriate documents.

Key Features

  • Custom Data Sources: Augment Magnus' internal knowledge with external data sources to enhance the accuracy and relevance of Magnus' outputs.
  • Real-Time Updates: Continuously expand and update your private data repository, ensuring Magnus adapts to new information and evolving business conditions.
  • Enhanced Security: The RAG knowledge base is hosted securely within your AWS private cloud or on-premises infrastructure, guaranteeing data security and privacy.
  • Intuitive Dashboard: The user-friendly interface simplifies management and interaction with the system.
  • Top-Notch Search Engine: Magnus leverages advanced algorithms for deep and effective querying across large datasets.