title: “RAG”

RAG

Retrieval-Augmented Generation (RAG) is a framework used to optimize the output of a Large Language Model (LLM) by retrieving relevant, authoritative information from an external knowledge base to augment the model’s context window.

Paradigms & Evolutions

• Traditional RAG: Relies on the retrieval of discrete, often static, document chunks to ground model responses in external data.
• LLM Wiki pattern:
  • Employs an LLM to autonomously maintain and evolve a structured wiki.
  • Focuses on a self-sustaining, continuously updating knowledge architecture rather than reactive retrieval of isolated snippets.
  • Reference: 2026 04 10 Karpathys LLM Wiki Beyond RAG for Persistent Knowledge Bases
  • [[concepts/summary|Summary:]]
    • Introduced by [[entities/andrej
• Domain-Specific Fine-Tuning: Uses linear adapters to optimize embedding models for specific domains without full retraining or re-embedding, enabling cost-effective performance improvements. (Reference: [[inbox
• Context Engineering: A new AI skill differentiating from prompt engineering, focusing on structured context provision for LLMs. Defined by Tobi Lütke (Shopify CEO) and Andrej Karpathy as “the art/science of providing all” [incomplete quote]. Emphasizes systematic context design over ad-hoc prompting. Reference: 2026 04 14 Context engineering by prompt engineering channel

Source Notes

• 2026-04-14: # RAG plus knowledge graphs using GRAPHITI --- --- https://www.youtube.com/watch?v=PxcOIINgiaA This video By Cole Medin provides a comprehensive overview of Retrieval Augmented Generation (RAG) and introduces Graphiti, an open-source platform designed to address RAG’s limitatio (RAG plus knowledge graphs using GRAPHITI)

Source Notes

• 2026-04-14: # Updated RAG with Google - Matt Penny channel - Applied AI --- --- https://www.youtube.com/watch?v=qxNLLqwhpb8 Here is a summary of the video content in Markdown format. # Google Just Made RAG Ridiculously Easy Google has released a new tool within the Gemini API called **File (Updated RAG with Google - Matt Penny channel - Applied AI)
• 2026-04-14: # Using docling and Llamaparse in RAG --- --- https://www.youtube.com/watch?v=eHw_6jhK8AM The video explores how to integrate diverse file formats into AI agents, focusing on Retrieval Augmented Generation (RAG) systems. The presenter introduces three document parsing solutions (Using docling and Llamaparse in RAG)
• 2026-04-23: https://www.youtube.com/watch?v=p7yRLIj9IyQ Prompt Engineering Channel: This video provides a comprehensive overview of Jina Embeddings v4, a new universal embedding model designed for multimodal and multilingual retrieval tasks. Here’s a detailed summary of the key points: (Prompt Engineering channel new RAG multi modal approach)
• 2026-04-23: https://www.youtube.com/watch?v=PxcOIINgiaA This video By Cole Medin provides a comprehensive overview of Retrieval Augmented Generation (RAG) and introduces Graphiti, an open-source platform designed to address RAG’s limitations, particularly in dynamic data environments. **1\ (RAG plus knowledge graphs using GRAPHITI)
• 2026-04-23: https://www.youtube.com/watch?v=qxNLLqwhpb8 Here is a summary of the video content in Markdown format. # Google Just Made RAG Ridiculously Easy Google has released a new tool within the Gemini API called File Search, which dramatically simplifies the implementation of Retri (Google Just Made RAG Ridiculously Easy)
• 2026-04-14: # Database search setup - Channel Tech with Tim --- --- https://www.youtube.com/watch?v=XEiQV4zRC-U Here is a markdown summary of the video transcript. # The Future of Database Search: pg_ (Database search setup - Channel Tech with Tim)
• 2026-04-14: # Prompt Engineering channel - new RAG multi modal approach --- --- https://www.youtube.com/watch?v=p7yRLIj9IyQ Prompt Engineering Channel: This video provides a comprehensive overview of Jina Embeddings v4, a new universal embedding model designed for multimodal and m (Prompt Engineering channel - new RAG multi modal approach)